Title Page

Factors Driving Individuals’ Usage

Intention of Artificial Intelligence (AI)
Assistants in E-commerce:
Perspectives of Users and Non-Users
by Ahlam Alnefaie

Thesis submitted in fulfilment of the requirements for

the degree of

Doctor of Philosophy

under the supervision of Dr. Kyeong Kang and Dr. Osama

Sohaib

University of Technology Sydney

Faculty of Engineering and IT

June 2024

i
 Certificate of Original Authorship

I, Ahlam Alnefaie, declare that this thesis is submitted in fulfilment of the requirements for the

award of Doctor of Philosophy, in the Faculty of Engineering and Information Technology at

the University of Technology Sydney.

This thesis is wholly my own work unless otherwise referenced or acknowledged. In addition,

I certify that all information sources and literature used are indicated in the thesis.

This document has not been submitted for qualifications at any other academic institution.

This research is supported by the Australian Government Research Training Program.

Signature:

Production Note:
Signature removed prior to publication.

24 June 2024

II
 Acknowledgement

First, I would like to thank my God "Allah," for granting me the patience, resilience, and

strength that brought me to this stage in my study journey. Alhamudulillah.

My sincere gratitude extends to my father, Eid Alnefaie, who believed in me but did not live to

see me complete this study. May Allah grant him the highest place in Jannah. Amen. I also

want to give special thanks to my mother and brothers for their continuous prayers, support,

and motivation during my PhD study. Additionally, I express my gratitude to my husband,

Engineer Osamh, and my children, Turki and Areen, for their encouragement and patience and

for accompanying me in completing this study. I extend heartfelt appreciation and thanks to

my supervisors, Dr Kyeong and Dr Osama, for consistently supporting my study, offering

valuable suggestions and providing guidance during my research.

I thank my family-in-law for their prayers and support during my study. Additionally, I would

like to thank my sponsor, the Ministry of Education of the Kingdom of Saudi Arabia, for their

support during my scholarship. Also, I would like to thank all my friends for their motivating

words and Support. Finally, I thank all the researchers I met during this study, who provided

me with a wonderful experience at the university.

III
 List of Publications

The following research papers were published from work undertaken by the author during this

PhD research study.

 Alnefaie, A., Gupta, D., Bhuyan, M. H., Razzak, I., Gupta, P., & Prasad, M. (2020).

End-to-end analysis for text detection and recognition in natural scene images. In 2020

International Joint Conference on Neural Networks (IJCNN) (pp. 1–8). IEEE.

 Alnefaie, A., Singh, S., Kocaballi, B., & Prasad, M. (2021). An overview of

conversational agent: applications, challenges, and future directions. In 17th

International Conference on Web Information Systems and Technologies.

 Alnefaie, A., Singh, S., Kocaballi, A. B., & Prasad, M. (2021). Factors influencing

artificial intelligence conversational agents usage in the E-commerce field: A

systematic. In Proceedings of the ACIS.

 Bin Sawad, A., Narayan, B., Alnefaie, A., Maqbool, A., Mckie, I., Smith, J., &

Kocaballi, A. B. (2022). A systematic review on healthcare artificial intelligent

conversational agents for chronic conditions. Sensors, 22(7), 2625.

 Alnefaie, A., Singh, S., Kocaballi, A. B., & Prasad, M. (2021). Investigating Consumer

Usage Intention of Conversational AI Agents. In Proceedings of the ANZMAC.

 Alnefaie, A., & Kyeong, K. (2023). Cultural Transformations of E-commerce

Consumer Behaviour and Intention Toward Using Artificial Intelligence (AI)

Assistants. In 57th Annual Hawaii International Conference on System Sciences,

HICSS 2024 (HICSS). [Accepted for Publication].

 Alnefaie, A., Kyeong, K., & Sohaib O., (2023). Attitudes and Usage Intentions Towards

Artificial Intelligence (AI) Assistants in E-commerce: A Mixed-Methods Investigation.

Journal of Technology in Society. [Submitted for Publication (i.e., under review)]

IV
 Abstract
The ongoing revolution of e-commerce has brought about significant transformations in the
global retail landscape, redefining how consumers interact with online platforms. In response
to this transformative trend, businesses increasingly adopt and integrate artificial intelligence
(AI) technologies, particularly AI assistants. AI assistants have gained significant traction to
enhance customer engagement, improve personalised experience, and streamline various
aspects of the e-commerce process. Companies across diverse industries and geographical
regions have recognised the potential of AI assistants in fostering more profound connections
with customers, providing real-time support, and bolstering sales through intelligent
recommendations. Consequently, investment in AI research and development has surged,
leading to remarkable advancements in AI assistants’ features and functionalities. Despite the
growing interest of the scientific community and business stakeholders in the topic, scholarly
research on the factors influencing e-commerce consumers’ attitudes and intentions toward
using AI assistants is still limited and provides contradictory evidence regarding some factors.
Moreover, no comparative studies in the e-commerce context empirically investigated the
attitudes of non-users and users toward AI assistant use. Also, several consumers'
demographics have been excluded from prior research, with no previous empirical research on
AI assistant use across different cultural backgrounds.

For these reasons, the study aimed to comprehend the factors influencing consumers'
behavioural intention to utilise AI assistants and to recognise the significant user differences
based on multiple perspectives. This study employed a unique research model based on the
technology acceptance model. It extended it with external factors of AI assistants’ capabilities
that still need to be tested together in AI assistant adoption for e-commerce consumers. This
research conducted a mixed-method approach. In the first phase (Phase A), a quantitative
method was employed to investigate the relationships between the constructs in the study
model, and the Partial Least Square Structural Equation Modelling (PLS-SEM) and several
statistical techniques were adopted. Furthermore, to account for cross-cultural differences and
identify potential variations in usage intentions towards using AI assistants between Eastern
and Western cultures, a multi-group analysis (MGA) was conducted. In the second phase
(Phase B), a qualitative approach was conducted by applying machine learning and natural
language processing techniques to analyse reviews of the Louis Vuitton brand's e-commerce
applications. The objective of this stage was to obtain supporting evidence for the results of the

V
first study and to gain deeper insights into consumer attitudes and experiences. Subsequently,
the results were integrated to provide multiple insights to answer the research questions and
strengthen the findings.

This study has confirmed some previous studies' results and provided new findings. The
attitude factor was the significant predictor of the intention to use AI assistants in non-users
and users, with a direct and positive effect. Perceived usefulness was found to be the
statistically significant predictor of attitudes in both non-users and users of AI assistants. The
additions to the original TAM model, specifically incorporating interactive communication and
personalisation, were statistically significant predictors of the attitudes of non-users and users
to use AI assistants with positive effects. In contrast, perceived ease of use was a non-
significant predictor of the non-users’ attitudes and positively impacted the users’ attitudes
towards using AI assistants. Furthermore, no significant differences existed in the relationships
among the primary factors influencing the intention to utilise AI assistants in e-commerce when
comparing Western and Eastern cultural groups.

This study contributes to both theory and practice by extending the TAM model with two
external factors enabling the assessment of the factors affecting the intention to use AI
assistants from consumer, social, and marketing perspectives and providing new empirical data
on this topic in technology adoption studies. The study also enables further research on this
topic and comparing study results, thus improving understanding of the phenomenon. It also
provides various e-commerce practitioners with valuable information and recommendations
regarding AI assistant use, enabling them to make better decisions in developing and
implementing AI assistant technologies.

VI
 Table of Contents

Certificate of Original Authorship ...................................................................................................... ii

Acknowledgement ............................................................................................................................ iii
List of Publications ........................................................................................................................... iv
Abstract............................................................................................................................................. v
Table of Contents .............................................................................................................................vii
List of Abbreviations ......................................................................................................................... x
List of Tables .................................................................................................................................... xii
List of Figures ................................................................................................................................. xiv
Chapter 1. Introduction ...................................................................................................................... 1
1.1 Research Background .............................................................................................................. 1
1.2 Purpose and Objectives of this Study ....................................................................................... 5
1.3 Research Questions .................................................................................................................. 8
1.4 Significance of this Study ........................................................................................................ 9
1.5 Definition of Key Terms ........................................................................................................ 12
1.6 Literature Review Overview .................................................................................................. 14
1.7 Research Method Overview ................................................................................................... 15
1.8 Thesis Layout ........................................................................................................................ 17
1.9 Chapter Summary .................................................................................................................. 19
Chapter 2. Literature Review ........................................................................................................... 21
2.1. Introduction to AI Assistants ................................................................................................. 21
2.1.1 History and Types of AI Assistants .................................................................................. 21
2.1.2 Roles and Challenges of AI Assistants ............................................................................. 26
2.2 Empirical Research on AI Assistant Adoption in E-commerce ................................................ 31
2.3 Cross-Cultural Perspectives on Technology Adoption and E-commerce .................................. 39
2.4 Factors Impact Behavioural Intention to Use AI Assistants ..................................................... 43
2.4.1 Perceived Usefulness (PU) .............................................................................................. 43
2.4.2 Perceived Ease of Use (PEU) .......................................................................................... 44
2.4.3 Interactive Communication (ICOM) ................................................................................ 45
2.4.4 Personalisation (PERS) ................................................................................................... 46
2.4.5 Attitude (AT) ................................................................................................................... 47
2.5 Chapter Summary .................................................................................................................. 48
Chapter 3. Theoretical Background and Research Model ................................................................. 49
3.1 Theoretical Background ......................................................................................................... 49
3.2 Research Model ..................................................................................................................... 56

VII
 3.3 Hypotheses ............................................................................................................................ 62
3.3.1 Perceived Usefulness and Attitude ................................................................................... 63
3.3.2. Perceived Ease of Use and Attitude ................................................................................ 64
3.3.3. Perceived Ease of Use and Perceived Usefulness ............................................................ 66
3.3.4. Interactive Communication and Attitude ......................................................................... 68
3.3.5. Personalisation and Attitude ........................................................................................... 69
3.3.6 Attitude and Intention ...................................................................................................... 70
3.4 Chapter Summary .................................................................................................................. 72
Chapter 4. Research Methodology ................................................................................................... 74
4.1 Research Paradigm ................................................................................................................ 74
4.2 Research Design .................................................................................................................... 79
4.3 Justification of Research Method Design ................................................................................ 82
4.4 Quantitative Study Approach Overview ................................................................................. 83
4.4.1 Measurement Design ....................................................................................................... 83
4.4.2 Sample and Sampling Size .............................................................................................. 86
4.4.3 Data Collection Method .................................................................................................. 88
4.4.4 Data Analysis Method ..................................................................................................... 89
4.5 Qualitative Study Approach Overview ................................................................................... 91
4.5.1 Data Collection Method .................................................................................................. 91
4.5.2 Data Pre-processing ........................................................................................................ 92
4.5.3 Data Annotation and Vectorisation................................................................................... 94
4.5.4 Data Analysis Method ..................................................................................................... 94
4.6 Integration of Data and Interpretation of Findings .................................................................. 95
4.7 Ethics of the Research ............................................................................................................ 97
4.8 Chapter Summary .................................................................................................................. 97
Chapter 5. Quantitative Data Analysis and Results ........................................................................... 99
5.1 Questionnaire Survey and Participants Demographics ............................................................ 99
5.1.1 Questionnaire Survey .................................................................................................... 100
5.1.2 Demographic Characteristics of Respondents ................................................................ 101
5.2 Data Examination ................................................................................................................ 105
5.2.1 Missing Data Analysis ................................................................................................... 106
5.2.2 Descriptive Data Analysis ............................................................................................. 108
5.2.3 Assessment of Normality .............................................................................................. 111
5.3 Research Model Assessment ................................................................................................ 114
5.3.1 Measurement Model Assessment ................................................................................... 115
5.3.2 Structural Model Assessment......................................................................................... 119

VIII
 5.3.3 Multigroup Analysis ...................................................................................................... 121
5.4 Hypothesis Testing ............................................................................................................... 122
5.5 Chapter Summary ................................................................................................................ 125
Chapter 6. Qualitative Analysis and Results ................................................................................... 126
6.1 Introduction ......................................................................................................................... 126
6.2 Description of Online Reviews............................................................................................. 127
6.2.1 Data Pre-processing of Online Reviews ......................................................................... 127
6.2.2 Data Annotation and Vectorisation ................................................................................. 128
6.3 Natural Language Processing Analysis ................................................................................. 129
6.3.1 Keywords Extraction ..................................................................................................... 129
6.3.2 Topic Modelling ............................................................................................................ 131
6.4. Thematic Analysis and Integrative Findings ........................................................................ 132
6.5 Chapter Summary ................................................................................................................ 134
Chapter 7. Discussion, Implications and Conclusion ...................................................................... 135
7.1 Research Aim and Questions ................................................................................................ 135
7.2 Findings ............................................................................................................................... 137
7.2.1 Usefulness..................................................................................................................... 139
7.2.2 Ease of Use ................................................................................................................... 141
7.2.3 Ease of Use Positively Influences Usefulness ................................................................ 143
7.2.4 Interactive Communication ........................................................................................... 145
7.2.5 Personalisation .............................................................................................................. 147
7.2.6 The Role of Attitude in the Intention to Use AI Assistants .............................................. 150
7.2.7 Comparison of Western and Eastern Consumers ............................................................ 152
7.2.8 Mixed-Method Findings ................................................................................................ 154
7.4 Contribution and Implication for this Study .......................................................................... 157
7.4.1 Theoretical Implications ................................................................................................ 157
7.4.2 Practical Implications .................................................................................................... 161
7.5 Conclusion........................................................................................................................... 164
7.6 Limitations and Direction for Future Research ..................................................................... 165
7.7 Chapter Summary ................................................................................................................ 168
References..................................................................................................................................... 169
Appendices.................................................................................................................................... 194
Appendix A. Factors Questions of the Survey Questionnaire ...................................................... 194
Appendix B. AI Assistant Interaction Simulations ...................................................................... 196

IX
 List of Abbreviations

AI: Artificial intelligence

ALICE: Artificial Linguistic Internet Computer Entity

AT: Attitude

E-commerce: Electronic Commerce

MGA: Multi-Group Analysis

ECA: Embodied Conversational Agents

FAQs: Frequently Asked Questions

HCI: Human-Computer Interaction

ICOM: Interactive communication

IS: Information Systems

ML: Machine Learning

MTurk: Mechanical Turk

PERS: Personalisation

PEU: Perceived Ease of Use

UI: Usage Intention

PLS-SEM: Partial least squares structural equation modelling

PU: Perceived Usefulness

TAM: Technology Acceptance Model

NLP: Natural Language Processing

TRA: Theory of Reasoned Action

UTAUT: Unified Theory of Acceptance and Use of Technology

SET: Social Exchange Theory

X
LDA: Latent Dirichlet Allocation

NLTK: Natural Language Toolkit

TF-IDF: The Frequency Inverse Document Frequency

SD: Standard Deviation

SE: Standard Error

AVE: Average Variance Extracts

XI
 List of Tables
Table 2.1. Timeline of Chatbot Development ................................................................................... 22

Table 2.2. Key factors affecting the adoption of AI assistants ........................................................... 34

Table 2.3. Summary of Recent Empirical Studies on AI Assistant Use.............................................. 36

Table 3.1. Definition of Factors Included in the Research Model ...................................................... 59

Table 3.2. Summary of Hypotheses .................................................................................................. 72

Table 4.1. Measurement Items ......................................................................................................... 84

Table 5.1. Demographic Profile of Respondents. ............................................................................ 101

Table 5.2. Reasons for Not Using AI Assistants by Respondents .................................................... 104

Table 5.3. Responses of Perceived Usefulness and Ease of Use ...................................................... 106

Table 5.4. Responses of AI Assistant Capabilities .......................................................................... 107

Table 5.5. Responses of Attitude towards the Use .......................................................................... 107

Table 5.6. Responses of Usage Intention ........................................................................................ 108

Table 5.7. Descriptive Statistics for Perceived Usefulness .............................................................. 108

Table 5.8. Descriptive Statistics for Perceived Ease of Use............................................................. 109

Table 5.9. Descriptive Statistics for Interactive Communication ..................................................... 109

Table 5.10. Descriptive Statistics for Personalisation...................................................................... 110

Table 5.11. Descriptive Statistics for Attitude................................................................................. 110

Table 5.12. Descriptive Statistics for Usage Intentions ................................................................... 110

Table 5.13. One-Sample Kolmogorov-Smirnov Test for PU ........................................................... 112

Table 5.14. One-Sample Kolmogorov-Smirnov Test for PE............................................................ 112

Table 5.15. One-Sample Kolmogorov-Smirnov Test for ICOM ...................................................... 113

Table 5.16. One-Sample Kolmogorov-Smirnov Test for PERS ....................................................... 113

Table 5.17. One-Sample Kolmogorov-Smirnov Test for AT............................................................ 113

Table 5.18. One-Sample Kolmogorov-Smirnov Test for UI ............................................................ 113

Table 5.19. Measurement model- Construct Reliability and Validity ............................................... 116

Table 5.20. Fornell-Larcker Criterion Analysis............................................................................... 117

XII
Table 5.21. Discriminant validity - Cross-loading .......................................................................... 117

Table 5.22. Model Fit .................................................................................................................... 117

Table 5.23. Latent Variables Correlation ........................................................................................ 118

Table 5.24. Path Coefficients ......................................................................................................... 119

Table 5.25. Multigroup Analysis .................................................................................................... 121

Table 5.26. Hypothesis Testing ...................................................................................................... 124

Table 6.1. Data Pre-processing....................................................................................................... 128

Table 6.2. Data Annotation and Vectorization ................................................................................. 129

Table 6.3. The Top 25 of Three-Word Phrase Extraction ................................................................ 131

Table 6.4. Thematic Analysis ......................................................................................................... 132

Table 6.5. Integration Analysis ....................................................................................................... 133

XIII
 List of Figures
Figure 1.1. Research Design ............................................................................................................ 16

Figure 2.1. Classification of AI Assistants ........................................................................................ 24

Figure 3.1. The TRA Model ............................................................................................................. 50

Figure 3.2. The TAM Model ............................................................................................................ 51

Figure 3.3. The UTAUT Model ........................................................................................................ 52

Figure 3.4. Research Model ............................................................................................................. 56

Figure 4.1. Research Design ............................................................................................................ 80

Figure 5.1. Gender of participants .................................................................................................. 102

Figure 5.2. Education level of the participants................................................................................ 103

Figure 5.3. Information Obtained from AI Assistants ..................................................................... 104

Figure 5.4. Two Main SEM Components ....................................................................................... 115

Figure 5.5. Scatter Plots for Relationship Paths .............................................................................. 118

Figure 5.6. Research Conceptual Model ......................................................................................... 119

Figure 5.7. Tested Model for Users ................................................................................................ 120

Figure 5.8. Tested Model for Non-Users ........................................................................................ 121

Figure 6.1. Positive and Negative Reviews .................................................................................... 128

Figure 6.2. Frequently Used Keywords in Positive and Negative Reviews ..................................... 130

Figure 7.1. Research Structural Model ........................................................................................... 136

XIV
Chapter 1. Introduction

This chapter provides an introductory overview of the research background, encompassing

various elements essential to the study (1.1), and addresses the research purpose and objectives

(1.2) and the research questions (1.3). Moreover, the significance of this study is expounded,

highlighting its relevance in both academic and practical contexts (1.4). Section (1.5) provides

the definition of key terms. In addition, the literature review and method adopted for data

collection and analysis are also introduced (1.6) and (1.7), and finally, the overall thesis layout

is presented (1.8). Section (1.9) concludes with a concise summary of the main points covered

in the chapter.

1.1 Research Background

Artificial intelligence (AI) is becoming a more popular tool for solving commercial difficulties,

such as e-commerce, by utilising natural language processing, machine learning, and deep

learning algorithms. Many sectors are turning to AI to develop new communication methods

with and engage customers (Libai et al., 2020). Recent text recognition and natural language

processing developments have significantly influenced AI and text analytics (Mah et al., 2022).

Chatbots, or AI assistants, are examples of AI systems that provide natural language user

interfaces that stimulate human-human conversation. These AI assistants use text-based or

voice-based inputs and outputs and are integrated into social media platforms or websites.

Moreover, AI assistants, formerly offered as a customer care solution in travel, banking, and

retail e-commerce, are now being utilised to engage customers in online shopping. Many

retailers also adopt AI assistants to provide customer service and customised online experiences

(Chung et al., 2020). For example, Louis Vuitton has introduced a chatbot that delivers

information about stores, products, and customer service (Fernandes, 2020). These AI

assistants, which rely on machine learning and AI algorithms, can provide the required

1
information quickly and directly. From the customers' perspective, AI assistants enable

convenient customer service and customised interaction without downloading an app (Akhtar

et al., 2019). As a result, experts anticipate that enterprises will produce more AI assistants than

traditional mobile apps (Brandtzaeg & Følstad, 2018). Furthermore, Insider (2020) predicts

that conversational agents will save $11 billion by 2023. According to Jovic (2020), business

experts estimated that by 2022, 90% of bank customer interactions will be handled without

human agents and with developing technologies such as machine learning applications and

chatbots. This anticipation is supported by a recent development in the environment, in which

face-to-face interaction with human agents became inaccessible during the COVID-19

pandemic. That is why customers now rely on social networking sites and online

communication applications such as AI assistants to seek information about purchasing

products and contacting brands for customer service.

Despite AI assistants’ various roles and capabilities and the increasing attention from

researchers and practitioners towards their adoption, the scholarly investigation concerning the

determinants of consumers’ behavioural intention to use AI assistants in e-commerce has been

notably constrained and insufficient (Jain et al., 2022). Moreover, it is noteworthy that previous

studies examining the adoption of AI assistants have predominantly relied on theoretical

models that overlook essential aspects such as social communication skills and personalised

services and provided conflicting findings regarding certain factors influencing the use of AI

assistants (Go & Sundar, 2019; Nordheim et al., 2019; Liew et al., 2021). At the same time,

social presence influences consumers’ trust in AI assistant usage (Toader et al., 2020), and

communication competence does not affect customer satisfaction (Chung et al., 2020). Also,

previous studies in AI assistant acceptance were mainly conducted in specific cultural contexts

(Sindermann et al., 2022), limiting the applicability of their findings to other cultural contexts.

On the other hand, although cross-culture research is interesting for global insights, empirical

2
research has yet to be conducted on cross-culture analysis for AI assistant adoption in e-

commerce (Perifanis & Kitsios, 2023; Bawack et al., 2022).

In the case of the e-commerce context, Oguntosin and Olomo (2021) and Syarova (2022)

conducted the adoption of chatbots in the context of online shopping. Still, they collected data

from only users and suggested other studies to focus on nonusers. Følstad and Brandtzaeg

(2020) pointed out a lack of studies focusing on users unfamiliar with AI assistants in customer

services. Moreover, no comparative studies in e-commerce consumer research empirically

investigated the attitudes of users and non-users toward AI assistants called chatbots (Følstad

& Brandtzaeg, 2020). Therefore, this study examines the factors affecting users’ and non-users’

attitudes toward adopting AI assistants for Western and Eastern e-commerce’ consumers.

Understanding users’ differences may assist enterprises and organisations seeking to introduce

new technology and modify their marketing and outreach activities to better align with their

target audience's needs (Choi et al., 2014).

From a practical point of view, the significance of AI assistants lies in their capacity to enhance

the interaction between companies and customers through diverse marketing initiatives.

Therefore, assessing consumers' perceptions of AI assistants becomes imperative, as these

perceptions can significantly impact their intention to use such technology. While there is an

abundance of research focusing on the technical aspects of AI assistants, there remains a need

for further investigation into the acceptance and consumer behaviour concerning the adoption

of AI assistants (Cheng & Jiang, 2021; Chung et al., 2020). Moreover, the investigation of

adoption behaviour, with a particular emphasis on usage intention, is the leading research

direction for many information system and marketing scholars due to its consideration of usage

intention as a reliable assumption of practical technology usage (Al-Adwan et al., 2023; Zhang

et al., 2021). While usage intention is defined as individuals’ intention to utilise the information

system (Bhattacherjee, 2001), based on this definition, the usage of AI assistants refers to the

3
extent to which a consumer intends to use the AI assistants. The initial development of AI

assistants was based more on technology capabilities than on market aspects; thus, customers'

adoption of AI assistants has been relatively limited since customer perceptions and demands

were not effectively addressed (Zierau et al., 2020; Zarouali et al., 2018). Usage intention, in

particular, is frequently recognised in the literature as a predictor of new technology efficacy

and leads to improved consumer behaviour about the latest technology (Al-Adwan et al., 2023;

Liao et al., 2009).

Meanwhile, due to COVID-19, consumers spend more time online, driving firms to adopt new

digital strategies for connecting with loyal customers. This challenge has compelled

organisations to devise or find innovative solutions to adapt to the new environment.

Lockdowns and social distancing measures have disrupted consumer behaviour worldwide,

driving businesses to seek new methods to engage with consumers locked at home and re-

establish their market position (Sharma & Jhamb, 2020). As consumers integrate new digital

technologies into their isolated lifestyles, their shopping habits and daily routines change (Ali,

2020). Overall, the current study sheds light on the evolving role of AI assistants in e-commerce

and helps businesses make informed decisions about the design of this technology to meet

consumers' changing needs and preferences.

As AI continues to transform the landscape of online shopping, understanding the drivers

behind consumers' decisions to engage with AI assistants is crucial for businesses and

policymakers alike. Recently, AI-powered assistants have gained popularity in e-commerce,

offering personalised and efficient customer experiences. However, despite their growing

prominence, there needs to be more understanding of how consumers’ usage patterns may

shape their acceptance and adoption of AI assistants. Also, recognising and comparing cross-

cultural consumers will provide valuable insights into tailoring AI-based services to suit diverse

consumer preferences, ultimately enhancing user satisfaction and loyalty.

4
This research will employ a comprehensive and multi-dimensional approach to investigate the

attitudes, usage intention, and perceptions of e-commerce users towards AI assistants in both

Western and Eastern contexts. Therefore, the study aims to comprehensively understand the

factors that influence behavioural intention to use AI assistants among non-users and e-

commerce users. The findings of this study will offer actionable recommendations to

businesses and e-commerce platforms seeking to implement AI assistants, ensuring their

strategies are attuned to consumers’ specific preferences and expectations. Additionally, from

an academic standpoint, this research will contribute to the growing body of knowledge

concerning the factors that influence consumers’ attitudes and behavioural intentions regarding

AI assistant usage to ensure the further development of AI assistants that can lead to successful

marketing and commercial objectives.

1.2 Purpose and Objectives of this Study

The previous section discussed the study's research background and identified research gaps.

This section addressed the study's research purpose and objectives. Given the increased

adoption of AI assistant applications, this study aims to investigate and comprehend the factors

influencing the attitudes and usage intention of AI assistants across users and non-users, as well

as to establish a conceptual framework concentrating on cultural differences in the relationships

between influencing factors, attitudes, and behavioural intention in the context of e-commerce.

However, some online purchasers still need to trust the capabilities of AI assistants (Uysal et

al., 2022). The development of new technologies that aim to provide services to global

consumers must include a consumer-centric approach (Zhang et al., 2021). In this context, the

study aimed to achieve multiple objectives:

- To explore key factors influencing the intention towards adopting AI assistants in e-

commerce by developing a unique research model through extending the technology

acceptance model for adopting AI assistants in E-commerce.

5
 - To empirically examine the model using both quantitative and qualitative methods to

determine which factors significantly influence the intention of users and non-users to

continue using AI assistants in the context of e-commerce.

- To examine the potential variations in the relationships among the primary factors

influencing the intention to use AI assistants across two cultural contexts to understand

their attitudinal differences towards AI assistant adoption. This aims to augment the

theoretical comprehension of this subject and provide insights into optimising the

design of AI assistants to accommodate diverse preferences, ultimately contributing to

heightened customer satisfaction.

The rapid advancement of AI technology has led to a proliferation of AI-powered assistants in

the e-commerce domain, significantly transforming how consumers interact with online

platforms. These AI assistants offer personalised recommendations, enhanced customer

support, and streamlined purchasing processes, among other benefits. The motivation behind

this research is multi-fold, including addressing a gap in the literature, enhancing e-commerce

strategy, cross-cultural insights, and academic contribution. Even though this study did not

cover major differences between the two cultural groups, it's still really important to consider

how culture affects AI technology use (Chi et al., 2023). Investigating the factors that influence

consumers' behavioural intention to use AI assistants in a cross-cultural context holds

significant theoretical and practical implications. It will advance scholarly understanding and

provide valuable insights for e-commerce businesses and technology developers striving to

create inclusive and user-centric AI assistant solutions. While AI's technological aspects have

garnered significant attention, a comprehensive understanding of its social and marketing

dimensions is essential for harnessing its full potential. The primary motive for the research

was to integrate AI's social and marketing perspectives, shedding light on its impact on

consumer behaviour and usage decisions. From a social perspective, this study explored how

6
interactive communication shapes consumers' interactions, attitudes, and usage decisions to

start and continue using this technology. From a marketing perspective, this study investigated

how personalised marketing efforts of AI assistants influence consumers’ attitudes towards

using AI assistants.

An additional factor intensifying the motivation for undertaking this investigation is the

absence of comprehensive empirical studies into this subject matter, considering both the

perspectives of users and non-users. Previous researchers have presented in a number of studies

contradictory results about the influence of certain factors among users. Følstad and Brandtzaeg

(2020) noted a lack of studies focusing on unfamiliar users with AI assistants. Another motive

for this research is to discover the crucial cross-cultural differences between Western and

Eastern consumers. Exploring diverse cultures is crucial because culture plays a big role in

shaping people's thoughts about and acceptance of technology. This study focuses on Western

and Eastern consumers for several reasons. Firstly, Western and Eastern cultural identities and

profiles are significantly different. Secondly, within the scope of cultural perspectives, it is

notable that Western culture places a greater emphasis on explicit knowledge and tangible,

individualistic motivational factors, in contrast to Eastern cultural views that lean towards

valuing tacit knowledge and abstract principles (Jelavic & Ogilvie, 2010). Thirdly, Eastern

consumers prefer a high-context communication pattern, while Western consumers prefer a

low-context one (Liao et al., 2008). Conducting cross-cultural research helps global business

practitioners identify differences in cultural attitudes towards adopting AI assistants. This, in

turn, empowers them to refine strategies for developing AI assistants that better suit diverse

markets.

Dwivedi et al. (2021) demonstrate that technology developers need adequate insights

concerning the social-cultural factors that might influence how end-users interact with AI. The

global reach of e-commerce businesses implies that AI assistants interact with a diverse

7
population with different social and cultural backgrounds (Balakrishnan & Dwivedi, 2021).

Therefore, understanding the cultural differences and similarities that emerge among the target

consumers can be instrumental in developing AI solutions that can adequately enhance online

business outcomes by effectively meeting the expectations of each customer. According to

Zhang et al. (2021), AI assistants can be instrumental in enhancing marketing capabilities in e-

commerce. In line with these sentiments, Davenport et al. (2020) suggest that strategic

marketing involves market research that covers the cultural characteristics of the target

population. As a result, understanding the impact different cultures have on the use of AI

assistants can help enhance the marketing strategies adopted with the application of such

technologies in e-commerce.

1.3 Research Questions

Although the AI assistant literature has seen significant advancements, notable research gaps

remain in understanding the factors influencing AI assistant usage and their implications on

consumers' attitudes and adoption of e-commerce. The main research question of this study is:

Main Research Question: What factors influence individuals' intention to use AI assistants in

e-commerce among users and non-users, and are there significant differences in these factors’

impact when comparing Western and Eastern cultures?

The main research question has been subdivided into several sub-questions as follows:

RQ1.1: How does perceived usefulness affect the attitudes of users and non-users towards

using AI assistants in e-commerce?

RQ1.2: How does perceived ease of use affect the attitudes of users and non-users towards

using AI assistants in e-commerce?

RQ1.3: How does perceived ease of use affect the perceived usefulness of users and non-

users of AI assistants in e-commerce?

8
 RQ1.4: How does interactive communication affect the attitudes of users and non-users

towards using AI assistants in e-commerce?

RQ1.5: How does personalisation affect the attitudes of users and non-users towards using

AI assistants in e-commerce?

RQ1.6: How does attitude affect the intentions of users and non-users towards using AI

assistants in e-commerce?

RQ1.7: Do significant differences exist in the relations among the primary factors

influencing the intention to utilise AI assistants in e-commerce when comparing Western

and Eastern cultures?

1.4 Significance of this Study

The rapid advancement of AI technology has significantly transformed and impacted various

aspects of society, including how consumers interact with online platforms. As AI assistants

become more widely integrated into e-commerce platforms, specifically text-based AI

assistants known as chatbots designed to provide overall customer support in the fashion

industry, it is essential to understand the key factors influencing and driving consumers’

intention towards their use. AI technologies are seen, expected, and accepted differently in

different consumers' experiences and technological skills. Therefore, investigating the users’

differences in consumer behaviour and intention concerning AI assistants can provide valuable

insights into the complicated interplay between technology and consumer research. By

exploring the important aspects that influence attitudes and usage intention, this research sheds

light on the underlying factors that drive consumers' attitudes towards using AI assistants in e-

commerce. Understanding how consumer beliefs, perceived usefulness, and perceived ease of

use affect consumer attitudes can assist e-commerce businesses in developing solutions that

are tailored to specific consumer preferences and expectations.

9
Additionally, investigating the relationship between AI features in interactive communication

and personalisation, consumer attitudes, and intentions towards AI assistants is also crucial. As

the pandemic has accelerated the adoption of online platforms and increased reliance on online

shopping, limited research has been conducted on the factors affecting users’ and non-users’

evaluation of AI assistant system performance (satisfaction and adoption) in e-commerce (Song

et al., 2019). Meanwhile, understanding the differences between Eastern and Western cultures

is crucial when investigating the research questions on e-commerce consumers' behavioural

intention towards using AI assistants. The significance lies in recognising that cultural

variations can significantly influence consumers' attitudes, perceptions, and preferences, thus

shaping their adoption of AI assistants in e-commerce. By investigating different types of

customers, researchers can find the differences in customers' attitudes towards AI adoption or

resistance to using AI assistants and provide valuable insights for new or potential users.

Moreover, by considering cultural nuances, researchers can identify how these factors may

differ or align between Western and Eastern consumers, providing valuable insights for

businesses to tailor AI assistant experiences that resonate better with specific cultural contexts.

Investigating whether a distinct variation exists in the intention to use AI assistants between

the two groups of consumers helps to address potential disparities in their acceptance and

utilisation. In essence, comprehending Western and Eastern consumers’ differences enhances

the applicability of research findings, facilitating the development of more effective strategies

for e-commerce businesses and marketers in respective regions. This understanding can

provide insights into strategies for effectively addressing consumer concerns and promoting

positive experiences with AI assistants. Hence, this study explores various factors impacting

customers' attitudes towards AI assistants employed within e-commerce platforms,

encompassing websites, mobile applications, or social media sites.

10
The current study is significant in developing a comprehensive understanding of how the broad

Western and Eastern cultures influence the use of AI assistants as a marketing strategy in e-

commerce. According to Song et al. (2018), customers from different cultural backgrounds

present varying reception to marketing strategies. Cornali and Tirocchi (2012) and Ding and

Saunders (2006) show that the transition to digital technologies and the diffusion of cultures

influenced by increased socialisation and the globalisation phenomenon has led to the

emergence of an international culture. As a result, there are many commonalities among online

users that foster the deployment of technologies that have similar features in the global e-

commerce ecosystem. The insights gained from the current study will help to demonstrate if

the differences among cultures call for variations in the AI solutions deployed in e-commerce

when targeting a culturally diverse customer base. Additionally, these insights can highlight

the social and technical implications in the development, implementation, and use of AI

assistants in e-commerce.

This study significantly contributes to multiple interdisciplinary knowledge domains, including

consumer culture, marketing science, information systems research, engineering, and design.

The research delivers the following valuable insights, enriching the existing body of knowledge

in these diverse fields. This contribution makes several significant contributions to

conversational AI assistants and their use in e-commerce.

1. It investigates state-of-the-art conversational AI assistants, providing valuable insights

into their capabilities and limitations.

2. The contribution presents a centric customer assessment framework to measure

customers' attitudes toward using conversational assistants for e-commerce. This unique

framework provides a valuable tool for developers and marketers looking to assess the

effectiveness of their conversational assistant technology.

11
 3. It provides new empirical data on the cross-cultural investigation for AI assistant

adoption while presenting a sound basis for further research into different cultural factors.

4. This research provides insights for developers and designers interested in developing AI

assistants for successful marketing and business objectives. These insights guide the

development of conversational assistants that meet the needs of both businesses and

customers.

5. This contribution provides valuable insights for online retailers and marketers looking to

develop online experiences incorporating conversational assistants as a customer service

and practical communication application.

1.5 Definition of Key Terms

AI assistants. They are also known as chatbots or virtual assistants. Maedche et al. (2019)

describe AI assistants as software applications that leverage AI technologies to foster

interactive and personalised services. According to Mekni (2021), AI assistants are human-like

tools designed to understand natural language input in their interaction with users and offer

relevant information or perform targeted actions. The current study is centred on how AI

assistants are employed in e-commerce to meet the needs and preferences of users and non-

users. Subsequently, the insights gathered can facilitate an understanding of the differences that

exist between people from Western and Eastern cultures involved in e-commerce.

Users and non-users. For this study, users and non-users allude to individuals who interact

with AI technologies as experienced and first-time customers in e-commerce businesses that

have adopted AI assistants. Therefore, users are individuals who are proactive in the use of AI

assistants in their regular online shopping activities (Malodia et al., 2022). These individuals

are tech-savvy and tend to have a positive attitude towards AI assistants by appreciating

specific features such as automation, efficiency, and personalisation. Non-users allude to

individuals with limited or no experience in the use of AI technologies and have limited

12
intentions to become active users of AI assistants in e-commerce (Lee et al., 2021). Therefore,

their limited exposure to AI technologies might compromise their attitudes and intentions

towards AI assistants. Understanding how users and non-users interact with AI shopping

assistants is essential in developing solutions that can enhance the application of such tools to

improve the customer experience and organisational performance.

Cultural background. The cultural background alludes to the shared beliefs, norms, customs,

values, behaviours, and practices shared by a group of people (Straub et al., 2002). The cultural

background influences how individuals from different groups perceive the world and their

behaviours towards businesses and customer services (Laroche et al., 2004). Assessing how

cultural backgrounds influence the adoption of AI assistants is critical in determining how

factors such as language, traditions, and identities impact the effectiveness of the technology

in enhancing customer satisfaction.

Western and Eastern cultures. These are broad generalisations of the cultural backgrounds

emerging in the West and East global geographical settings. Hong et al. (2007) state that the

categorisation of Western and Eastern cultures is a reflection of the diversity in beliefs, values,

practices, and behaviours in a multicultural world. Although globalisation and increased

socialisation have fostered increased cultural exchange, there are distinct characteristics

associated with Western and Eastern cultures. Nunn (2012) describes Eastern cultures as the

cultural traditions, values, and norms that emerge from the Asian and Middle East parts, while

Western cultures allude to those experienced in Europe and North America. Focusing on

Eastern and Western cultures is important to establish how AI shopping assistants deliver to a

multicultural customer base.

13
1.6 Literature Review Overview

The literature review presents a comprehensive and critical discussion of concepts, theories,

and findings from previous publications. The chapter covers the history and types of AI

assistants, the roles and challenges they present, empirical research on AI assistant adoption in

e-commerce, and cross-cultural perspectives on technology adoption and e-commerce. These

subtopics foster an assessment of the issues of interest highlighted in the research objectives

and questions. A brief history of AI assistants demonstrates a gradual development to

sophisticated modern-day systems and increased diversity in their capabilities and use. Since

the first chatbot, Eliza, was introduced in 1966 by Joseph Weizenbaum, there have been

substantial developments in AI technologies and specialisation in their functions, including

general-purpose conversational agents (e.g., Siri, Alexa), task-specific agents (e.g., customer

service chatbots), menu/screen-based agents, and more sophisticated text/voice-based agents

that require machine learning algorithms (Weizenbaum,1966; Aro, 2011; Guzman, 2017). AI

assistants have been applied in various sectors such as banking, healthcare, tourism, and e-

commerce (Siebra et al., 2018; López et al., 2017; Kalia et al., 2017; Ranoliya et al., 2017;

Dahiya, 2017). They perform a variety of tasks that include customer service, interacting and

offering users information, and enhancing the companies' performance (Okuda & Shoda, 2018;

Milhorat et al., 2019; Kowatsch et al., 2017; Oh et al., 2017; Argal et al., 2018; Solem, 2016).

Studies also highlight that AI assistants are effective in enhancing marking strategies and

customer satisfaction (Chung et al., 2018; Zarouali et al., 2018). The use of AI in e-commerce

is challenged by issues associated with the quality of services and user satisfaction, among

other limitations emerging in the use of digital systems. Some of the major concerns emerging

in empirical research concerning the use of AI assistants include user satisfaction, perceived

performance, trust, and corporate reputation (Ashfaq et al., 2020; Eren, 2021). These

considerations influence the users' intentions to continue using AI assistants. Perceived

14
usefulness is a strong predictor of user satisfaction and intentions to users (Lubbe & Ngoma,

2021). Cultural factors have an impact on trust and communication, which influence the

decision to engage with AI assistants (Nordheim, 2018). Therefore, cultural norms shape the

attitudes individuals have towards technology as personalisation becomes an Inherent aspect

of how AI assistants create competitive advantage (Kull et al., 2021; Merhi, 2021; Akour et al.,

2022; Peña-García et al. (2020). The key factors that influence the behavioural intentions to

use AI assistants addressed in the literature review are perceived usefulness (PU), perceived

ease of use (PEU), interactive communication (ICOM), personalisation (PERS), and attitude

(AT).

1.7 Research Method Overview

This section provides an overview of the research design and method used in the study. This

study is an explanatory research project that aims to explain the main aspects of the research

phenomenon. This study focuses on the fundamental constituents that constitute the research

model. Explanatory research is vital to uncover causal relationships, particularly investigating

'what is the impact?' by exploring cause-and-effect associations among specific research

phenomena (Bryman & Bell, 2015). Examining causal relationships within the primary

research clusters is also a central aspect of this study. Thus, the explanatory approach and other

research types, like exploratory and descriptive methods, are deemed relevant and appropriate

for this investigation. The research employs a mixed-method approach, incorporating both

positivism and interpretivism. Recently, mixed methods have gained popularity in social

science and management studies (Liu & Huang, 2023; Alakwe, 2017) for the following reasons.

Firstly, it allows researchers to draw upon the strengths of both quantitative and qualitative

methods, providing a more robust understanding of the phenomenon under investigation.

Combining numerical data and in-depth narratives enables a deeper exploration of complex

research questions. Additionally, the mixed-method approach helps researchers triangulate

15
findings, corroborating results from different sources, thus increasing the reliability and

credibility of the study. It also allows for a more holistic perspective, considering diverse

viewpoints and contexts, thereby enhancing the applicability of the research outcomes in real-

world scenarios. By employing mixed methods, researchers can tackle multidimensional

research questions, leading to more comprehensive and insightful conclusions that have a

broader impact on theory and practice. A mixed-method approach is adopted because it

provides a comprehensive understanding of this phenomenon, strengthens the validity of

findings, offers practical implications, and facilitates triangulation for more robust conclusions

(Dawadi et al., 2021). Meanwhile, there is limited mixed-method research on technology

adoption (Fan et al., 2016).

Figure 1.1. Research Design

The study began with a quantitative research method that used cross-sectional data collection

to test the hypotheses. The cross-sectional analysis aims to collect data simultaneously to

compare different groups of respondents (Belli, 2008). The study targets the e-commerce sector

and includes participants from different cultural backgrounds. Chapter 4 provides a deep

presentation of the research methodology employed in this study. The examination of the

16
relationships between various parameters in the research model was undertaken by using the

Partial Least Square Structural Equation Modelling (PLS-SEM) approach through the software

SmartPLS version 4 (Ringle et al., 2014). This technique allows us to examine complex

relationships between latent variables and facilitates comprehensive data analysis. The multi-

group analysis (MGA) was utilised to analyse the model between two groups of the collected

data. Furthermore, the qualitative research approach was employed to analyse reviews of the

Louis Vuitton brand's e-commerce applications, utilising machine learning and Natural

Language Processing (NLP) techniques to acquire deeper insights and construct a

comprehensive understanding of the phenomenon under investigation.

1.8 Thesis Layout

This section outlines the organisational structure of this thesis, detailing the chapters and the

primary contents presented within each chapter. This thesis comprises seven chapters: the

introduction, literature review, research methodology, statistical analysis and outcomes,

discussion, and conclusion.

Chapter 1 [Introduction] presents the thesis overview. This chapter starts with the background

of the research phenomenon and identifies the research knowledge gaps. The following

sections of Chapter 1 are the study's scope, the research questions, the study's objectives, the

significance of this research, and the research design. This chapter culminates with an

exposition of the research methodologies in this thesis and an outline of the thesis structure and

its main concepts.

Chapter 2 [Literature Review] critically examines the literature related to the research topic,

offering a comprehensive theoretical background for the primary research phenomenon. This

chapter begins with the history of AI assistant development. After that, this chapter presents

theoretical discussions on the roles of AI assistants in various domains and the factors

17
influencing AI assistant usage in the e-commerce field. Thus, this section of the thesis expands

upon the existing research factors and formulates pertinent dimensions. The study aims to

construct a comprehensive conceptual framework that addresses the research questions by

integrating these essential elements. The overarching goal of this chapter is to establish a solid

theoretical groundwork for the integrated research phenomenon. Building on this foundation,

the researcher can subsequently conceptualise the research framework, which will serve as a

fundamental structure for this investigation.

Chapter 3 [Theoretical Development and Research Model] delves into the theoretical

underpinnings and models that form the basis of the research model, aiming to bridge the

identified knowledge gap. Building upon the comprehensive literature review conducted in

Chapter 2, this chapter presents the proposed research model and its associated hypotheses.

Each idea integrated into the research model is substantiated by sound justifications, aligning

with the existing body of knowledge. Consequently, the output of this chapter is the

development of a well-defined conceptual model, which lays the groundwork for the

subsequent phases of the study.

Chapter 4 [Methodology] explains and justifies the core methodology used in this thesis,

including the research approach, the research design, and the research methods. The following

section of this chapter is the detailed questionnaire development, including the choice of the

design creation of the questionnaire structure and measurement scales. Chapter 4 also discusses

the data collection adopted in this study and ethical considerations. The chapter concludes with

details of the pilot study and the primary survey with the natural language processing

techniques for analysing online customer reviews.

Chapter 5 [Quantitative Data Analysis] and Chapter 6 [Qualitative Data Analysis] The data

analysis approach for this study involves employing PLS-SEM testing, which is justified based

18
on their suitability for investigating the research hypotheses. Chapter 5 presents both the

measurement model assessment and the structural model assessment. Chapter 6 provides a

detailed account of the results obtained through the machine learning approach. The study

analyses customer reviews using NLP technologies to gain further insights. The outcomes of

this NLP analysis are presented alongside the PLS-SEM results and assess whether they are

supported or rejected.

Chapter 7 [Discussion and Conclusion] discusses the study outcomes for each proposed

hypothesis. Chapter 7 critically discusses the research questions, investigating the relationships

between the research model factors and its hypotheses. Furthermore, this chapter also revisits

the research objectives and summarises the answers to the research questions, concluding the

findings of this study. The researcher has presented the study's main contributions and critical

managerial implications. Finally, this chapter concludes with the research limitations and

recommends future directions suggesting further research opportunities.

1.9 Chapter Summary

This chapter offers a comprehensive overview of the research, encompassing the research

background, problems, study scope, questions related to the research, objectives, and plan. This

chapter presented the research objective and questions with their research methods. In addition,

this chapter discussed the research contribution in terms of academic and practical significance.

Furthermore, this thesis's research approach and design have been explained and justified.

Finally, this chapter has demonstrated the overall structure of the thesis idea, covering the main

points presented within each chapter. Overall, the researcher provided Chapter 1 (Introduction)

to demonstrate the foundation for this thesis and based on this foundation. The next chapter

will discuss the concept of conversational AI agents in e-commerce and the influencing factors

that impact the usage of this AI application in terms of customers' perception and usage

intention. As with any research project, a level of iteration of research stages should exist to

19
demonstrate the consistency of the research process (Peffers et al., 2007). Therefore, this thesis

attempts to narrate the study's approach from the starting point to the completion point with a

reasonable level of implementation of each stage, presenting the consistency of the research

process and its logical flow.

20
Chapter 2. Literature Review

In this section, we explore the extensive literature surrounding AI assistants, establishing the

groundwork for the research model described in Chapter 3. The organisation of this chapter

comprises key segments: an initiation into the realm of AI assistants, encompassing their

definition, historical evolution, primary classifications, roles, and challenges; a scrutiny of past

empirical investigations concerning the utilisation of AI assistants, both in a general context

and specifically within the domain of e-commerce; and an examination of the factors

incorporated in the current study to probe into attitudes and the intention to use AI assistants in

the e-commerce area.

2.1. Introduction to AI Assistants

This section introduces AI assistants by describing the history of the development of chatbots,

their main types, roles, and the challenges of applying AI assistants.

2.1.1 History and Types of AI Assistants

In the early years of 1950, many studies brought attention to the relationship between

computers and humans based on communication theory. Since then, AI assistant platforms have

received immense interest. An article by Alan Mathison in 1950, titled "Computing Machinery

and Intelligence," offered a report based on the test of the computer's ability to think like

humans, which has served as the foundation for understanding AI agents (Pinsky,1951). Table

2 presents the history of the leading chatbot development. In 1966, Joseph Weizenbaum

developed the very first chatbot, Eliza. The primary reason for this development was to act as

a Rogerian psychotherapist. The chatbot was developed based on the model, which performs

by mirroring the prompt of the previous user. However, this model had some limitations as it

failed to keep up with the flow of ongoing conversation and was also limited when recognising

human-like feelings (Weizenbaum,1966). Later, in 1972, Kenneth Colby developed another

21
chatbot named Parry at Stanford University. This chatbot responded and played a role in

activating paranoid schizophrenia in a person. Depending on the embodied conversational

interface, Parry was more developed and had more features than Eliza (AbuShawar & Atwell,

2015).

The Artificial Linguistic Internet Computer Entity (ALICE) was created by Richard Wallace in

1995. To determine the heuristic dialogue rules, this entity used a platform called AI Markup

Language (Tabet et al., 2000). Later, in 2001, Windows created another platform called Smarter

Child and integrated it into Windows Messenger to interact with people as a customised

conversational interface. Apple introduced its platform in 2010, which operated as Siri, a voice

assistant. Siri uses voice and text interaction modes to interact with users in their native

language (Aro, 2011; Guzman, 2017). In 2015, based on Apple's model, Amazon launched its

voice-based assistant, Alexa, an intelligent speaker. Inspired by this model, in 2016, Google

also launched its smart speaker named Google Assistant. Samsung also adopted this trend by

developing Bixby, an intelligent voice-based assistant similar to Siri in 2016.

Table 2.1. Timeline of Chatbot Development

Chatbot Year Interaction mode Role

Eliza 1966 Text Rogerian psychotherapist

Parry 1972 Text Simulate a person with paranoid

schizophrenia
ALICE 1995 Text Practice human-like conversation

Smarter Child 2001 Text

Siri 2011 Text/Voice

Alexa 2015 Voice Virtual AI Assistant

Bixby 2017 Text/Voice

Meena 2020 Text

Blender 2020 Text

22
Google recently announced the launch of Meena, the most advanced AI assistant. Meena is an

end-to-end trained neural conversational model with 2.6 billion parameters. Compared to

earlier chatbots, this model facilitates more intelligent discussions (Adiwardana et al., 2020).

The creation of Blender, a more sophisticated AI assistant, has been unveiled by Facebook.

Compared to Meena, this assistant is anticipated to look more lifelike. However, Blender has

several limitations, such as its inability to filter Reddit datasets that contain objectionable

language that can affect Blender's replies (Gjurkovi & Najder, 2018).

Investigations on chatbots or AI assistants highlight the need for these technologies to facilitate

human-like engagements with customers. According to Davenport et al. (2020), AI assistants

can learn and respond to changes emerging among customers. Therefore, the customer

experience in using AI technologies is an underlying factor that impacts their attitudes and

intentions to continue using AI assistants in e-commerce. The particular characteristics

emerging from the different cultures justify an exploration of how different AI assistants are

adopted and implemented in e-commerce. Users can be described as individuals who routinely

utilise AI assistants in e-commerce, while non-users are those with limited or no experience in

the use of such technologies. According to Maedche et al. (2019), users are characterised by

positive attitudes towards AI technologies because they have adequate skills and knowledge to

leverage the benefits emerging from such solutions. However, non-users might be aware of the

AI technologies but lack insights on how to utilise the AI assistants to enhance their shopping

experience adequately. Lee et al. (2021) suggest that converting users to non-users is premised

on enhancing their experience and knowledge concerning AI assistants.

The AI assistants have been divided into two types according to their functions and engagement

styles. The first category hinges on discussing the various functions played by conversational

agents, while the second group concentrates on technical issues, such as defining user

interaction styles. Other frameworks are used in typical AI applications, such as an avatar, a

23
conversation system, or an expert framework for more effective question processing. Figure

2.1 presents the broad type of AI assistant classification. The general-purpose conversational

agent is referred to as an agent who is a personal virtual assistant and is also a multi-tasking

agent. Such an agent is usually integrated into smart speakers, mobile, and desktops (Siebra et

al., 2018). These applications are responsible for performing general tasks, and the users can

also ask public questions. For instance, the users can ask questions about the nearest restaurant

and weather, adjust the calendar, open email applications, or any other personal inquiries.

Among the examples of virtual assistants that play the role of general purpose are Alexa, Siri,

Cortana, Bixby, and Google Home (López et al., 2017).

Figure 2.1. Classification of AI Assistants

The task-specific conversational agent performs tasks specifically for the users as an assistant

agent for any particular domain, such as customer service, therapist, or online tutor (Kalia et

al., 2017; Ranoliya et al., 2017). Such agents perform on integrated platforms like social media

applications and websites. Facebook Messenger is among the popular virtual assistants used to

target consumers by many firms (Pereira & Díaz, 2018). In addition, WhatsApp, Kik

Messenger, Twitter, and WeChat application are also among the chatbot platforms (Xie et al.,

2019; Yamaguchi et al., 2018).

The menu/screen-based conversational agent performs using predefined rules through which

they can produce only a limited number of answers. The users are allowed to ask the predefined

24
questions only, in response to which the chatbot's knowledge base generates the solution. The

user interface of such a conversational agent has only end-user prompts in limited numbers.

Implementing such conversational agents is very straightforward and mitigates the need for

machine-learning algorithms for implementation (Dahiya, 2017). However, the menu/screen-

based conversational agent is limited when answering the questions not included in the

predefined list in the dataset. The benefits of such an agent are that it enhances the ease-of-use

factor and navigates the conventional flow of information (Hornbæk & Hertzum, 2017). Its

drawback of limited response results in the users' expressive capacity constraint.

One of the most advanced conversational agent types is the text/voice-based agent, which

requires an ML algorithm to generate appropriate responses. Such conversational agent

interacts directly with the end-users by allowing them to interact through unconstrained input,

either through voice-by-speech, typing text, or using long sentences during the conversation.

For instance, in case the user asks, "Where is the nearest saloon to my location?" the agent

takes the keywords "nearest," "saloon," and "location" to determine the reply to be given to the

end-user. Text/voice-based agents utilise deep learning algorithms to develop the ability to

recognise and detect keywords (Khanpour et al., 2016). Such conversational agents have a

dialogue manager central to the entire design because it collects the keywords from the

conversational interface sent to the knowledge engine. The knowledge engine is responsible

for classifying the questions and searching for appropriate responses from digital libraries

(Setiaji & Wibowo, 2016).

All these data sets are available as open platforms (Serban et al., 2015). Compared to menu-

based conversational agents, the text and voice-based agents allow the users to search with

more text/voice-based agents for more flexibility when expressing their prompts. However, this

may also lead to the misinterpretation of the prompt, which has a high chance of returning with

a negative response to the user, leading to a poor experience. The AI assistants have

25
applications in various domains like finance, travel, education, E-commerce, banking, and

healthcare. The following section discusses these applications.

2.1.2 Roles and Challenges of AI Assistants

Conversational systems have applications in several sectors, including healthcare, business,

education, travel, and finance. Table 2 summarises the AI assistant's role in various domains.

The conversational application aids learning by utilising the appropriate dialogue scenario

design in the educational sector, resulting in a less complex knowledge structure. Previous

studies have covered using AI assistants in the classroom and for student learning. For instance,

Sánchez-Daz et al. (2018) created a formal technique to support the university-level course

with a virtual assistant-based tutor. Clarizia et al. (2018) presented an ontology-based virtual

assistant in education. Using NLP techniques, this virtual assistant finds the keywords to give

pupils the correct answers. To assist teachers and students with learning tasks like spotting

spelling or grammar errors, reviewing homework, assigning projects, etc., virtual AI platforms

are helpful. In their research, Kerlyl et al. (2006) suggested that a negotiated open learner model

would be more advantageous for creating conversational agents and putting in place intelligent

tutoring systems to assist students in their learning process. The researchers proposed that

introducing a virtual assistant that has the ability not only to negotiate but also to incorporate

small talk can have a beneficial impact as it serves the purpose of enjoyment in engagement

and interaction with the chatbot, which has the capability of improving the students' experience

of learning. Hien et al. (2018) developed a virtual assistant to provide services for academic

staff and students. Such assistants have high user intent identification along with context

extraction accuracy. These virtual assistants show promising results regarding context

information and intent identification, i.e., if the context information extraction score is high, it

indicates that the virtual assistant can provide the correct answers.

26
AI virtual assistants work as customer service agents for large-scale inquiries for clients in the

financial sector, providing information about services like home loans, car loans, and FAQs for

those customers who own car loan contracts. Okuda and Shoda (2018), when examining the

features of the AI virtual assistant implemented in Sony Bank, found that it has developed a

function of the user stream, which visualises the number of users that have passed through

different contexts. Visualising the user stream function is essential as it helps gather

information about the script locations that require amendments regarding the detailed response

for developing the AI virtual assistant's conversation suitability. Altinok (2018) put forth a

framework for the banking and finance industry. The framework was created for German

language banking using the financial AI assistant to improve the dialogue between consumers

and the virtual assistant. Although this architecture showed promising results, work is currently

being done to introduce success metrics and assess the dialogue manager module. Duijst (2017)

presented an AI virtual assistant for the banks, which serves the purpose of investigating the

factors of personalisation, which can be beneficial in improving the user experience of chatbots.

The scholar also demonstrated no significant impact of personalisation on the user experience

of chatbots when it comes to the finance industry.

In the banking industry, conversational agents assist end users in various ways. These

responsibilities may include paying bills, responding to user account-related questions,

processing credit card payments, carrying out transactions, and setting up meetings. Milhorat

et al. (2019) study looked into the impact of AI virtual assistants in banking. The researchers

investigated it by creating a dialogue management system that can respond adequately and

reduce the incidence of any generic fallback. Two hundred twenty-six users participated in the

study, of whom 187 gave accurate answers, and 39 gave fallback utterances highlighting the

coherent statement response component.

27
Conversational agents are also used in the healthcare sector to inform patients and caretakers

about their medical issues (Kowatsch et al., 2017; Oh et al., 2017). Huang et al. (2018) created

a virtual AI healthcare assistant using user data. This virtual assistant can advise diabetic

patients on what foods to eat and stay away from as part of their diet. AI in medicine has

potential benefits since it enables personalised information delivery to individuals in various

practical contexts (Kocaballi et al., 2019a). However, additional research is still required to

guarantee the security of the patient (Laranjo et al., 2018). Recent studies have demonstrated

the potent influence of AI technologies in providing knowledge for averting the COVID-19

pandemic (Miner et al., 2020). This is made feasible by creating an AI-based chatbot that, in

addition to helping to battle COVID-19, also helped strengthen the influence of healthcare

management (Martin et al., 2020). An example is the creation of a web-based AI virtual

assistant was suggested by Espinoza et al. (2020) as a solution to the COVID-19 pandemic,

which is advantageous in filtering and rerouting consumers via the links dispersed over many

channels, such as email, social media, and text messages. The screening questions are handled,

and answers are generated to direct people to the best healthcare options.

The AI virtual assistant is vital in tourism and travel, providing services via conversation-like

interactions. The AI assistants are helpful in how consumers book their trips, discover new

experiences, plan their vacations, and make reservations at their preferred hotels. Argal et al.

(2018) developed an AI assistant for improving user-machine interaction in the travel industry.

This assistant is helpful as it collects the user data and preferences for generating the desirable

results and recommending accordingly to the user, along with providing accurate information

about their travel. Sano et al. (2018) also implemented a tourism AI virtual assistant based on

the agglomerative nesting algorithm and hierarchical cluster analysis, which balances the time

and quality of tours on different tourism sites.

28
Their primary objectives in terms of business are to increase sales and improve client

engagement and services (Solem, 2016). Combining AI technology makes it simple to achieve

these objectives (Brynjolfsson & McAfee, 2017). Researchers claim that Facebook chatbot

applications have a significant impact on brand engagement. According to studies (Leong et

al., 2018; Shareef et al., 2018), user-generated content and consumer participation on Facebook

have positively impacted consumer engagement with brands. According to Lee and Ko's (2019)

research, AI virtual assistants favour branding relationships, creativity, hedonic value, and

customising functionality. Bhawiyuga et al. (2017) created an artificial intelligence (AI)

assistant system to engage with consumers using the Telegram service and automatically

generate responses to customer-to-seller inquiries within 5 seconds.

AI virtual assistant plays many vital roles in the e-commerce domain, including asking for

customer feedback, sending messages for advertisement, collecting data regarding the

customer's preferences, and improving the consumer's engagement with the brand. Two studies

support the opinion that chatbots play a vital role in enhancing the marketing strategies of

brands and leading to customer satisfaction (Chung et al., 2018; Zarouali et al., 2018).

However, there is still a need to investigate the practical impact of using AI virtual assistants

in e-commerce as a marketing channel and consider all the minor details, like the customer's

perception. AI assistant plays the role of enhancing the online shopping experience of the

customers as it provides relevant recommendations and information for the customers online.

This improves the browsing of the customers' products, as finding the preferred outcome

among the massive variety of products on the website can be much more time-consuming and

challenging. For example, a website-based chatbot developed by Gupta et al. (2015) works as

an online automated assistant by suggesting products that match customers' preferences.

Today, almost all the existing brands depend upon social media sites to improve customer-

brand relationships and distribute information to customers. However, there is a limitation to it

29
as social media sites contain content in massive amounts, and a poorly generated post can lead

to increasing difficulty for customers in finding information about the brand quickly. In such a

case, a conversational AI agent is influential as it helps the customers engage with the brand

and find their information. However, there are still many challenges when it comes to

designing, making conversations, and evaluating the satisfactory experience for the customers

(Kocaballi et al., 2019b). Despite this, AI virtual assistant shows improved results regarding

customer service technologies and marketing channels. The AI virtual assistants already

existing in the marketing channels have several cons, including the need to support third-party

integration, poor interactive user interface, and multilingualism. They cannot detect customer

emotions (Nuruzzaman & Hussain, 2018).

Some recent studies have investigated and sided with the effectiveness of using AI virtual

assistants for brand strategies and online marketing (Chung et al., 2018; Zarouali et al., 2018).

Also, much research is being carried out to determine the potential of online AI assistants when

working as an online marketing tool. These studies also focus on the enhancement of customer

engagement using conversational agents. Further, in the future, these researchers might shift

their focus to investigating the customer's attitude toward the involvement of the conversational

agents in marketing along with the factors that influence the user's satisfaction with the

conversational agents. Currently, consumer experience quality assessment is a trendy topic

under the HCI discipline (Kocaballi et al., 2019b). Further, there is an area of improvement in

the design of the conversational application depending on the feedback of the customer's

experience. Also, certain areas of the conversational agents that need improvement can be

found through the customer's perspective to enhance the conversation's outcomes and

capabilities.

30
2.2 Empirical Research on AI Assistant Adoption in E-commerce

This section elaborates on studies investigating AI assistant use in e-commerce to provide a

comparative analysis of related work and results. Identified pertinent studies through electronic

databases, including Scopus, Web of Science, and Science Direct (Ashfaq et al., 2020;

Kasilingam, 2020; Pillai et al., 2020; Zarouali et al., 2018). To gain comprehensive insights

into the topic, the subsequent sections encompass research on the utilisation of a conversational

AI agent in e-commerce domains. The objective was to establish a suitable theoretical and

empirical foundation for constructing the research model in this study and pinpoint the most

relevant factors for investigating attitudes and behavioural intentions to use AI assistants in e-

commerce.

Previous studies show that user satisfaction is inherent in developing the intention to use AI

assistants. Findings by Ashfaq et al. (2020) show that user satisfaction with chatbot e-services

was a key determinant and predictor of continuous intention to use, highlighting the need to

improve the quality of services and information offered to consumers. The authors

recommended the integration of human service employees to enhance user satisfaction with

digital services. Aligned with these findings, Luo et al. (2019) found that in cases where chatbot

identity was disclosed to the customers before a conversation, there was a 79.7% decrease in

purchase rates, which shows that customers value human interactions compared to chatbots.

According to Eren (2021), customer satisfaction in using chatbots in the banking sector is

influenced by perceived performance, trust, and corporate reputation. Insights from Lubbe and

Ngoma (2021) show that perceived usefulness (PU) is the strongest predictor of satisfaction

among customers using chatbots. Kim and Chang (2020) found that the chatbot service quality

does not affect user satisfaction and reliability. Further insights from the authors show that the

achievements made in satisfaction, reliability, and immersion when using chatbots influence

the customers' reuse intentions.

31
The perceptions users have concerning chatbots influence the decisions made by an individual.

According to Araujo (2018), chatbots that use human-like language or names effectively

develop the perception that the systems could satisfy user needs like human workers. However,

Go and Sundar (2019) stress that although chatbots increasingly replace humans in online

customer services, their interactions are often machine-like. According to Sheehan et al. (2020),

anthropomorphic chatbots can lead to enhanced customer satisfaction because they can meet

their social desires for human transactions.

The customers' intention to use AI assistants is influenced by access to information that can

enhance their understanding and perceptions of the technology. According to Rese et al. (2020),

the intention to use chatbots is influenced by the customer's insight, including utilitarian aspects

such as the authenticity of conversation between the users and the chatbot and the perceived

usefulness established by individuals. The researcher adds that the intention and frequency of

use can be impacted by the customers' concerns about privacy when using the system. Van den

Broek et al. (2019) show that the perceived helpfulness and usefulness of the chatbot influence

the perceived intrusiveness of chatbot advertising and patronage intentions. This implies that

individuals with adequate access to information, including the chatbot capabilities, can make

informed decisions on its use. Table 2.2 synthesises the key factors affecting the intention to

adopt AI assistants.

Previous studies show that the increased application of chatbots in e-commerce has contributed

to its acceptance with improved customer experience. Moriuchi et al. (2021) show that users

develop a positive attitude toward using chatbots, leading to increased engagement, while

higher user satisfaction enhances the revisit intention on the online platforms. Sfenrianto and

Vivensius (2020) show a positive correlation between customer experience and an array of

factors, including information quality, system quality, service quality, e-trust, e-satisfaction,

and e-loyalty. These findings are supported by insights from Pillai et al. (2020), who show that

32
the innovativeness and optimism among consumers influence the perceived ease and usefulness

of AI-powered systems in the retail sector. Furthermore, Pillai et al. (2020) established that

insecurity in using AI negatively impacts the perceived usefulness, which, when integrated with

perceived ease of use, enjoyment, customisation, and interactivity, are predictors of the

customers' intention to shop in AI-powered automated stores.

Studies demonstrate that anthropomorphism in using chatbots is a critical consideration in

developing brand engagement and purchase decisions. Research by Han (2021) shows that

human-like features in online shopping and customer service assistants improve customer

engagement and the subsequent intention to purchase. In the same context, Kull et al. (2021)

show that the increased use of chatbots in customer services requires effective messaging

tailored to individual visitors, enhancing customer-brand connections.

Although chatbots have enhanced customer service and support, various considerations are

critical in their effective implementation. Nordheim (2018) shows that the effective use of

chatbots requires developing trust among users. Additional considerations include chatbot-

related factors such as expertise, absence of marketing, fast response, and anthropomorphism,

which are characteristics of an effective chatbot. The researcher also established that access to

human operators enhanced the use of chatbots because it increased trust in the technology.

Mimoun and Poncin (2015) demonstrate that using embodied conversational agents (ECA)

requires improving the users' perceptions of shopping value to enhance their experience.

According to Selamat and Windasari (2021), the strategic objective in adopting chatbots is to

create a positive customer experience, especially when implemented in SMEs. This can be

achieved by developing responsive systems, triggering customer actions, and fostering

humanised conversations and personalisation. Li et al. (2020) found that AI customer services

enhance users' online shopping experience. However, there are challenges in fostering a

seamless bridge between technology and human agents. The use of AI systems is also

33
influenced by customer attitudes that vary among customers with different characteristics.

Therefore, businesses must foster a strategic and orderly approach to using human agents

versus AI chatbots. Rhee and Choi (2020) suggest that customers can be more inspired to use

chatbots with voice-based conversational agents that enhance personalisation. In this light, the

personalisation and social aspects of the chatbots enhance the positive attitudes among

customers promoting their application in shopping.

Table 2.2. Key factors affecting the adoption of AI assistants

Key Factor Study

Perceived Usefulness, Perceived Ease of Use, Perceived Balakrishnan and Dwivedi, 2024
Anthropomorphism, Perceived Intelligence, Perceived Animacy

Perceived Competence, Trustworthiness Choudhury et al., 2024

Personalisation, Conversational Tone, Autonomy, Responsiveness Guo and Luo, 2023

Usefulness, Ease of Use, Anthropomorphism, Sociability, Singh et al., 2024

Enjoyment, Privacy, Trust

Information Quality, Service Quality, Perceived Enjoyment, Ashfaq et al., 2020

Perceived Usefulness, Perceived Ease of Use

Human-like Cues, Framing, Anthropomorphism Araujo et al., 2018

Responsive Features, Social Conversations, Personalized Selamat et al., 2021

Recommendations

Authenticity of Conversation, Perceived Usefulness, Perceived Rese et al., 2020

Enjoyment

Initial Message, Brand Self-distance, Miscommunication, Kull et al., 2021

Anthropomorphism

Message Interactivity, Social Role, Involvement Go and Sundar, 2019.

Rhee and Choi, 2020

Information Quality, System Quality, Service Quality, e-Trust, e- Sfenrianto and Vivensius, 2020
Satisfaction, e-Loyalty

Expertise, Anthropomorphism, Response Time Nordheim, 2018

34
Previous studies employed different methodologies and explored factors associated with

applying AI assistants (Balakrishnan & Dwivedi, 2024; Choudhury et al., 2024; Guo & Luo,

2023). The studies also provided contradictory evidence on factors influencing AI assistant use.

Table 2.3 provides a summary of recent empirical studies on AI assistant use. In contrast,

certain aspects of technology adoption models have not been examined within the realm of AI

assistant utilisation in e-commerce (Singh et al., 2024; Araujo et al., 2018; Selamat et al., 2021).

Furthermore, previous research on AI assistants has predominantly focused on Western

countries or specific cultural contexts (Istiqomah & Alfansi, 2024; Kull et al., 2021; Sheehan

et al., 2020; Mimoun & Poncin, 2015), making their findings potentially less applicable to

individuals in different contexts due to cultural and other variations. Notably, entire regions

have been excluded from prior investigations, or studies have predominantly explored

countries diverging from the Eastern context and culture (Li et al., 2020; Pillai et al., 2020;

Kull et al., 2021). Therefore, the current knowledge gaps are:

- Not enough study on how people adopt technology: Researchers haven't closely examined all

aspects of how individuals decide to use AI assistants in e-commerce contexts.

-Some countries and cultures are excluded: Most studies have focused on AI assistants in

Western countries or specific cultures. This might mean their findings don't fit well for users

or potential users in other places.

-Looking at how different users use technology: More research is needed to compare how users

and non-users perceive AI assistants in online shopping.

Thus, the aim of this research is to address these gaps by examining distinctions between

Western and Eastern consumers. The following sections elaborate on the cultural differences

and cross-cultural perspectives on technology adoption and e-commerce.

35
Table 2.3. Summary of Recent Empirical Studies on AI Assistant Use

Study Purpose Method Findings Limitations

Balakrishnan Explore factors TAM model and SEM for Perceived ease of use, Non-random
and influencing users' data analysis. perceived usefulness, sampling and
Dwivedi, attitudes and purchase 440 participants in India perceived focusing only on
2024 intentions towards with prior experience anthropomorphism, Indian users.
digital assistants. using digital assistants perceived intelligence,
and perceived animacy
all had a positive impact
on users' attitudes
towards digital
assistants. Users'
positive attitudes also
increased their purchase
intentions.
Choudhury Investigate how users An online survey was Perceived competence The data was
et al., 2024 perceive and use conducted with adults in and trustworthiness of collected only from
ChatGPT in healthcare the United States who used ChatGPT are crucial in U.S. adults, and
decision-making. ChatGPT. 607 healthcare. Security and only a small
respondents. persuasiveness are not number of them
related to healthcare used ChatGPT.
contexts. Non-random
sampling and
reliance on self-
reported data.
Guo and Examine factors Offline survey. Personalisation, Focused only on
Luo, 2023 influencing purchase 428 valid questionnaires conversational tone, Generation Z
intention in intelligent were collected. autonomy, and consumers in
personal assistants. responsiveness China.
positively impact
informational and
emotional support.
Informational and
emotional support
positively affects
purchase intention.
Singh et al., Examine acceptance of Used the TAM Model. Usefulness, ease of use, Focused on
2024 Online Shopping 272 participants completed anthropomorphism, accepting assistants
Assistants in e- the online survey. sociability, enjoyment, in low-risk
commerce privacy, and trust consumer products.
significantly impact the
acceptance of online
shopping assistants.
Ashfaq et Drivers of users' Model: Expectation- Information quality and Did not include
al., 2020 satisfaction and confirmation model, service quality positively marketing and
continuance intention information system influence satisfaction. communication
toward chatbot-based success, and TAM Perceived enjoyment, factors. Limited
customer service. Country: United States perceived usefulness, sample.
Sample: 370 actual users and perceived ease of
from Amazon's use are significant
Mechanical Turk. predictors of
continuance intention.
Araujo et al., How human-like cues Model: Experimental The study established The study uses a
2018 such as language style design that a human-like agent small sample that is
and name and the Country: Netherlands was associated with limited to
framing Sample: 207 Facebook higher levels of mindless Facebook users.
used to introduce the users and mindful
chatbot to the customers anthropomorphism than
influences their a machine-like agent.
perceptions of the social
aspects,
mindful and mindless
anthropomorphism

36
Selamat et Determining the chatbot Model: The critical The study lacks to
al., 2021 features and elements Anthropomorphism, considerations for consider the
that are appropriate perceived enjoyment, chatbots are responsive diversity among
based on customers' perceived ease of use, features, simple steps to customers and their
characteristics perceived usefulness, trigger customer actions, requirements to
Country: social conversations, and meet customer
Sample: 315 customers personalised needs in different
recommendations. industries.

Rese et al., Contrasting the Model: TAM and uses and The acceptance of the The research used a
2020 technology acceptance gratifications theories chatbot was influenced small sample that
model (TAM) with the Country: Germany by utilitarian factors reflects the German
uses and gratifications Sample: 205 German such as the authenticity setting.
theories in determining Millennial of the conversation and
the acceptance of the perceived usefulness, as
text-based "Emma" well as hedonic aspects
chatbot among target like perceived
consumers enjoyment. There was an
insignificant deviation
between the models.
Van den Predicting perceived Model: Multiple linear The perceived The study used a
Broeck et intrusiveness of chatbot regression intrusiveness of chatbot small research
al., 2019 advertising based on the Country: advertising relies on the sample and was
perceived helpfulness Belgium/Netherlands users' perceived limited to
and usefulness and Sample: 245 Facebook helpfulness and Facebook users.
subsequent influence on users between 18 and 35 usefulness.
patronage intentions years

Kull et al., To determine a strategic Model: Experimental tests Chatbots start The small sample
2021 approach for managers Country: United States conversations using a and geographical
to tailor the initial Sample: 82 university warm message that limitations show
message of the chatbot students increases engagement, that the findings
to enhance consumer- and the message can be are not universal.
brand connections and mediated by brand self-
engagement distance that allows
individuals to feel closer
to the brand.
Sheehan et Relationship between Model: None Lack of effectiveness in The study uses a
al., 2020 miscommunication and Country: United States addressing small sample and
adoption of customer Sample: 200 Americans miscommunication focuses on the
service chatbots from Amazon's errors reduces American context
Mechanical Turk (MTurk) anthropomorphism and
adoption intent
Go and To establish what can be Model: Factorial There is a high level of The study relies on
Sundar, done to humanised experiment message interactivity a small population
2019 chatbots Country: United States that compensates for the sample
Sample: 141 e-commerce impersonal nature of a
customers who interacted chatbot that is low on
with chatbots anthropomorphic visual
cues
Rhee and Determining the Model: Dual modes of There is a significant The study relied on
Choi, 2020 persuasion mechanism information-processing interaction effect for the a small sample.
used in product models social role and
recommendations made Country: South Korea involvement when using
by a voice-based Sample: 122 voice-based
conversational agent conversational agents.
and the impact
personalisation of
content has on
consumers' attitudes in
voice shopping.

Lubbe and How perceived ease of Model: TAM, exploratory Perceived ease of use, The sample
Ngoma, use, perceived factor analysis and perceived playfulness focused only
2021 playfulness, and multiple regression and perceived usefulness focused on
perceived usefulness of analysis significantly and millennials in
chatbots influence Country: South Africa positively impact South Africa.

37
 customer experience Sample: 333 South African customer experience and
and satisfaction in millennials satisfaction when using
emerging markets chatbots.

Moriuchi et The consumers' attitude Model: The theory of A positive attitude The study uses a
al., 2021 and engagement with conversation and partially towards using small sample of
chatbots in customer observed Markov decision technology leads to millennial
service in a retail process increased engagement consumers and
environment Country: United States and high satisfaction, lacks adequate
Sample: 68 respondents increasing the revisit insights into the
from a private university intention. role of technology
engagement using
different
technological
solutions in various
types of businesses,
such as business-
to-business
Sfenrianto Determining the factors Model: IS Success Model The correlation between The study uses a
and, that affect customer Country: Indonesia Information quality, small sample from
Vivensius experience when using Sample: 385 individuals system quality, service Indonesia and lacks
2020 e-commerce services who engaged in active quality, e-trust, e- to address diverse
that have a chatbot that transactions in e- satisfaction, and e- e-commerce
helps individuals to do commerce platforms that loyalty influences services such as
transactions have implemented chatbot customer experience in banking.
technology e-commerce that uses
chatbots
Istiqomah Expolre the role of 270 respondents who made Perceived benefit and Focused on
and Alfansi, attitude towards the use at least two fashion perceived ease of use consumers in
2024 of AI-enhanced fashion purchases online. positively influence Indonesia’s fashion
e-commerce actual use. Perceived e-commerce sector.
benefit and perceived
ease of use positively
affect attitude towards
use.
Han, 2021 Assessing the impact of Model: None Anthropomorphism is The study uses a
anthropomorphism on Country: United States critical in small sample based
consumers' perceptions Sample: 170 university shaping the consumers' in New York.
concerning mobile students with mobile intentions to purchase
messenger chatbots and phones from businesses using
its impact on chatbots
behavioural decision-
making
Eren, The impact of Model: None Customer satisfaction in Limited research
(2021). perceived trust in Country: Turkey chatbots is significantly samples from
chatbots and reputation Sample: 240 customers affected by perceived Turkey.
on customer satisfaction who transacted using a performance, trust, and
chatbot corporate reputation.
Mimoun and Determining how the Model: Embodied It is imperative to The findings of the
Poncin 2015 use of ECA can conversational agents account for utilitarian study are
improve the users' (ECA) usage and hedonic features to applicable in the
perceptions concerning consequences: playfulness, understand ECA outputs French context.
shopping value and the decision quality, and social in e-commerce sites.
consequences for their presence
purchase intentions and Country: France
satisfaction with the Sample: 576 French
website consumers
Nordheim Determining factors that Model: Framework on The users' trust is The limitations
2018 affect trust in chatbots trust in websites affected by expertise, emerge in using
Country: Norway anthropomorphism, low questionnaires in
Sample: 154 users of risk and not trust data collection,
customer service chatbots relevant/no trust, fast which could have
response, absence of introduced bias and
marketing, brand, and interpretation
access to a human inaccuracies.
operator.

38
 Kim and The impact of chatbot Model: IS success and User satisfaction is not The study involved
Chang, 2020 service quality on user service quality model affected by the chatbot a small sample, and
satisfaction and Country: Korea service quality. Reuse the findings are
reliability Sample: 218 users in their intention is influenced limited to the
teens and 70s who had by satisfaction, Korean context.
experience using chatbot reliability, and
services immersion in chatbot
services.

2.3 Cross-Cultural Perspectives on Technology Adoption and E-commerce

Culture plays a critical role in the adoption of new technologies. Various conceptualisations

and perceptions demonstrate how cultural factors impact the adoption of new technologies in

the e-commerce context. The cultural background influences the behavioural intentions that

emerge among a group of people concerning the use of particular technologies (Faqih, 2016).

The culture shared among the target customer base can help determine how individuals make

decisions concerning the use of particular technologies (Yi et al., 2016). Understanding culture

sets the premise for designing AI solutions that can adequately meet the specific needs and

preferences of different customers. According to Lin and Hsieh (2007), the impact culture has

on behavioural intentions reflects how technologies that satisfy individuals from a particular

group can be designed, adopted, and implemented to benefit individuals and the companies

involved.

According to Merhi (2021), the variations that emerge among cultures and societies in using e-

commerce can be attributed to people's cultural values, beliefs, norms, and attitudes toward

new technologies. For instance, Akour et al. (2022) established that people's trust in e-

commerce varies among people from different cultures, which impacts adoption and

widespread usage. In this context, some cultures value face-to-face interactions and human

relationships with companies, which limits their uptake of e-commerce services. Drawing from

Hofstede's cultural dimensions, Faqih (2022) show that the power distance dimension affects

the trust and confidence individuals from some cultures have towards online transactions.

Filieri and Mariani (2021) argue that cultures characterised by low power distance are more to
39
explore innovations and smaller e-commerce platforms, while high-power distance cultures are

more attracted to established brands and online platforms. Mou et al. (2020) argue that

individuals from cultures with high uncertainty avoidance are characterised by risk-averse

behaviours in using new technologies, which implies that e-commerce is more likely to

proliferate in cultures that demonstrate low uncertainty avoidance. According to Li and Wang

(2023), adopting e-commerce among different cultures is subject to people's preferred

communication styles. The study suggests that some cultures are interested in services that

entail explicit and direct communication, such as straightforward descriptions of products and

transparency in transactions. In contrast, in some cultures, such considerations might not

impede the adoption of e-commerce. Sobti (2019) found that payment methods diver across

cultures, considering that in some countries, the available options and infrastructure facilitate

seamless electronic transactions such as credit cards and mobile payments, while in others,

there is a high preference for using cash.

An emphasis on communalism characterises Eastern cultures through the interconnectedness

that exists among the people. In the East religions that include Buddhism, Hinduism, and Islam

have a significant influence on values and practices (Mokhlis, 2009). Western cultures embrace

individualism that is characterised by autonomy and self-expression (Cherrier, 2007).

Therefore, cultural backgrounds influence how individuals perceive technology as a

community and in their individual contexts. Cultural differences influence the attitudes that

exist among individuals in the use of e-commerce. Peña-García et al. (2020) suggest that

cultural norms and values impact the attitudes that build up toward using technology,

considering that some societies value the stability associated with their traditions. In contrast,

others have become more open and accept innovation and technological advancements. A study

in Taiwan by Daowd et al. (2021) established that local languages were a critical consideration

among online consumers when adopting online shopping platforms. Therefore, e-commerce

40
applications developed in the local language were more likely to be accepted and used by

consumers due to the enhanced ability to understand information and communicate with the

service or product providers (Elsholz et al., 2019). Chawla and Kumar (2022) noted that e-

commerce platforms that demonstrate high levels of information security are more likely to be

embraced in cultures where trust is inherent in the relationship between the companies and their

consumers. The attitude towards e-commerce is also shaped by social influences whereby

word-of-mouth and fear of missing out (FOMO) play a critical role in how individuals engage

online (Kamalia et al., 2022; Rana & Arora, 2022).

Cross-cultural investigations addressing the use of AI assistants are critical to understanding

how consumers from different cultural backgrounds interact with the technologies and how

such interactions are influenced by their characteristic values, preferences, and norms.

Subsequently, the insights gathered can help technology develop to adapt the business

processes, products, or services in alignment with the expectations and needs of the target

markets. According to Nam et al. (2021), personalising online services is challenging because

it requires an in-depth understanding of the target consumers to enhance their experience and

create a competitive advantage. Chi et al. (2023) recommended that in a diverse global virtual

market, it is critical to understand cultural nuances to ensure that the services offered resonate

with users' beliefs and values. These insights reflect sentiments expressed by Tsai et al. (2021),

who highlight that ethical considerations should be embedded in using AI assistants to ensure

an enhanced experience from virtual customer care agents similar to what individuals would

experience when dealing with a human representative. Insights from Skjerve et al. (2023) show

that cross-cultural perspectives in developing AI assistants are necessary in the health sector to

address language barriers and improve communicative processes in the services offered. Such

considerations can enhance customer satisfaction and create opportunities for expansion in new

markets.

41
The potential implications of cultural differences in the acceptance of AI assistants among

online consumers are reflected in individuals' trust and confidence that the technologies can

meet their needs and expectations compared to human representatives. Furthermore,

Shneiderman (2020) stressed that the successful application of AI technologies in commercial

services must demonstrate high reliability, security, and privacy and meet customer needs and

expectations. AI developers must offer attractive benefits in the context of personalisation in

terms of language, user interface, and design. Aligning such features with the cultural

preferences of the target market can increase adoption and value for the companies and

customers. Additionally, the relevance of culture in e-commerce shows that the AI assistants

have to express social and emotional intelligence to establish interpersonal relationships and

emotional connections reflected where people are involved in the communication processes

(Drigas et al., 2023). Therefore, cultural sensitivity is at the heart of developing AI assistants

that can adequately engage with people from diverse backgrounds, considering that consumer

attitudes, perceptions, and communication between Western and Eastern cultures vary

significantly.

Despite the growing body of literature on how cross-cultural differences impact the use of AI

assistants in e-commerce, further research is needed to address the trends and changes in the

use of these technologies over time in different cultures. There is also a need to compare Eastern

and Western cultures in the context of cultural dimensions such as power distance to highlight

how cultural nuances impact the adoption of AI assistants. Further research can address

multilingual AI assistants' impact on e-commerce and establish if cultural sensitivity and ethical

considerations are integrated when using different languages to engage with a diverse customer

base. The subsequent sections provide detailed insights into the factors employed in the current

study to examine the intention to use AI assistants among consumers in both Western and

Eastern contexts.

42
2.4 Factors Impact Behavioural Intention to Use AI Assistants

In the preceding sections, elements subjected to both theoretical and empirical analysis were

identified, shedding light on how consumers make decisions regarding the adoption of

innovative technologies, such as AI assistants, across various contexts and domains. Further

exploration of previous research on the use of AI assistants and related areas is carried out

below to discern and delve into the most pertinent factors for examining the current research

topic.

2.4.1 Perceived Usefulness (PU)

Perceived usefulness is a critical aspect of technology acceptance and adoption. It highlights

the extent to which individuals believe that a particular technology can help them by improving

their performance, productivity, and effectiveness in meeting specific objectives or completing

a task (Tahar et al., 2020). From this perspective, PU in AI assistants can be described as the

perception that technology can add value or benefit the user engaged in e-commerce. This

implies that PU is a subjective evaluation by the user shaped by the experiences and

expectations concerning the capabilities of the AI assistants. De Kervenoael et al. (2020)

suggest that the use of technology among people is influenced by its relevance to their tasks.

Therefore, the perception that AI assistants can support or enhance the user's experience

increases their likelihood of adopting it. McLean et al. (2021) show that PU is linked to task

efficiency, and individuals are more likely to use AI assistants if they believe it can simplify

complex issues. In this context, if individuals feel that the AI assistant leads to substantial

improvements compared to alternatives such as human actors, then its PU increases. Further

insights from Shiau et al. (2020) show that PU improves the intention to use a technology where

there is a clear understanding of the advantages and benefits it brings in performing a particular

task. Learning and understanding the features and functionalities of the AI assistant is critical

in how individuals consider it a helpful tool. Despite this, there is a lack of adequate insights

43
addressing how the AI assistant's PU contributes to the increased adoption of e-commerce

among consumers from different cultural backgrounds. Addressing the PU concept from this

perspective can enhance how developers and e-commerce companies implement AI assistants

that add value to how consumers utilise online platforms.

2.4.2 Perceived Ease of Use (PEU)

The PEU alludes to the subjective assessment of individuals using or learning a particular

technology (Tahar et al., 2020). From this perspective, PEU, in the context of AI assistants,

highlights the users' feelings concerning the overall user-friendliness of the features and

functions of the technology. According to Sarkar et al. (2020), the PEU is influenced by the

user interface design that facilitates interaction between the user and the system. Alves et al.

(2020) suggest that PEU can be enhanced when the user interface is intuitive, visually

appealing, and informative for more efficient navigation and user interaction. Selamat and

Windasari (2021) note that PEU influenced online platform use because it determined how

quickly users learned and effectively used the features, including AI technologies.

Suggestively, a complex and challenging system for users to learn can discourage the continued

use of the technology. Yen (2023) argued that a high PEU among users fosters a seamless and

coherent interaction that is critical in meeting customer satisfaction in the facilitation offered

by online platforms. Focus on improving PEU for users engaging with AI assistants can

enhance their ability to perform tasks efficiently and limit the frustrations associated with using

such features in e-commerce (Gümüş & Çark, 2021). Al-Adwan et al. (2023) recommend

implementing new features on online platforms, such as AI customer services, which should

be associated with facilitation in training and provision of adequate information to enhance

PEU. Harrigan et al. (2023) established that PEU in using online technologies can be influenced

by social factors such as peers' recommendations and the users' technological capabilities.

However, there is a research gap in how PEU shapes the decision to use AI assistants in e-

44
commerce, considering the technological disparity and the differences in digital skills among

people from different cultural settings.

2.4.3 Interactive Communication (ICOM)

According to Billion and Van den Abeele (2007), ICOM is the communicative process that

entails the active participation of two or more parties in a dynamic and bidirectional exchange

of information, including real-time responses and feedback. ICOM can be experienced through

various channels, such as face-to-face interactions, phone calls, video conferencing, social

media, and chatting on online platforms. Kim et al. (2023) demonstrated that the effectiveness

of chatbots in online platforms can be attributed to the real-time communication between the

system that represents the company and its users. This interaction facilitates timely responses

and access to customer needs information. Therefore, customers are more attracted to systems

where ICOM makes them key stakeholders in determining the quality of services offered in a

personalised context (Choi et al., 2020). Dwivedi et al. (2022) highlighted that ICOM allows

active engagement of all parties, which creates an opportunity for developing inclusive services

where all users' opinions, concerns, and interests are considered in the services offered. Cheng

and Jiang (2022) found that the immediate feedback from chatbots contributed to the customers'

intention to use online services when seeking information about the company offerings and

positively impacted their decision to purchase. These insights show that ICOM promotes

positive attitudes towards AI technologies, companies, products, and services. Furthermore, an

effective ICOM gives customers an enhanced understanding because they can seek clarification

and additional information to make informed purchase decisions. Schanke et al. (2021) found

that the increased use of chatbots compromised customer satisfaction with the company

because it eliminates the human touch, which is critical in developing healthy customer

relationships. The research gap in this context is that despite ICOM being a valuable form of

communication, there is a need to establish the various underlying factors that the AI assistants

45
must meet to enhance the relationship with the e-commerce companies to achieve better

outcomes than in interactions with human representatives.

2.4.4 Personalisation (PERS)

Hsu et al. (2021) highlight that each consumer is unique, making it imperative that companies

consider tailoring products and services to meet their needs and characteristics. Chandra et al.

(2022) described personalisation as providing the right services or products at the right time,

in the right place, and to the right person. In the e-commerce context, Selamat et al. (2021)

suggest that personalisation is premised on accessing, analysing, and acting on large volumes

of consumer data that might be historical or real-time and using it to understand and predict

customer behaviours with the intent of offering services or products they can relate with. A

study in the banking sector by Kaaniche et al. (2020) found that customers are substantially

attracted to online banking providers that demonstrate enhanced capacity to offer quality,

secure, and personalised services. Saniuk et al. (2020) supported these findings by stating that

modern consumers expect personalisation in online services. Companies with online operations

have responded to these expectations by implementing sophisticated technological solutions

such as AI and ML technologies, enabling them to respond to each customer’s needs adequately

and effectively. These developments are aligned with the increased focus on the data-driven

approach in decision-making that entails using customer data from the organisation and other

sources to track customer behaviours and preferences, which enhances the ability of the

companies to offer a unique experience for all customers. The research gap emerging in this

context occurs in the lack of adequate insights into how personalisation is considered an

imperative across different cultures and how the AI assistants used in e-commerce are designed

to foster it premised on each user's unique characteristics in a highly diverse customer base.

46
2.4.5 Attitude (AT)

The attitude individuals have towards a product or service has a substantial impact on their

decision-making. Jain and Weiten (2020) describe attitude as a psychological construct that

reflects individuals' feelings and beliefs when evaluating a particular service, product, event,

phenomenon, or entity. Attitudes shape human behaviours and how individuals make decisions.

Nevertheless, attitude is influenced by various factors such as the environment, social contexts,

and access to information. Insights from Araujo et al. (2020) show that attitude can be positive

or negative based on how an individual evaluates the subject. Positive attitudes are favourable

toward the issue, while negative attitudes are unfavourable. Duffett (2020) attributes the

varying attitudes that emerge among people to an individual's affective, cognitive, and

behavioural aspects. Focusing on adopting and using new technologies, Xiao et al. (2019)

determined that attitudes tend to be consistent among people with similar characteristics.

However, some individuals within a group can change their attitude due to their experience or

access to new information. Furthermore, Ajzen (2020) argues that in the context of the theory

of planned behaviour, attitudes are subjective, which implies that the perceptions held by one

person concerning a particular subject can differ significantly from those of another person. In

the cultural context, Rawlings (2020) suggests that attitude can be deeply ingrained in one's

beliefs that emerge from cultural influences. Such individuals are more likely to practice

behaviours that align with their positive attitudes towards something while avoiding what

contradicts their cultural beliefs. The gap in research emerging in this context is the lack of

adequate insights that reflect how the Eastern and Western cultures and belief systems shape

individuals' attitudes towards using AI assistance in e-commerce. Understating culture's

influences on beliefs about AI assistance can influence positive attitudes and behaviours in how

individuals utilise such services.

47
2.5 Chapter Summary

This chapter has reviewed previous research on adopting AI assistants and cross-cultural views.

It began with a brief history of chatbots, presenting its main components and discussing how it

has impacted various domains. This was followed by a discussion of empirical studies on

general AI assistant adoption, particularly e-commerce. Furthermore, the cross-cultural

perspectives on technology adoption were discussed in detail. The critical factors applied in

the AI assistant adoption were described in detail: perceived usefulness, perceived ease of use,

interactive communication, personalisation, and attitude.

An analysis of prior studies revealed a significant void in the literature, specifically, the dearth

of research and conflicting evidence concerning the intention to use AI assistants. Furthermore,

empirical research on this topic from the viewpoints of both users and non-users in the e-

commerce domain is notably scarce. There is also a limited body of research that compares the

attitudes and intentions of Western and Eastern consumers regarding the use of AI assistants.

Therefore, the anticipated outcomes of the current study aim to address and bridge this gap by

offering original and comprehensive data on the intentions of users and non-users to utilise AI

assistants among global e-commerce consumers.

48
Chapter 3. Theoretical Background and Research Model

The following section explores the theoretical foundations relevant to the adoption of

technology and the acceptance of AI assistants. Special attention is given to elucidating the

foundational aspects of the Technology Acceptance Model (TAM), which serves as the

fundamental framework for the investigative paradigm in this scholarly inquiry. Moreover, this

chapter introduces the research model employed in the current study to address the gaps

identified in the literature in Chapter 2. Additionally, hypotheses derived from the research

model are articulated in this chapter, accompanied by a presentation of the pertinent literature

for each. This chapter theoretically develops the research hypotheses through theoretical

linkages in the hypothesis development section. The research model of this study is presented

in Section 3.2. The summary of the research hypotheses is presented in Section 3.4. Finally,

Section 3.5 concludes this chapter with a summary.

3.1 Theoretical Background

Businesses increasingly apply emerging technologies to meet their customers' needs.

Therefore, its successful development is based on customers' attitudes towards using them

(Guo & Luo, 2023; Foroudi et al., 2018). AI assistants remain in their nascent stage,

characterized by limited scope and capabilities. Consequently, the exploration of AI assistants

and their utilisation has emerged as a compelling subject within scientific research, particularly

over the past decade. However, existing studies are constrained, omitting entire regions and

yielding contradictory results on certain factors (Balakrishnan & Dwivedi, 2024; Collins et al.,

2021). Some critical theoretical backgrounds that can help explain customers' attitudes and

behavioural intentions in this context are; the TAM, a widely used model in technology

adoption that explains users' acceptance of new technologies ( Singh et al., 2024; Li, 2023;

Wang et al., 2023a). The TAM model is proposed and derived from the theory of reasoned

49
action (TRA) (Fishbein, 1979). The TRA theory proved that pre-existing consumer attitudes

successfully predict and explain behavioural intentions that lead to a specific behaviour (Davis,

1989). Figure 3.1 presents the original factors of the TRA model: attitude toward behaviour,

subjective norms, and behavioural intention.

Figure 3.1. The TRA Model

Source: Ajzen & Fishbein, 1980

The primary purpose of the TAM model is to explain how internal factors, such as individuals'

attitudes and intentions, are affected by external factors. The factors of the TAM can be

observed in Figure 3.2. It suggests that users' attitudes and intentions to use technology are

determined by two primary factors: perceived usefulness and ease of use. However, Davis

(1989) has not measured a critical example of external factors and mentioned that users’ general

characteristics could be an affected external factor. Although different researchers have

developed several new models for technology acceptance, the importance of the TAM model

is still recognised by both academic and industrial researchers, especially in the AI assistant

use field (Xiong et al., 2024; Wang et al., 2023a; Sciarelli et al., 2022; Chi, 2018; Zarouali et

al., 2018).

50
 Figure 3.2. The TAM Model

Source: Naeini, 2012, p. 5288

Consequently, this section comprehensively synthesises pertinent scholarly works concerning

utilising the TAM model in the context of novel technology assimilation and using AI

assistants. This evaluative examination is the foundation for formulating a suitable conceptual

framework for the current research. Moreover, it aids in the judicious selection of factors

delineated within the literature as the most auspicious in elucidating the underlying

determinants of intent to employ AI assistants, specifically from the vantage point of online

shoppers. The ensuing enumeration expounds upon the principal rationales driving the

integration of the TAM model in this study.

1. The TAM model is specifically formulated to assess the propensity of individual users to

embrace a particular technology, in contrast to the Unified Theory of Acceptance and Use of

Technology (UTAUT), as shown in Figure 3.3, which operates at the organisational level

(Wang et al., 2023a; Ammenwerth, 2019; Lai, 2017). Hence, the present study is grounded in

examining individual customers’ perceptions, elucidating the rationale behind the preferential

suitability of the TAM model for adoption.

2. The TAM model is recommended to be used where the acceptance of novel technology

innovations is nascent (Acikgoz & Vega, 2022). This proposition is particularly pertinent within

emerging technological advancements, such as AI assistants in the e-commerce domain, which

are progressively gaining prominence for augmenting customer service and online shopping

51
experiences. The intricate dimensions of AI assistants’ functionalities elucidate consumers’

adoption behaviours concerning their utilisation in the e-commerce sector, which remains

multifaceted (Li & Wang, 2023; Adam et al., 2021). Notably, the study encompasses a cohort

of e-commerce consumers who have yet to engage with this technological paradigm,

positioning them as nonusers. Because of these circumstances, the employment of the TAM

model is advocated as an appropriate analytical framework.

Figure 3.3. The UTAUT Model

Source: Venkatesh et al., 2003

Meanwhile, understanding users' attitudes and acceptance of technology is crucial for ensuring

its successful adoption and implementation. It posits that two primary factors influence user

acceptance: Usefulness and Ease of use. These factors are crucial in shaping consumers'

attitudes about technology, ultimately influencing their opinions and behaviours regarding

technology adoption (Kelly & Palaniappan, 2023). Usefulness refers to the degree to which

consumers believe new technology will enhance their experience and provide tangible benefits

(Eleonora & Loredana, 2012). In the context of shopping assistants, customers are likely to

form positive beliefs if they perceive that AI technology can offer personalised product

recommendations, streamline the shopping process, and save time and effort (Jan et al., 2023;

Ameen et al., 2021). This perceived usefulness is influenced by various factors, such as the

assistant's ability to understand customer preferences, its accuracy in suggesting relevant

products, and its capacity to adapt to individual shopping behaviours over time (Ho & Chow,

52
2023). Similarly, perceived ease of use is another critical factor in shaping consumers' beliefs

about an emerging technology (Huda, 2023). It relates to how users perceive that interacting

with the technology is effortless and intuitive (Malodia et al., 2021). Perceived ease of use is

vital. Customers are more likely to have positive opinions if they find the technology easy to

navigate, understand, and incorporate into their routines (Jo, 2022) and shopping activities

using AI assistants (Kautish et al., 2023). Moreover, it influences users' perceptions of the

technology's usefulness, as the more manageable the system, the greater its utility can be

(Mikalef & Gupta, 2021). Factors contributing to this perception include the assistant's

interface design, conversational capabilities, and the absence of complex or confusing

processes. These factors are selected based on the understanding that users' beliefs and attitudes

significantly influence their intentions to adopt and continue using technology, such as mobile

applications (Galetsi et al., 2023), social media (Wang et al., 2023b), virtual reality (Lv et al.,

2022), and AI shopping assistants (Klaus & Zaichkowsky, 2022).

According to TAM, positive beliefs in both usefulness and ease of use lead to a positive attitude

toward the technology, which, in turn, drives the intention to use the technology (Teo et al.,

2009). Thus, by examining and understanding customers' attitudes to perceived usefulness and

ease of use, businesses can identify the determinants of their opinions and adjust their AI

shopping assistants to effectively meet customer needs and expectations. Moreover, other

theoretical perspectives complement TAM in understanding the significance of perceived

usefulness and ease of use in shaping customers' opinions toward AI shopping assistants. The

customer experience concept emphasises that a positive experience is fundamental to

influencing customer opinions and building loyalty (Srivastava & Kaul, 2016). A user-friendly

and useful AI shopping assistant can enhance the customer experience, leading to more

favourable opinions and increased usage (Ruan & Mezei, 2022; Zimmermann et al., 2023).

53
Furthermore, the Social Exchange Theory (SET) highlights the impact of social interactions

with AI shopping assistants, affecting customers' satisfaction (Jiang et al., 2022). Additionally,

the Theory of Reasoned Action emphasises subjective norms and the influence of the social

environment on customers' attitudes (Roh et al., 2023). Meanwhile, trust shapes customers'

opinions toward AI shopping assistants. Customers' attitudes are influenced by their confidence

level in the AI assistant's reliability, security, and competence. Trust is built over time through

positive experiences and consistent performance of the AI shopping assistant in meeting

customers' needs and preferences (Pitardi & Marriott, 2021). Moreover, users’ technology skills

and ethical concerns factors also shape customers' opinions toward AI shopping assistants

(Mantelero, 2022). Customers with different technological skills may have varying attitudes

toward AI and personalised shopping experiences (Pillai et al., 2020). Ethical concerns, such

as data privacy and transparency in AI decision-making processes, can influence customers'

opinions positively or negatively, depending on how well companies address these issues

(Aldboush & Ferdous, 2023).

Similarly, interactive communication and personalisation also affect customer attitudes and

lead to the establishment of good behavioural intentions (Kim & Gambino, 2016). The

theoretical background for interactive communication and personalisation is crucial in shaping

customer attitudes and fostering positive behavioural intentions toward a technology (Chandra

et al., 2022; Mariani et al., 2023). In the context of AI assistants, through interactive

communication, AI assistants engage in dynamic and responsive interactions with users as two-

way communication (Guzman & Lewis, 2020), fostering a sense of connection and perceived

social presence, leading to heightened trust and likability (Tsai et al., 2021). In contrast, as

contradictory results, Chung et al. 2020 have found that communication competence does not

affect customer satisfaction. Personalisation of service reflects an individual’s interests and

preferences so that with a greater depth of personalisation, the content will be more relevant

54
and useful (Serrano-Malebran & Arenas-Gaitan, 2021). Despite users’ concerns over personal

information leakage in the technology environment, individuals’ needs for personalised

services are increasing, creating a conflicting phenomenon (Lee, 2019). Drawing from theories

like the TAM, SET, and personalisation, the positive user experiences facilitated by interactive

communication (two-way communication) and personalisation lead to increased perceptions

of AI assistants as valuable and reliable tools, resulting in the establishment of favourable

behavioural intentions, such as continued usage and word-of-mouth recommendations, which

ultimately contribute to the successful integration of AI assistants into users' lives (Gao &

Huang, 2019; Khalid & Ali, 2017).

The factors underlying the selection of behavioural intentions concerning the adoption of AI

within the context of e-commerce are grounded in pertinent theoretical constructs. The

rationale for including these factors in the study is substantiated by their alignment with

established theoretical frameworks and empirical evidence from existing literature. Drawing

from the TAM, the chosen factors encapsulate critical determinants influencing users'

inclinations to engage with AI assistants in e-commerce settings. These factors encompass

perceived usefulness, perceived ease of use, social influence, and facilitating conditions. These

dimensions have been widely acknowledged and explored in prior research as influential

predictors of users' intentions to adopt and use technological innovations.

The decision to incorporate these factors stems from their comprehensive coverage of the

cognitive, social, and contextual aspects that underpin users' decision-making processes.

Empirical support from seminal studies further reinforces their significance in elucidating the

complexities of technology adoption, including AI assistants, in e-commerce. Through a

synthesis of scholarly references and theoretical underpinnings, this study substantiates the

rationale behind selecting these factors by elucidating the theoretical basis and explaining the

linkage between the chosen factors and the phenomenon under investigation. However,

55
understanding these theoretical elements is critical for organisations and developers seeking to

increase customer adoption and usage of AI shopping assistants, which can enhance the overall

e-commerce experience. Additionally, it is important to note that customers' opinions are

diverse and multifaceted, so combining these theories with personalisation is necessary for a

comprehensive analysis. Social, consumer, and contextual factors will also significantly shape

customers' attitudes toward AI shopping assistants.

3.2 Research Model

Figure 3.4 shows the research model designed for this study. The proposed model examines

consumers' attitudes and intentions towards using Artificial Intelligence (AI) service shopping

assistants in e-commerce as the primary dependent variables. This study validates the

relationship between attitude and usage intention in commerce AI assistants. Further, this study

has explored and added additional constructs, which are perceived usefulness, perceived ease

of use, and AI shopping assistant capabilities (interactive communication and personalisation),

to make the research model more meaningful in understanding consumers' acceptance and

usage of AI assistants in e-commerce. Table 3.1 summarises the definitions of the critical

factors used in the research model.

Figure 3.4. Research Model

56
The model was based on the hypothesis that consumers' attitudes towards AI assistants are the

fundamental determinant driving their decision to adopt these technologies, consequently

impacting their tangible utilisation of the AI assistant. The perceived usefulness and ease of

use encapsulate an individual’s belief in the potential of AI assistants to augment task

performance or productivity/efficiency, accompanied by the notion that interaction with AI

assistants will be devoid of undue complications. Moreover, within this theoretical framework,

the attitude held by an individual towards the utilisation of AI assistants encapsulates their

favourable or emotional disposition. Concurrently, behavioural intention pertains to an

individual's ultimate decision to adopt AI assistants, contingent upon their orientation. Prior

research has found that belief, attitude, and purpose are strongly related and have significant

relationships (Foroughi et al., 2019). However, previous studies have argued that perceived

ease of use and perceived usefulness may only partially reflect consumers' attitudes,

necessitating an investigation of additional predicting factors (Ladkoom & Thanasopon, 2020).

Therefore, this research tested a model that assesses other elements from the literature, such as

AI shopping assistant capabilities, which could influence consumers' attitudes toward AI

shopping assistants.

The research model included seven hypotheses concerning the relationships among the model’s

distinct components. The initial hypothesis posited a favourable association between usefulness

and attitudes toward adopting AI assistants among users and non-users. The second hypothesis

postulated a positive relationship between the attributes of AI assistants and the corresponding

attitudes towards their ease of use among users and non-users. The third hypothesis inferred

that perceived ease of use positively influences perceived usefulness among users and non-

users of AI assistants. The fourth hypothesis perceives interactive communication positively

influences the attitudes of users and non-users towards the use of AI assistants. The fifth

57
hypothesis conjectured that personalisation positively influences the attitudes of users and non-

users in utilising AI assistants in e-commerce. The sixth hypothesis inferred that attitude

positively impacts behavioural intention to use AI assistants among users and non-users. The

seventh hypothesis postulated that the relationships among the primary factors influencing the

intention to use AI assistants in e-commerce differ between Western and Eastern cultures. The

research model in this study is used to explain the adoption of using AI assistants. Therefore,

the research model is significant and distinguished compared to previous studies for several

reasons:

1. Focus on Consumer Perceptions: The research model focuses on consumer perception

and usage intentions. This focus is essential because it acknowledges that accepting AI

platforms and their success depends on consumers' attitudes toward their adoption.

2. Simplicity: The research model is simple and easy to understand. It consists of only

four key constructs: perceived usefulness, perceived ease of use, and AI capabilities

(interactive communication and personalisation). This simplicity makes the research

model accessible to researchers and practitioners.

3. AI applications: The research model specifically addresses the trended AI assistant

technologies across global consumers, including users and non-users of AI assistants.

This trended AI technology has helped establish the research model as an essential

framework for understanding AI assistant adoption. Indeed, previous studies have

investigated the effectiveness of chatbots among Facebook users (Zarouali et al., 2018).

Still, they have not included AI capabilities, a critical factor before consumers’ attitudes.

4. User Differences: The research model considers the consumers’ differences toward

using AI assistants in e-commerce from multiple perspectives, unlike recent studies

focusing on just users and a specific cultural context (Lee & Lin, 2023; Shin, 2010).

58
Table 3.1. Definition of Factors Included in the Research Model

Constructs Definition Sources

Perceived Perceived usefulness and ease of using AI shopping assistants are Zarouali et al.,
usefulness and
practical and effortless and will enhance consumers' productivity or 2018
ease of use
task performance.

AI capabilities: The extent to which AI shopping assistants facilitate two-way Chung et al.
Interactive
communication between consumers to provide personalised assistance 2020,
communication
and and support tailored to their specific needs and preferences. Srinivasan et
personalisation
al. 2002

Attitude towards Refer to the feeling or opinion towards using AI shopping assistants. Moon and
using Kim, 2001

Usage Intentions Refer to the willingness to use and continue using AI shopping assistants Zarouali et al.,
in the future and recommend them to others. 2018

One major theoretical factor that drives customers' adoption and use of AI shopping assistants

is their sense of usefulness and usability (Ajisoko, 2020; Maryanto & Kaihatu, 2021).

Consumer utility beliefs are their assessment of how valuable and beneficial AI shopping

assistants are in terms of increasing productivity and task performance. This opinion is based

on the idea that AI assistants can provide helpful features such as tailored suggestions, efficient

search functionality, and overall shopping ease (Ajisoko, 2020). Consumers are more likely to

accept and use AI shopping assistants in their e-commerce activities if they are helpful.

Similarly, ease-of-use perception heavily influences consumer adoption and use of AI shopping

assistants. Ease of use refers to users' perceptions of how simple it is to interact with AI helpers

and traverse their interfaces. Perceived ease of use impression is heavily influenced by intuitive

design, clear information display, and simplicity of communication (Sudaryanto et al., 2023).

Customers who view AI shopping assistants as user-friendly are more likely to have good

attitudes regarding utilising them and are more likely to incorporate them into their shopping

habits.

59
Another key theoretical issue is how AI shopping assistants promote two-way dialogue and

customised consumer guidance (Huang & Rust, 2021). Interactive communication is the ability

of AI assistants to participate in dynamic and responsive interactions with customers (Adam et

al., 2021). This includes answering inquiries, clearing uncertainties, and offering real-time

assistance. Interactive communication increases consumer happiness and perceptions of the AI

assistant's effectiveness by fostering a sense of involvement and trust. Personalisation is

another essential feature of AI shopping assistants. Personalised help means adapting advice,

ideas, and product offers to meet consumers' requirements, preferences, and interactions

(Suresh et al., 2023). AI shopping assistants increase consumer happiness and establish

relevance by offering individualised experiences, which can favourably affect attitudes and

intentions toward adopting the technology (Zimmermann et al., 2023). These factors are

essential in shaping users' emotional connection and perceived social presence with the AI

assistant (Zhang et al., 2021). By examining how interactive communication fosters dynamic

and engaging interactions while personalisation tailors the experience to individual

preferences, the researcher could explore how these aspects influence users' attitudes and

subsequent behavioural intentions. Moreover, AI shopping assistants can be programmed to

interact with consumers conversationally, using natural language processing to understand and

respond to consumers' requests (Bahja, 2020). Personalised communication can enhance the

consumer experience of using AI assistant platforms. Attitude towards using refers to

consumers' overall perspective about using the AI shopping assistant. Understanding the

predisposition and emotions users associate with AI assistants is pivotal in comprehending how

receptive they are to integrating these tools into their daily lives.

Attitude is an individual's general feeling or opinion about using AI shopping assistants. It

includes customers' evaluative judgments, emotional reactions, and attitudes about the worth

and use of AI assistants (Pillai et al., 2020). Positive sentiments regarding AI shopping

60
assistants are expected to be influenced by perceptions of usefulness, usability, interactive

communication, and personalisation (Yin & Qiu, 2021). Attitudes affect consumers' readiness

to accept and employ AI shopping assistants in e-commerce, impacting their intent to use them.

Positive attitudes often lead to greater acceptance and willingness to continue using AI

assistants, while negative attitudes may hinder adoption and create barriers (Acikgoz et al.,

2023). Finally, the factor of behavioural intention reflects users' inclination to use AI assistants

in the long run. Behavioural intention is the likelihood that consumers will use AI shopping

assistants in the future and recommend them to others. It denotes expected conduct based on

people's views, perceived usefulness, and subjective standards. Consumers with favourable

beliefs and stronger perceptions of service, usability, interactive communication, and

personalisation are more likely to acquire strong intentions to continue using AI shopping

assistants and to urge others to do the same.

Analysing the attitudes of users and non-users towards AI, as well as comparing these

perspectives between Western and Eastern consumers, provides valuable insights into the

cultural and experiential factors that shape individual perceptions of technology. Users often

exhibit a higher level of comfort and familiarity with AI due to direct interactions, which can

result in a more nuanced understanding of its benefits and limitations (Jiang et al., 2022).

Conversely, non-users may hold reservations rooted in uncertainty or misinformation (Kreutzer

& Sirrenberg, 2020). When considering Western and Eastern consumers, different societies,

such as the West and East, often contrast their attitudes toward Technology and AI (Jecker &

Nakazawa, 2022; Lim et al., 2021). Western societies, typically characterised by fast-paced

technological adoption, may showcase a more diverse range of attitudes due to varying levels

of innovation and education. Meanwhile, Eastern cultures might have unique societal

expectations that could demonstrate attitudes influenced by communal values or past

technological experiences (Frank et al., 2021). Therefore, this mode tested a difference between

61
Western and Eastern consumers regarding their attitudes and intentions to use this service for

e-commerce.

Incorporating these factors into this study offered a holistic and robust analysis of the complex

dynamics between users and AI assistants. Examining how usefulness, ease of use, AI assistant

capabilities, and attitudes collectively influence behavioural intentions leads to identifying key

drivers and barriers to successful AI assistant adoption. It offered valuable insights into the

mechanisms that lead to positive outcomes, such as increased satisfaction, trust, and continued

usage, which are vital for successfully establishing and integrating AI assistants in e-commerce.

Through systematically exploring hypotheses, investigators can gain enhanced insights into the

pivotal determinants that influence the adoption of AI assistants. This process subsequently

empowers the formulation of strategies to enhance the embracement and integration of

sophisticated AI assistants within online retailers and e-commerce enterprises. In light of these

findings, marketing practitioners are impelled to cultivate and refine their customer service

stratagems, as these measures effectively facilitate consumers' interactions with e-service

platforms, consequently mitigating the temporal extent devoted to online product exploration

and shopping undertakings.

3.3 Hypotheses

The adaptable nature of IS adoption research models facilitates their application across various

technological domains, allowing for the incorporation of context-specific determinants. As a

result, examining the adoption of AI assistants in e-commerce involves considering several

factors that have proven to be impacted from e-commerce consumers’ perspectives, affecting

both users’ and non-users’ attitudes. These key factors were discussed in Chapter 2, and their

hypotheses were elaborated on in the following sections.

62
3.3.1 Perceived Usefulness and Attitude

The concept of perceived usefulness in the context of AI deployment within e-commerce holds

significant implications for individuals' attitudes toward its utilisation. The perceived

usefulness of AI technologies, encompassing their potential to enhance decision-making,

optimise shopping experiences, and provide personalised recommendations, fundamentally

shapes consumers' attitudes and receptiveness towards incorporating such technologies into

their online shopping practices. A positive perception of AI's usefulness can foster a favourable

attitude, engendering a willingness to embrace and engage with AI-driven features in e-

commerce platforms and reinforcing the symbiotic relationship between perceived usefulness

and user attitudes in this dynamic digital landscape.

In addition, recent studies have shown that perceived usefulness positively influences

technology adoption and acceptance (Ohk et al., 2015). Additionally, past empirical evidence

has demonstrated that usefulness perception can influence consumer attitudes, such as attitudes

toward social media adoption (Alduaij, 2019). Moreover, the study's findings demonstrate that

each proposed hypothesis has yielded favourable and statistically significant outcomes. In

contrast, the variable perceived usefulness emerged as the most robust predictor influencing

the individual's overall experiential outcome for using chatbots (Lubbe & Ngoma, 2021). In

addition, another study's results also reveal that factors such as the perceived usefulness of the

mobile app have noteworthy effects on the user's intentions to employ the mobile application

and their overall attitude towards it (Hasan, 2022). Nevertheless, trust, personal innovativeness,

and attitude solely impacted the inclination toward usefulness.

Meanwhile, including "usefulness" in studies about using AI in e-commerce is motivated by

its significance in shaping individual attitudes. This factor is recognised as pivotal due to its

potential to influence how individuals perceive the applicability and advantages of AI within

the e-commerce context. Investigating the relationship between usefulness and attitude among

63
users and non-users helps understand how the perceived usefulness of AI technologies may

contribute to shaping overall positive or negative attitudes toward their adoption and

integration in e-commerce settings (Teodorescu et al., 2023). This relationship is expected to

provide valuable insights into the underlying mechanisms that drive individuals' behavioural

intentions and decision-making processes in AI use in e-commerce (Rodgers et al., 2023). In

summary, perceived usefulness has been approved as a strong predictor of how people feel

about adopting AI assistants for various tasks, including online shopping. Nevertheless, there

remains a lack of evidence regarding its impact on the intention to use AI assistants among

both non-users and users. To test this, the following hypotheses were developed:

H1. Perceived usefulness positively influences the attitudes of users and non-users towards

using AI assistants in e-commerce.

3.3.2. Perceived Ease of Use and Attitude

The perception of the ease with which AI technologies can be employed within the context of

e-commerce significantly shapes individuals' attitudes toward their utilisation. This cognitive

assessment of the convenience and ease of use of AI-driven tools and systems, ranging from

personalised product recommendations to virtual shopping assistants, profoundly influences

stakeholders' inclinations to embrace and engage with such technological solutions (Ho, 2021).

Consequently, enhanced perceived ease of use tends to foster a more positive attitude among

consumers and potential users, fostering an environment conducive to the widespread

acceptance and integration of AI advancements in e-commerce (Kumar et al., 2019).

Furthermore, the perceived ease of use directly impacts the intention of customers to make a

purchase (Nofirda & Ikram, 2023). Additionally, trust positively impacts perceived ease of use,

correlating with perceived usefulness and user attitudes towards AI (Wang et al., 2023a). In

contrast, the empirical evidence suggests that perceived ease of use does not correlate

statistically with the dependent variable (Tan & Lim, 2023).

64
The significance of the "ease of use" factor in utilising AI stems from its pivotal role in

determining the successful adoption and integration of AI technologies by individuals and

organisations. This factor directly influences the accessibility, comprehensibility, and user-

friendliness of AI systems, affecting users' willingness and efficiency to engage with and

harness AI capabilities. A high degree of ease of use in AI systems is instrumental in mitigating

potential barriers that impede the acceptance and effective utilisation of these advanced

technologies. Complex and unintuitive AI interfaces or applications can lead to user frustration,

resistance, and decreased adoption rates. On the contrary, when AI solutions are designed with

a focus on ease of use, they empower a more comprehensive range of users, including those

with limited technical expertise, to seamlessly interact with AI-driven functionalities.

Moreover, the ease-of-use factor enhances productivity and task performance. Intuitive AI

interfaces enable users to quickly grasp the functionalities and features of AI systems, thereby

expediting the learning curve and facilitating proficient utilisation. This, in turn, bolsters users'

confidence and motivation to leverage AI tools to their full potential, leading to improved

decision-making processes and outcomes.

From an organisational standpoint, the ease of use of AI technologies can drive higher levels

of adoption across various departments and personnel, fostering a culture of innovation and

efficiency. This can result in quicker integration of AI-driven insights into business strategies

and operations, ultimately yielding competitive advantages and improved operational

outcomes. To succinctly encapsulate, perceived ease of use emerges as a robust precursor that

delineates the disposition towards embracing AI assistants across multiple domains, including

online shopping and its associated technologies. Nevertheless, a discernible dearth of

substantiated indications pertains to its sway upon the proclivity of both non-users and users to

engage with AI assistants. Accordingly, the following hypothesis was tested:

65
H2: Perceived ease of use positively influences the attitudes of users and non-users towards

using AI assistants in e-commerce.

3.3.3. Perceived Ease of Use and Perceived Usefulness

The assessment of AI assistants' efficiency in e-commerce is intrinsically tied to two pivotal

dimensions: perceived ease of use and usefulness (Wang et al., 2023a). The perceived ease of

use encapsulates users' perceptions of the intuitive nature and simplicity of interacting with AI-

driven systems, shaping their attitudes and predisposition towards employing such technology.

Simultaneously, perceived usefulness encompasses how AI assistants are perceived as capable

tools that enhance efficiency, convenience, and effectiveness within e-commerce contexts.

These dimensions collectively shape individuals' inclinations and behaviours regarding

adopting and accepting AI assistants in e-commerce settings. Meanwhile, Jannach et al. (2021)

found that trust in AI is materially shaped by variables encompassing transparency,

acquaintance with diverse AI applications, and the perceived usefulness and ease of use of AI

recommendation systems. In order to foster extensive approval and integration among users,

e-commerce enterprises must accord paramount importance to cultivating trust in these

emergent technological paradigms (Teodorescu et al., 2023). Moreover, Noreen et al. (2023)

demonstrated a noteworthy and affirmative correlation between the intention to embrace AI

and critical factors encompassing perceived usefulness and ease of use of AI technology.

The intricate interplay between the perception of ease of use and perceived usefulness in

employing AI within e-commerce warrants an examination of its mechanisms and underlying

rationales. How perceived ease of use influences perceived usefulness in using AI in e-

commerce can be elucidated through cognitive and psychological processes and has a

favourable impact on the perception of usefulness (Mollick et al., 2023). When users perceive

an AI-driven system or platform as easy to navigate and interact with, they are more likely to

develop a positive cognitive disposition towards the technology. This positive cognitive stance

66
can enhance their perception of the system's usefulness. Simplified and intuitive interactions

engender a sense of proficiency and mastery, subsequently fostering a perception that AI

technology is efficacious in achieving the intended goals within the e-commerce domain.

Furthermore, a perceived easy-to-use diminishes cognitive load and reduces the mental effort

required for engagement. Reducing cognitive burden liberates cognitive resources, enabling

users to process information and evaluate the system's utility more effectively (Koć-Januchta

et al., 2022). Consequently, perceived ease of use enhances the cognitive efficiency of users,

resulting in an amplified perception of the AI's overall usefulness.

The rationale for this influence can be attributed to the principle of cognitive alignment, where

congruence between the cognitive demands posed by the technology and the cognitive

capability of the user fosters a harmonious interaction. The lessened cognitive friction arising

from a user's perception of ease aligns with the cognitive processes required for appraising

usefulness. Users tend to evaluate the utility of AI within e-commerce based on their cognitive

investment, wherein reduced cognitive load stemming from the ease of use aligns with the

positive valuation of the technology's usefulness. In summation, the constructs of perceived

usefulness and ease of use have demonstrated robust predictive capabilities concerning the

disposition towards adopting AI assistants across diverse domains, including pertinent

technologies within online commerce. Nevertheless, an empirical void remains concerning

their potential impact on the inclination to utilise AI assistants, both for individuals

unacquainted with their usage and those already versed in their application. In order to probe

into this phenomenon, the subsequent set of hypotheses has been formulated:

H3: Perceived ease of use positively influences perceived usefulness among users and non-

users of AI assistants.

67
3.3.4. Interactive Communication and Attitude

Research has shown that AI assistant characteristics and interaction styles are crucial in human–

chatbot communication (Wilkinson et al., 2021; Roy & Naidoo, 2021; Liew et al., 2021). For

example, the social characteristics of AI assistants can meet users' expectations, which can help

to avoid frustration and dissatisfaction (Chaves & Gerosa, 2021). Li and Wang (2023) found

that informal language styles of AI assistants can help create a more natural and interpersonal

communication experience with consumers, ultimately leading to positive service outcomes.

While some studies on digital service assistants have found that the complete communication

skills of these tools do not impact customer satisfaction (Chung et al., 2020), the literature has

consistently indicated that AI assistants should be designed with expertise capabilities in mind

for e-commerce (Petersson et al., 2023; Liew et al., 2021). Therefore, interactive

communication is adopted from the marketing and communication literature to demonstrate

the expert level of AI assistants in e-commerce. Previous empirical studies have found that

interactive communication predicts attitudes toward online communicators (Lee et al., 2015).

Meanwhile, Interactive communication notably influences attitudes towards AI utilisation in e-

commerce due to its capacity to personalise interactions, enhance transparency and

understanding, empower users, and foster social presence. Understanding these dynamics is

essential for e-commerce platforms seeking to optimise user experiences and adopt AI-driven

features (Haleem et al., 2022; Nicolescu & Tudorache, 2022). Therefore, the limited number

of studies on the use of AI assistants and related fields, along with their varied outcomes, raises

an open question regarding the impact of the communication skills of AI assistants on the

attitudes and usage intentions of both users and non-users in e-commerce. Accordingly, the

following hypotheses were tested:

H4: Interactive communication positively influences the attitudes of users and non-users

towards using AI assistants in e-commerce.

68
3.3.5. Personalisation and Attitude

Personalisation in e-commerce applications gives the power to modify the service's

characteristics to suit their needs or preferences (Srinivasan et al., 2002). Meanwhile, AI

facilitates personalisation in various facets, including personalised profiling, navigation,

nudges, and retention strategies, which are discernible across the customer journey's sequential

phases. These personalised AI-driven interventions come to the fore in reaction to the

challenges encountered throughout this journey (Gao & Liu, 2022). The application of

personalisation exerts a favourable impact on consumer attitudes and their intentions to share

personal information with a digital assistant (Kronemann et al., 2023). Meanwhile,

personalisation significantly impacts shaping individuals' attitudes towards AI deployment in

the e-commerce landscape. This influence is underpinned by relevance, efficiency, and

psychological perceptions of companionship. Understanding these mechanisms provides

valuable insights for e-commerce businesses aiming to enhance user acceptance and

engagement with AI-driven systems.

Current e-commerce AI platform developments are focused on offering customisation options

to meet individual preferences (Wu et al., 2020). For example, Gucci uses a chatbot to send

personalised advertisements to target customers interested in customised products (Chung et

al., 2020). Much consumer research has found that customisation positively impacts

consumers' attitudes toward marketing communication and mobile advertising (Jeong et al.,

2020). Furthermore, recent findings showed that customisation could significantly impact

attitudes toward using mobile commerce applications (Marinkovic & Kalinic, 2017).

Moreover, the empirical investigation centred on a cohort of female fashion retail consumers

in the United Kingdom subjected to such promotional propositions. The findings from this case

study delineate that these consumers exhibit a proclivity for seeking price reductions on items

of interest, alongside an aspiration for enhancing their in-store experiences. However, they

69
exhibit aversion towards disruptions and generic promotional overtures. Additionally, a

conspicuous inclination towards self-governance becomes apparent, coupled with endeavours

to exert control over private information's divulgence and ameliorate the quality of

recommendations received. Importantly, our analysis brings inherent incongruities within

customers' anticipations about personalised interactions, necessitating the tracking of

geographic coordinates (Canhoto et al., 2023). Moreover, investigations indicate that

consumers swiftly adopt AI for their daily activities because it can afford individuals prompt

and valuable information access. Furthermore, AI technologies have demonstrated their

capability to cater to customers' personalisation by providing contextually pertinent and

meticulously tailored content, all while being disseminated instantaneously (Brill et al., 2022).

A limited body of research about the utilisation of AI assistants and its associated domains,

coupled with inconclusive findings, has given rise to an unresolved inquiry concerning the

impact of personalised interactions demonstrated by AI assistants on the attitudes and intention

to use, both among users and non-users of AI assistants within the area of e-commerce.

Consequently, the present study examined the following hypotheses:

H5: Personalisation positively influences the attitudes of users and non-users towards using AI

assistants in e-commerce.

3.3.6 Attitude and Intention

Consumers’ attitudinal experience using an AI shopping assistant positively relates to their

behavioural intention, especially their usage intention. The relationship between attitude and

use intention is well-documented in IS research (Harasis et al., 2018). More precisely, this

study focuses on 'consumers' choice to use AI assistants to communicate with online retailers

for e-commerce. Recent studies have found that attitude is a crucial factor influencing

consumers’ intention to use virtual AI as a shopping assistant (Kasilingam, 2020). The theory

70
of reasoned action and technology acceptance model states that the user's behavioural intention

to use an information system determines user acceptance and actual behaviour (Davis, 1989).

Behaviour refers to a consumer's intention to take specific actions (Purwanto & Loisa, 2020).

Therefore, the intention will determine use behaviour (Davis, 1989). Purwanto and Loisa

(2020) found that mobile banking systems positively and significantly predicted the actual

banking system use behaviour. Moreover, Tandi and Questier (2014) also found that the

behavioural purpose of using communication systems predicted actual usage. Thus, this

research hypothesis is that the attitudes will positively influence the usage intentions of users

and non-users toward using AI assistants in e-commerce. (H6).

The discernible distinction in consumers' attitudes and behavioural intentions between Western

and Eastern cultural contexts regarding adopting and utilising technologies can be attributed to

cultural, societal, and psychological factors. Western cultures, characterised by individualism

(Kim et al., 1994) and valuing autonomy, often foster an environment that encourages rapid

technology adoption, as it aligns with notions of self-expression and personal efficiency

(Leidner & Kayworth, 2006). In contrast, Eastern cultures, which frequently emphasise

collectivism and social harmony (Triandis, 2015), tend to exhibit a more cautious approach

towards technology integration, valuing stability and tradition (Erez & Earley, 1993).

Furthermore, cultural dimensions such as uncertainty avoidance and power distance influence

the level of comfort individuals feel when engaging with novel technologies. Psychological

theories, including the TAM, suggest that cultural norms and values shape cognitive and

affective responses to technology ( Lai et al., 2016; Srite & Karahanna, 2006). Additionally,

variations in language, communication styles, and aesthetics shape perceptions of technology.

These multifaceted factors collectively contribute to the distinct patterns of consumers'

attitudes and intentions in Western and Eastern cultural contexts when adopting and utilising

technologies. Accordingly, the model was tested between Western and Eastern groups to

71
explore the differences in the relationship between all antecedents and intention towards the

use of AI assistants. Therefore, this research hypothesises that the relationships among the

primary factors influencing the intention to use AI assistants in e-commerce differ significantly

between Western and Eastern cultures (H7). Table 3.2 outlines the hypotheses.

Table 3.2. Summary of Hypotheses

Sub-questions 1.1 – 1.7 Hypotheses

Main Question
How does perceived usefulness affect the H1. Perceived usefulness positively
attitudes of users and non-users towards influences the attitudes of users and
What factors influence individuals' intention to use AI assistants in e-commerce among users and
non-users, and are there significant differences in these factors’ impact when comparing Western

using AI assistants in e-commerce? non-users towards using AI assistants in

e-commerce.

How does perceived ease of use affect the H2. Perceived ease of use positively
attitudes of users and non-users towards influences the attitudes of users and
using AI assistants in e-commerce? non-users towards using AI assistants in
e-commerce.

How does perceived ease of use affect the H3. Perceived ease of use positively
perceived usefulness of users and non-users influences perceived usefulness among
of AI assistants in e-commerce? users and non-users of AI assistants.

How does interactive communication affect H4. Interactive communication

and Eastern cultures?

the attitudes of users and non-users towards positively influences the attitudes of
using AI assistants in e-commerce? users and non-users towards using AI
assistants in e-commerce.

How does personalisation affect the H5. Personalisation positively

attitudes of users and non-users towards influences the attitudes of users and
using AI assistants in e-commerce? non-users towards using AI assistants in
e-commerce.

How does attitude affect the intentions of H6. The attitudes will positively
users and non-users towards using AI influence the usage intentions of users
assistants in e-commerce? and non-users toward using AI
assistants in e-commerce.

Do significant differences exist in the H7. The relationships among the

relations among the primary factors primary factors influencing the
influencing the intention to utilise AI intention to use AI assistants in e-
assistants in e-commerce when comparing commerce differ significantly between
Western and Eastern cultures? Western and Eastern cultures.

3.4 Chapter Summary

This chapter has described the theoretical models pertinent to technology adoption and the

acceptance of AI assistants. It also described factors used in previous associated studies to

establish the theoretical linkages of the research model. As mentioned above, the research

72
model was based on the TAM Model, and the researcher intended to extend it by incorporating

AI assistant capabilities while considering user and cultural differences that have been found

to be impactful for AI assistant adoption in e-commerce. Furthermore, seven research

hypotheses have been developed for empirical testing based on the theoretical arguments

concerning the main research factors, such as usefulness, ease of use, interactive

communication, personalisation, and usage intentions. The following chapter discusses the

research methodology adopted for this study.

73
Chapter 4. Research Methodology

4.1 Research Paradigm

Years of scientific inquiries have led to the development of various considerations for

researchers to account for in establishing how they should engage in a specific investigation.

These approaches have been grouped into research paradigms that have emerged and developed

across different disciplines. According to Mackenzie and Knipe (2006), the research paradigm

alludes to the framework that integrates the theories and practices aligned with the researcher’s

discipline to develop an appropriate research plan. Antwi and Hamza (2015) suggest that the

research paradigm alludes to the set of ideas, beliefs, or understanding that sets the basis on

which theories and practices function in creating or improving knowledge. The research

paradigm is premised on the aim of the study, research objectives, questions, instruments and

measurements, and analysis methods. According to Mackenzie and Knipe (2006), adopting a

research paradigm is critical for any study because it clarifies the investigation and improves

the quality of the methods and analysis employed in developing knowledge.

Furthermore, Antwi and Hamza (2015) highlight that the research paradigms reflect the beliefs,

assumptions, and biases that impact the research process, outcomes, and application of the

findings. There are three key pillars of research paradigms. These are ontology, epistemology,

and methodology. Ontology alludes to the nature of reality, which can be single or multiple

realities. Epistemology focuses on the study of knowledge by highlighting how researchers can

gain knowledge and how they can validate it. The methodology is how the research is

conducted and validated to gain new knowledge. Therefore, the research paradigms offer

insights into how ontology, epistemology, and methodology are integrated to develop a

practical structure and process for conducting the investigation. According to Goduka (2012),

74
the research paradigm establishes the philosophical basis that entails how data is collected and

analysed to develop knowledge.

Although various research philosophies exist, the majority of the paradigms are premised on

the positivism and interpretivism approaches. Alharahsheh and Pius (2020) note that positivism

and interpretivism are key influences in developing other research paradigms due to their direct

link to quantitative and qualitative research approaches. The variations between the two

approaches highlight their unique strengths in addressing issues under investigation. According

to Antwi and Hamza (2015), positivism and interpretivism present contrasting philosophies

that guide social science investigations through different ontological and epistemological

approaches. Positivism is described as a philosophical approach rooted in natural sciences that

advocates for gathering and analysing empirical evidence using scientific methods. The

underlying assumption in positivism is that investigating the social world should be

accomplished by focusing on the regular patterns and laws in a phenomenon similar to what is

experienced in the natural world using observations and measurements (Cecez-Kecmanovic,

2005). Positivism is objective, implying that researchers aim to eliminate bias and personal

values in the research process. This is accomplished by applying quantitative methods to gather

large volumes of data from large sample sizes to enhance the generalisation of the findings and

conclusions. It also embraces deductive reasoning, which entails starting with a theory or

hypothesis that can be tested using the data collected and analysed in the study (Casula et al.,

2021). Interpretivism contrasts positivism as a philosophical approach that focuses on

subjective meanings, assuming that reality is a social construct that can be understood from

multiple perspectives. According to De Villiers (2005), interpretivism challenges the

perceptions held by positivists by highlighting the importance of integrating the intricacies of

human experiences and behaviours from the multiple layers emerging in the social context

under investigation. Interpretivists demonstrate that researchers play a critical role in gathering

75
and interpreting data from qualitative methods that foster a subjective approach for deeper

exploration of meanings and contexts. Therefore, adopting interpretivism offers an opportunity

to integrate human elements such as beliefs, actions, relationships, and interactions in

developing new insights.

The current study seeks to demonstrate how AI assistants in e-commerce can be improved

premised on the users’ experiences with the technologies. Therefore, gathering quantitative and

qualitative data is critical in addressing the research objectives. According to Kaushik and

Walsh (2019), a pragmatic research paradigm is a philosophical approach that bridges the gap

between positivism and interpretivism and offers an opportunity for researchers to integrate

qualitative and quantitative research methods in the same study. Pragmatism is a problem-

oriented approach that uses different research methods based on their appropriateness in

addressing the research problem (Morgan, 2014). Although the pragmatic research philosophy

could have been adopted in the current study, employing both positivist and interpretivist

research philosophies offers a more strategic approach to addressing the research problem and

influencing outcomes that substantially contribute to the wider social science field. In this

context, addressing the research problem requires a sequential approach where quantitative

data is employed in developing the basis for developing the qualitative investigation that, when

the insights are integrated, can result in practical solutions for the research problem. The current

study is pertinent for social sciences by demonstrating different research philosophies can be

used in the same study. Moon and Blackman (2014) review that researchers in social sciences

often focus on a single philosophical approach. This study challenges this by highlighting that

a sequential approach can entail following different philosophical approaches to enhance the

insights gained from each research methodology adopted.

Quantitative research refers to structured and systematic techniques used to gather and analyse

numerical data that can be interpreted to explain a particular phenomenon. Quantitative studies

76
entail measuring variables and establishing the relationships between them, leading to

generalisations based on the statistical analysis conducted. Often, data collection entails

administering surveys to large samples or conducting experiments that involve manipulating

variables to gain insights into the cause-and-effect relationships. The statistical analysis

includes descriptive statistics such as mean, median, and standard deviation and inferential

statistics (Simpson, 2015). Qualitative research methods premised on interpretivism entail

gathering non-numerical data that fosters an in-depth understanding of human experiences and

social phenomena. Qualitative studies are effective in developing meanings and interpretations

and in contextualising social experiences that are difficult to quantify. Qualitative data can be

collected from secondary sources through literature review or document analysis and primary

research involving interviews, focus groups, and observations in natural settings, among others

(Kaplan & Maxwell, 2005). The quantitative and qualitative methods can be used in the same

study through the mixed-method research methodology. This study adopts a mixed-method

approach for its research methodology, integrating both quantitative and qualitative methods

to encompass positivism and interpretivism perspectives in faithfully representing the

phenomenon under investigation (Petter & Gallivan, 2004). Quantitative and qualitative

methods can effectively complement each other within a mixed-method research design, a

framework that aligns with critical realism. Critical realism emphasises understanding both

observable regularities and the underlying causal mechanisms that shape phenomena

(Mukumbang, 2023). This study adopts a mixed-method approach to explore the adoption of

AI assistants in e-commerce, integrating quantitative surveys and qualitative interviews. The

quantitative aspect follows a positivist approach, utilizing established theories to formulate

hypotheses and employing surveys to quantify consumers' attitudes and intentions towards AI

assistants. Concurrently, the qualitative component adopts an interpretivist stance, providing

deeper insights into the survey results and exploring the contextual factors influencing adoption

77
behaviours. This methodological pluralism not only enhances the breadth of understanding but

also aligns with critical realism's goal of uncovering both the surface-level patterns and the

deeper structures that influence technology adoption in real-world settings.

The primary emphasis is on a quantitative (positivist) approach, which involves studying well-

established theories and pertinent literature to formulate hypotheses and a conceptual model.

The conceptual model is subjected to quantitative analyses, utilising a survey method to gauge

consumers' attitudes and intentions towards using AI assistants. Subsequently, a qualitative

(interpretivist) approach is employed to provide supplementary insights that enhance the

understanding of the quantitative results. Machine learning techniques were employed as a

supplementary procedure for this type of analysis. Therefore, the research is accomplished

through a sequential mixed methodology.

The mixed method research strategy using qualitative and quantitative techniques is widely

used in information systems (IS) fields (Petter & Gallivan, 2004). According to Ivankova et al.

(2006), the mixed-methods sequential explanatory design focuses on explanation and

understanding through distinct quantitative and qualitative data collection and analysis phases.

According to Venkatesh et al. (2013), this mixed-methods design is instrumental in offering a

richer and more meaningful understanding of the issues under investigation by combining

quantitative and qualitative data. In this context, the quantitative phase offers statistical

evidence, while the qualitative phase develops an in-depth exploration, contextualisation, and

explanations behind the findings. Fielding (2012) suggests that mixed methods allow

triangulation of data from findings in the sequential phases that allow comparison and

integration that enhances the credibility and validity of the findings. According to Polit and

Beck ( 2010), the quantitative and qualitative research approaches contribute to the robustness

of the research that fosters practical implications based on a thorough understanding. From

these insights, the rationale for adopting the mixed method approach in the current study is to

78
address the multifaceted nature of IS to develop a more comprehensive and in-depth

understanding of the systems and users, allowing meaningful contributions to theory, practice,

and policy in the field.

4.2 Research Design

In any research study, the research design is essential for a comprehensive perspective of the

research framework. The research design involves a series of stages made by the researcher

that establish the strategic decisions made in addressing the research problem, questions, and

objectives (Carcary, 2009). Figure 4.1 represents the research design and stages to achieve the

study's main objectives. The current study was accomplished based on the mixed methods

design. According to Kerrigan (2014), the mixed methods design alludes to using qualitative

and quantitative data collection and analysis approaches and comparing or relating the insights

gathered from each approach to better understand the issues under investigation. New

knowledge emerges in the interpretation and discussion of the areas of convergence or

divergence experienced in the qualitative and quantitative results (Abdalla et al., 2018).

Othman et al. (2020) suggest that the mixed method sequential explanatory research design

facilitates the development of comprehensive analysis using mixed methods from diverse

qualitative and quantitative data. This design enables the researcher to capture the richness of

research aspects through qualitative methods while obtaining numerical data for statistical

analysis necessary for quality, credibility, and reliability (Asenahabi, 2019). Researchers using

this research design prioritise the methods equally, although the data analysis is independent.

The results are integrated to develop an overall interpretation based on the convergence

approach.

The sequential quantitative and qualitative design is systematically employed through a

sequential approach, where the initial quantitative phase entails generating and testing the

research hypotheses and exploring the potential factors influencing the phenomenon under

79
study (Harlap et al., 2016). The quantitative phase informs the qualitative phase, considering

that the data collected allows the researchers to delve deeper into quantitative results. Although

the sequential mixed method design requires the two research methods, the current study

largely relied on the quantitative approach. According to Conboy et al. (2012), quantitative

research is considered a stronger approach to developing new knowledge compared to

qualitative methods due to the statistical analysis and generalisation that is not realised in

qualitative studies. Based on the research model of the quantitative study, a qualitative method

was conducted via Machine Learning (ML) techniques to provide deeper insights into the

research factors.

Figure 4.1. Research Design

This study adopted an explanatory research approach to address the main issues emerging in

the research model. It facilitated the generation of findings highlighting the importance of

‘What is happening?’ or ‘What is the impact? The approach helped to discover the cause-and-

effect relationships in different variables of the phenomenon under investigation (Bryman &

80
Bell, 2015). This study was accomplished in three stages. The first was conducting a literature

review to get a broad perspective on AI assistants and cultural theories with their application

for different domains. The second stage was completing the quantitative method, which

involved conducting the pilot study and the primary survey. The third stage was an ML study,

which involved analysing reviews by online customers. The positivist research philosophy

guided the quantitative phase of the study, while the interpretivist research philosophy was

adopted for the qualitative phase. The positivist research philosophy fosters the use of

quantitative methods that offer empirical evidence on the phenomenon under investigation. The

interpretivism research philosophy suggests that reality is a social construct and knowledge can

be developed from the experiences of different subjects (Holden & Lynch, 2004).

Firstly, a literature review was conducted to build a theoretical background and identify the

knowledge gap that guided us in developing the research questions. A conceptual model was

generated to demonstrate the research questions premised on the insights emerging from the

interpretation of literature sources. Using the mixed method sequential explanatory research

design, the quantitative phase was applied, followed by the qualitative research, which entailed

collecting data that could contribute to a more nuanced interpretation of the research outcomes

and provide context-specific understanding.

A quantitative survey was employed for data collection in the quantitative phase. According to

Fakis et al. (2014), surveys are adequate in practically examining research models and

hypotheses because they facilitate the collection of large volumes of data on the issues under

investigation. The deductive approach was instrumental in the quantitative analysis when

testing the research model regarding the customers’ intention to use AI assistants in e-

commerce. In this light, the literature review on cultural factors and AI assistant platforms

highlighted the customers’ attitudes and intentions to use this type of AI assistant for e-

commerce. Chalmers (1990) suggested employing a deductive approach facilitates the

81
development of a logical conclusion from the insights emerging in testing the hypotheses

developed in the study. The proposed hypotheses focused on the pertinent theoretical aspects

emerging from the information systems and marketing literature. Chapter 2 and Chapter 3

present the literature review and conceptual framework that was developed through a deduction

process.

An analysis of online reviews was conducted to focus on the customers’ use of e-commerce

platforms. The ML method was included in this context because the research objective was to

explore this phenomenon in-depth (Bryman & Bell, 2015). Subsequently, a theoretical model

was established to demonstrate the relationships between the various concepts and variables

from the literature review that were empirically assessed. As a result, the theoretical model was

developed to address intention and consumer behaviour using AI assistants in Chapter 3. The

findings of this research contribute to the current knowledge in the area of AI assistants and

interdisciplinary research within culture, providing new opportunities for future research.

4.3 Justification of Research Method Design

As stated earlier, this study applies a mixed methods research strategy described by Ivankova

(2006) using sequential quantitative and qualitative techniques. A mixed-method research

design is a process for collecting, analysing, and mixing both qualitative and quantitative

methods in a particular study to understand a single research problem (Ivankova, 2006).

Quantitative data using survey questionnaires have been used primarily to collect numerical

data to test and validate the research conceptual model. More precisely, quantitative data has

been utilised to develop an initial understanding of the relationship between factors that

influence e-commerce consumer attitudes and intentions. However, online reviews using

qualitative data have been used to identify themes for designing AI assistants that reflect the

key influencing factors on attitudes and intentions towards using AI assistants. Additionally,

natural language processing techniques are utilised to clarify and explore the aspects of the pre-

82
themes that reflect associated variables. The advantages of using a mixed-method approach

have been clearly shown in the information systems literature (Petter & Gallivan, 2004). This

study adopts a mixed-method approach based on the following arguments: According to

Kaplan and Duchon (1988), gathering various types of data using different methods and from

different sources enhances the researchers’ capacity to capture a broader perspective of the

research problem, which leads to the better contextual basis of interpreting and validating the

findings; Greene et al. (1989) suggest that a mixed method approach fosters stronger

conclusions because the different research approaches complement and confirm each other.

Considering the limited research on consumer culture in AI technologies in e-commerce, a

mixed-method approach has been adopted to provide more details about consumers’

knowledge, experience, and adoptions.

4.4 Quantitative Study Approach Overview

Quantitative research was done to evaluate the proposed conceptual model and the critical

relationships among its constructs. This was accomplished by employing various statistical and

analytic approaches in interpreting the data collected from the online survey. A multivariate

data analysis was used to evaluate the relationships in the model. According to Hair et al.

(2013), correlation analysis is used to establish the dependence among two or more factors that

generate results. This analysis focused on how and the degree to which one of the factors

associates with another (Cavana et al., 2001). An analysis using the Partial Least Squares

Structural Equation Modelling (PLS-SEM) was used to highlight the complex cause-effect

relationships emerging in the model.

4.4.1 Measurement Design

Surveys are often used in quantitative studies to present research participants with structured

questions that increase efficiency and accuracy in data collection. Therefore, quantitative data

collected from surveys can be adequate in standardisation, making generalisations to a larger

83
research population (Sidi et al., 2012). According to Kelley et al. (2003), quantitative survey

data can be easily analysed using various statistical techniques to respond to the issues under

investigation. Surveys are instrumental where investigations require high levels of

confidentiality and anonymity and enhance the respondent’s willingness to offer truthful

information. They are also cost-effective to administer, unlike interviews requiring researchers’

participation in the data collection process (Goertzen, 2017).

An online survey was administered to gather adequate data for testing the research hypotheses.

The Qualtrics platform was used for designing the questionnaire and managing the collected

data. The Qualtrics platform is a cloud-based tool used for creating and distributing surveys to

the large online community. It allows users to collect data, analyse the results, and export the

data in various formats. The questionnaire consisted of statements for each factor included in

the conceptual model, as shown in Table 4.1, that required responses using a five-point Likert

scale, ranging from 1 (strongly agree) to 5 (strongly disagree). Appendix A presents the

questions for the factors in the survey questionnaire.

Table 4.1. Measurement Items

Factors Code Items References

Perceived Usefulness PU1 The AI assistant provides useful information. Hsieh & Liao, 2011;
Çelik & Yilmaz, 2011
PU2 The AI assistant provides sufficient content.

PU3 The AI assistant makes it easy to find the content

required.
Perceived Ease of Use PE1 Learning to use the AI assistant is easy for me. Chung et al., 2020

PE2 The interaction with the AI assistant is clear and

understandable.
PE3 I would find it easy to use the AI assistant to
search for what I want.
Personalisation PERS If I changed to another brand, the products and Chung et al., 2020
services would not be as customised as I have
now.
PERS The AI assistant offers products and services that
would be difficult for me to find.
PERS I feel that using the AI assistant and transacting
with it may meet my personal needs.

84
 PERS The AI assistant provides information about
products according to my preferences.
Interactive ICOM My interactions with the AI assistant can be more
Communication productive than face-to-face interactions with in-
store personnel.
ICOM Using AI assistants can be more efficient than
other forms of communication.
ICOM AI assistants can save a tremendous amount of
time.
Attitude AT1 I would have positive feelings towards using the Hassanein & Head,
AI assistant. 2007
AT2 The thought of browsing a product from the AI
assistant is appealing to me.
AT3 It would be a good idea to find the right product
using the AI assistant.
Usage Intention UI1 I will use the AI assistant regularly in the future. Moon & Kim, 2001

UI2 I will frequently use the AI assistant in the future.

UI3 I will strongly recommend others to use the AI

assistant.

The measurement scales considered that the adopted items were edited to fit the objectives of

this study. Different studies have integrated measurement subscales to create a single

measurement scale instead of using multiple separate scales, which enhanced their reliability

(Polkosky & Lewis, 2003; Lewis & Hardzinski, 2015). However, this alternative has some

limitations concerning internal reliability and construct validity. Therefore, the measurement

scales for each construct were validated before the statistical analysis. The AI assistant

interaction simulations were designed for performing most e-commerce functions, such as

searching for information, recommending products, and contacting the brand. The survey

consisted of two scenarios of AI assistant interaction to demonstrate variety in conversation

outcomes. Appendix B presents the AI assistant interaction simulations. The measurement

properties for the latent constructs and their indicators were calculated for construct reliability

and internal consistency. The construct reliability and internal consistency were established

using Cronbach alpha that was of a value higher than 0.7 (Bland & Altman, 1997). The average

85
variance extracted values are higher than the value of 0.5. Therefore, the research instrument

was adequate for data collection in this study.

4.4.2 Sample and Sampling Size

The target sample for this study was individuals who engage in e-commerce activities,

regardless of whether they have used AI assistants or not. The goal was to investigate the usage

intentions towards AI assistant adoption among users and non-users of AI assistants. The

participants of this study were from various geographical locations, providing a cross-cultural

perspective on the usage patterns of AI assistants in e-commerce. This was achieved using the

sample from the professional survey platform Amazon Mechanical Turk (MTurk), assuming

that its user base represents a diverse range of backgrounds and experiences. The inclusion

criteria focused on ensuring the participants demonstrated prior experience with online

shopping and e-commerce activities. The rationale behind this is to ensure that offered

responses are relevant to the current study concerning interactions with AI assistants in e-

commerce. Therefore, participants with no previous experience with AI assistants were

engaged in this study based on the notion that their attitudes could be compared with those with

such experience. This comparison aimed to identify the factors that drive their interest or lack

of interest in using AI assistants in e-commerce. MTurk allows researchers to receive data with

IP addresses to enhance the quality of the responses (Forkus et al., 2022). Then, the sample was

divided into two groups during the data analysis: Eastern and Western groups were identified

based on geolocation techniques of the IP addresses related to the responses (Burnette et al.,

2022). Using two groups from different cultures (Western and Eastern) through the MTurk tool

can be justified for several reasons. The MTurk platform helps researchers reach a diverse

group of participants from different areas, making the study more interesting by showing how

people from different cultures think and act differently (Keith et al., 2024). This makes the

study more valid because it can apply to more cultures than just one (Hitchcock & Nastasi,

86
2024). MTurk also uses IP addresses and location techniques to accurately group people into

Eastern and Western cultures, which helps researchers compare how cultural differences affect

the study's results (Aguinis et al., 2012). This method helps find patterns and differences

between cultures under AI assistants as a recent development technology. Finally, MTurk is

efficient and cost-effective, making it easier to include people from different cultures in studies

than other data collection methods (Keith et al., 2024). However, the grouping was premised

on the assumption that the physical location of the individual aligned with the cultural

characteristics associated with the region. Furthermore, the limitations in using internet

services in some jurisdictions imply that some users might use VPNs and proxies that

misrepresent their geographic location (Khan et al., 2023). Some users from different cultures

might have migrated or travelled to their current physical places. Using the IP addresses also

lacked a clear reflection on the subcultures among the two primary cultures that might

substantially highlight differences in the use of advanced technologies. Therefore, data

collection focused on engaging a diverse population to ensure that participants were

representative of people from different regions. The main consideration for participation was

the willingness to contribute to the current study by giving consent and answering screening

questions about e-commerce activities and AI assistants. The diversity in the research sample

enhanced the potential for generalisation to a broader audience and the practical application of

the research findings.

For structure equation modelling, several analysts and experts adopted several rules of thumb

in predicting sample sizes for investigations that utilise structural equation modelling as an

analysis method. For instance, Hair et al. (2013) noticed that a sample size of 150-300 (6-10

indicators) is appropriate for critical structure equation modelling studies. For this study, the

sample size was determined using the Australian Bureau of Statistics calculator, assuming that

the population of e-commerce users is unknown. The parameters used for determining the

87
sampling size are the following (Confidence level: 95% - Population Size: the option of the

sampling size is left blank due to the population size being very large. The proportion has a

value of 0.5 due to the unknown proportion, and it should be set to 0.5, as this produces a

conservative variance estimate with a confidence interval of 0.05. Consequently, a sample size

of 385 participants was selected for the current study. The two main factors affecting the sample

size calculation were the P value and statistical power. The statistical P value for significance

is 0.05, and the statistical power is generally between 80% and 95%, depending on the size of

the research sample (Whitley & Ball, 2002). According to the Nomogram form for calculating

power, the power of 400 and 385 sample size is 97%, achieving higher power for the testing

validity and sampling size (Whitley & Ball, 2002). Therefore, 400 participants were recruited

for this survey and after pre-processing the data, the participants included were 397.

4.4.3 Data Collection Method

The final survey was conducted in December 2021. At the beginning of the study, the

participants had instructions regarding the definition of AI assistants, which helped to establish

if they had such an experience or not. During the survey, they were shown two examples of AI

assistant interaction simulations to help understand what websites or applications could be

considered AI assistants in e-commerce. During the survey, participants were asked about their

online purchasing history, their experiences with AI assistants and what information they

needed to obtain from them. The survey was developed on the Qualtrics software, and a link

was sent to the MTurk platform. The MTurk users shared the link to access the survey on the

Qualtrics platform, where the data was automatically recorded and stored. MTurk is an online

platform that facilitates crowdsourcing, making it easy for organisations and individuals to

outsource tasks to a diverse and widely distributed workforce in a virtual working environment

(Aguinis et al., 2021). The platform facilitates various tasks ranging from simple data

validation tasks to more complex and subjective undertakings, such as participating in surveys

88
and content moderation. The platform guarantees data accuracy by ensuring that participants

can only engage once on a task (Aguinis et al., 2021). This enhances the quality of the data

collected and contributes to the credibility and reliability of the application of the insights

gathered. Additionally, the global reach of MTurk offers an opportunity to develop insights and

applications that are relevant to a diverse population.

Additionally, the platform offers a strategic approach that reduces the costs and time required

at each stage of collecting data (Ashfaq et al., 2020). This is attributed to the ease of gathering

and annotating large volumes of data that are required in survey studies and the continuous

iterations and corrections. As a sampling technique, Mturk is appropriate for empirical studies

(Chandler & Shapiro, 2016). The application of this approach is premised on previous

marketing and business studies in virtual assistants that used Mturk for their online survey

(Kull & Monahan,2021; Ashfaq et al., 2020; Sheehan & Gottlieb, 2020; Go & Sundar, 2019;

Araujo, 2018). Before conducting the main survey, a pre-testing was conducted to establish if

the items were aligned with the issues of interest and to improve the quality of these questions.

Pre-testing a questionnaire highlights problems that might emerge, compromising it in

collecting primary data. It facilitates gathering preliminary insights to improve clarity, making

it easy for the participants to complete the survey (Ghazi et al., 2018). The pilot study was

conducted with 26 participants. The feedback from the participants helped in modifying the

questionnaire, mainly on the inquiries related to the AI assistant usage and the participants’

previous experiences.

4.4.4 Data Analysis Method

The data analysis approach was drawn from Hair et al. (2010). These are the primary techniques

of multivariate analysis, which are multiple regressions and structural equation modelling. The

various statistical analyses used Excel, IBM SPSS, and SmartPLS software. Several previous

studies have used multivariate analysis techniques and the PLS-SEM to address more than one

89
variable and analyse more than one statistical outcome (Sheehan & Gottlieb, 2020; Araujo,

2018). PLS-SEM is often implemented in marketing and social science research and is

appropriate for studies that collect nonnormal data and can facilitate the use of any size of

research sample (Hair et al., 2019). The PLS-SEM approach allows the validation of the

hypothesis to establish if there is synchronous modelling of correlations experienced between

the independent and dependent variables (Hair et al., 2013). The PLS-SEM was used to identify

the relationships that exist among the independent and dependent variables. Additionally, a

range of experimental and descriptive analysis procedures were used to assess the relationships

between the influential constructs outlined in the theoretical model. Evaluating the correlations

helped to predict the relationships, and multivariate techniques were employed to verify the

relationships identified in the correlations (Hair et al., 2013).

Field (2009) suggests that statistical analysis of any kind should be done after checking for

missing data. Also, normal distribution is necessary to ensure that the data is adequate for

credible and reliable results. The risk of missing data was mitigated by leveraging the features

of the Qualtrics platform in conducting online surveys that prevent incomplete submissions.

The 5-point Likert scale was used to record the survey responses, followed by transforming the

categorical data via SPSS to test the collected data’s normality. After testing the normality of

the data survey, the Spearman correlation approach and multiple regression analysis were used

for the correlation test because the collected data are non-normality distribution, and the model

included various constructs. Measurement scale analysis was part of PLS-SEM strategies to

establish the meanings associated with variables in the model by assessing the reliability and

validity of the scale. Cronbach’s alpha, the composite reliability, and the average variance

extracted tests were used to assess the scale’s reliability and validity. The analysis to confirm

the measurement model was coupled by establishing the relationships between the variables

emerging in the hypothetical structural model and the latent variables. After testing

90
measurement scales to establish reliability and validity, the conceptual model and path

coefficients were assessed to test the significance of the hypotheses. To investigate the

relationships of the research model, the PLS-SEM was employed for path analysis (Hair et al.,

2019).

4.5 Qualitative Study Approach Overview

This section is about the qualitative study that applied ML and natural language processing

techniques for collecting and analysing online reviews. The Latent Dirichlet allocation (LDA)

model is the most used topic model for analysing online reviews (Sutherland et al., 2020). The

LDA model was used as a relative topic model to extract the common topics and explore deep

insights to support the outcome of the quantitative study. The thematic analysis was further

conducted to synthesise and comprehend each topic’s themes of the LDA model. Therefore,

online reviews of e-commerce tools (website and mobile applications) were collected to

provide valuable insights for the AI assistant designs and support the findings of the

quantitative study.

4.5.1 Data Collection Method

Data collected focused on online consumer reviews, which, according to Xie et al. (2014), are

critical sources of insights into consumer decision-making processes. Amblee and Bui (2011)

suggest that online consumer reviews offer social proof concerning the perceptions and

behaviours that emerge among people using a product or service. Suggestively, they enhance

transparency and accountability in research by including the consumers’ perspective about the

products and services (Zhang et al., 2014). Therefore, focusing on online consumer reviews

offered an opportunity to enhance the quality of the current study and its contribution to theory

and practice. The online reviews of technology products of the Louis Vuitton brand were

collected as a case study in the fashion industry e-commerce. The current virtual assistants used

by Louis Vuitton have limited geographic locations, and the CEO of Louis Vuitton is planning

91
to develop virtual assistants to increase their available locations and platforms (Arthur, 2017).

Understanding consumers’ needs and challenges towards the existing online communication

tools in e-commerce could support developers in enhancing the design of virtual assistants.

Therefore, this study aimed to analyse the reviews of the current technologies of Louis Vuitton

to provide practical conclusions for how the brand can enhance its virtual assistants.

For this study, the data was collected from three platforms: Trustpilot, Apple App Store, and

Google Play app store review platforms. Trustpilot is a well-known technology review site. It

is a digital platform connecting companies and customers to build trust and encourage

cooperation. Trustpilot facilitates feedback that can help customers make confident decisions

when purchasing and provides substantial insights for companies that can be used to improve

customer experiences. Apple App Store and Google Play allow users to review applications

that highlight their experiences using the technical aspects of what is offered. Therefore, a web

scraper code was developed in Python language for collecting the reviews of the online e-

commerce applications of Louis Vuitton. Web scraping is an intuitive way of extracting vast

data from sites that are unorganised and scattered. Web scraping assists in collecting

unstructured data and storing it in an organised format. The methods for scraping webpages,

including internet services, APIs, and developing specialised programs. Therefore, the code

was programmed by adopting existing libraries of Python programming language to develop

this study’s reviews scraper. The total collected data was 996, including three attributes: date,

review, rank, and description.

4.5.2 Data Pre-processing

The pre-processing technique was conducted to clean the data before the data analysis (Famili

et al., 1997). Pre-processing entails cleaning raw data to make it usable and transform it into

intelligible formats. Raw or actual data is often inadequately formatted; it can be incomplete

and can be compromised by human errors. Data pre-processing resolves these challenges,

92
making the datasets more adequate for analysis. Therefore, the data pre-processing step was

conducted twice to improve this study’s results. The review pre-processing is an important step

that can influence the success of ML and data mining projects. It speeds up data analysis from

datasets and can eventually impact the efficiency of ML frameworks.

The first step of review pre-processing was changing the case or letter format. Changing the

case entails transforming the texts to lowercase or uppercase so that the word strings are

consistent. Lowercasing is the most common choice in NLP analysis. A lower method was used

for this analysis to convert all review texts to lowercase. The lower method is a Python function

that converts strings to lowercase. Lowercasing is critical for generating a single token of

lowercase and uppercase variants of the same word (Ferrario & Nagelin, 2020). The next step

was removing emoticons, pictographs, transport, and map symbols from the reviews. This step

was essential to remove any non-text object from the reviews. The regular expression model

with a compiled method was used to specify a set of patterns matching the required removed

forms. The next step was removing punctuation marks and digits from using the natural

language toolkit Python library. Then, removing extra spaces and multiple lines methods were

used as the next pre-processing step for cleaning the collected reviews. Strip in Python was

applied to delete the additional lines. “Stop-words” are the frequent words used in sentence

construction. Examples of stop-words in English are the, are, of, in, and. For NLP applications,

such as sentiment analysis, document categorisation, and spam filtering, the stop-words are

removed at the pre-processing stage because they are redundant (Sarica & Luo, 2021).

Therefore, the second time of pre-processing, a method of removing stop words was added to

improve the results. The Natural Language Toolkit (NLTK) library is a standard Python library

offering various algorithms for NLP. The up-to-date packages were downloaded from the

NLTK library to remove the stop words. The following section is about data annotation and

vectorisation, which are essential steps for NLP analysis.

93
4.5.3 Data Annotation and Vectorisation

The collected reviews were classified negatively and positively based on provided ratings and

the sentiment analyser’s outcomes. The ratings of numbers 1, 2, and 3 referred to negative

reviews, and 4 and 5 referred to positive reviews. The collected reviews included 533 negative

and 454 positive reviews. The sentiment analyser-based classification introduced 105 negative

reviews and 233 positive reviews. The data annotation techniques were applied to make text

identifiable to system text recognition to develop such datasets. The data vectorisation method

was applied after data annotation, as it was required for the NLP analysis, such as keywords

extractions and topic modelling algorithms (Hasan et al., 2019). Data vectorisation refers to the

practice of transforming words into digits. It is a technique in NLP that maps phrases

and words from a lexicon to a matching vector of real numbers.

4.5.4 Data Analysis Method

The NLP analysis was used to understand how systems analyse a document's content,

subtleties, and emotions. NLP was adequate because it could precisely extract the data and

insights included within the document and then arrange them into the appropriate categories.

The NLP techniques used in this study entailed a method to address the research questions and

for keyword extraction from collected reviews. Keyword extraction provides a straightforward

and adaptable method for extracting top-frequency words from documents. The Frequency

Inverse Document Frequency (TF-IDF) model was adopted to identify the most frequently

occurring words in the collected reviews. The TF-IDF model could assign more excellent

relevance ratings to words appearing in fewer corpus documents. The TF-IDF compares a

word’s frequency in a document to its overall frequency within-corpus. The next step was

identifying the most frequent of the three-word combinations and the most frequent of the top

words using the TF-IDF model for the negative and positive collected reviews. Then, the

produced data visualisations consisted of plot bars and word clouds representing negative and

94
positive reviews. A cloud of words is also a text representation technique that defines the

appearance of words in a document.

A topics model was conducted for this study to capture the critical topics of the current

communication tools in e-commerce that matter to consumers. AI-powered text analysis

employs a broad range of approaches or techniques to interpret language naturally; topic

modelling is one of them (Nikolenko et al., 2017). Topic modelling is an ML approach that

automatically evaluates text data to discover a cluster of words from documents (Maier et al.,

2018). Topic modelling was conducted to understand what consumers say about certain aspects

of the current communication tools. The collected reviews included for analysis were 987 rows

with 235892 words. The collected reviews had twenty topics, with ten passes found via the

topic modelling for better-displaying topics.

After the topic modelling method, the thematic analysis was applied to capture the theme of

each topic of this study. Thematic analysis is a qualitative strategy that entails familiarising

with the dataset to establish, evaluate, and report on patterns and themes (Kiger & Varpio,

2020). In this context, the deductive thematic analysis was conducted to find the pre-defined

themes based on the main variables of the research model in the quantitative study.

4.6 Integration of Data and Interpretation of Findings

The application of the mixed methods started with quantitative data collection and analysis and

built to qualitative data collection and analysis, which led to the interpretation. The quantitative

results were used as a new instrument for the qualitative strand. The rationale for conducting

the mixed method design allows for a comprehensive interpretation of the research topic by

triangulating quantitative and qualitative data. This design enables the researcher to capture the

richness of research aspects through qualitative methods while obtaining numerical data for

statistical analysis (Asenahabi, 2019). The design of this method facilitates a sequential

95
approach, where the initial quantitative phase helps generate hypotheses and explore potential

factors influencing the phenomenon under study. This quantitative exploration informs the

subsequent qualitative phase, allowing for the validation and complement of findings.

Furthermore, the sequential mixed methods design enables the researcher to dig deeper into

quantitative results. Therefore, the qualitative data were collected to interpret the research

outcomes better and provide context-specific understanding (Subedi, 2016).

The integrated quantitative and qualitative methods are intrinsically linked and mutually

illuminating, thereby creating a comprehensive understanding of issues under investigation

(Woolley, 2009). In mixed methods research, integration between two methods refers to the

specific relationship between these methods that retain their paradigmatic elements, although

they are aligned in pursuing the shared objective of improving and creating new knowledge

(Guetterman et al., 2018). Mixed methods foster the integration of findings and connection of

the two methods to develop meta-inferences that create a better understanding (Ivankova et al.,

2006) and more insights concerning a particular phenomenon from different perspectives

(Shannon-Baker, 2016). Consequently, mixed-method research outcomes can be unhelpful if

there is a failure to link, connect, or integrate the quantitative and qualitative insights (Moran-

Ellis et al., 2006). The concept of mixing methods extends beyond using two or more research

methods in the same study. Uprichard and Dawney (2019) argue that data integration must lie

in the extent to which data is interpreted from different sources meaningfully. Suggestively,

there are three stages where integration can be experienced in a mixed methods study– design,

methods, and interpretation and reporting (Fetters et al., 2013).

In the current study, there was no integration at the design level because a sequential mixed

methods design was adopted whereby the quantitative phase was conducted first and informed

the qualitative phase. However, the merging approach was adopted, which entails comparing

and interpreting the qualitative and quantitative results and establishing how the insights from

96
each method complement the other. Data merging is carried out following the quantitative

statistical analysis and an analysis of the qualitative data (Jantsch & Neves, 2022). Deductive

thematic analysis was employed premised on predetermined codes for the qualitative data. This

approach ensured that the qualitative data was aligned with the key constructs measured in the

quantitative phase, which increased efficiency in merging the data sets. Data integration in the

narrative can follow many approaches, such as weaving, contiguous, or staged approaches

(Fetters et al., 2013). The contiguous narrative approach was adopted, which entailed

presenting and separating the narrative for the quantitative and qualitative findings in Chapters

5 and 6, respectively. However, the insights from both approaches were later merged in Chapter

6.

4.7 Ethics of the Research

This research involves human participants for data collection. To ensure the anonymity and

ethical protection of the participants, the Human Research Ethics Committee (HREC) – at the

University of Technology Sydney approved the ethical form along with the protocol number

approval, ETH 20-5011.

4.8 Chapter Summary

This chapter discusses the overview of the research methodology adopted in the current

investigation to test the proposed conceptual model concerning AI assistant applications in e-

commerce. This chapter addresses the research methodology, the research approach, and the

relevant analytical techniques used. The study was accomplished using a mixed-method design

that integrates quantitative and qualitative approaches, including machine-learning research

approaches. The approach began with a review of relevant literature sources and establishing

logical relationships between different variables. This was followed by a quantitative phase

that offered further insights using empirical evidence. Following the quantitative phase, the ML

97
approach provided deep insights into the identified constructs of the research model. The

quantitative data set analysis involved various statistical techniques that include primary

descriptive analyses, multivariate and correlation techniques, and PLS-SEM using Excel, IBM

SPSS and Smart PLS (Version 4) software. The analysis allowed testing of the conceptual

model and hypotheses to develop the final empirical model, highlighting the interrelationships

emerging among the different variables. The second phase involved an ML approach for

analysing the reviews of e-commerce applications of the fashion brand Louis Vuitton. The

following chapter addresses the data analysis of the quantitative phase.

98
Chapter 5. Quantitative Data Analysis and Results

The primary research objective centred on exploring users' perspectives on the intention to use

AI assistants in e-commerce. Following the initial data analysis, the conceptual model's

representation was formulated. A pivotal aspect of the conceptual model involves the

application of the Technology Acceptance Model within the context of AI shopping assistants.

This model clarifies the connections between the adoption of AI assistants among both users

and non-users of online shoppers. Quantitative data analysis was conducted to test the proposed

research model. The quantitative data analysis includes two parts: first, examining the

appropriateness of the collected data in terms of validity and reliability to ensure the construct

validation is satisfactory; second, testing the hypotheses on the relationships between the

research constructs. Therefore, this chapter presents several summarised tables and figures

describing the findings of the quantitative data analysis.

Initiating with a descriptive analysis of the survey and the demographic data of respondents,

this chapter progresses to undertake data cleaning, testing, and validation for normality and

effective representation. The subsequent stages involve the application of the structural

equations modeling procedure, the development of the conceptual model, and the formulation

of hypotheses. Following these steps, the process includes data screening, conducting validity

checks for the final structural model, and an investigation into the effects. In the end,

multigroup analysis was presented for testing the model related to cross-culture investigation.

5.1 Questionnaire Survey and Participants Demographics

As described in Methodology Chapter 4, a survey instrument was developed to examine the

factors influencing online consumer attitudes and intention to engage with AI assistants in e-

commerce. The survey questionnaire contained four constructs and two internal factors with

99
nineteen items. The five-point Likert scale was used for the survey instrument. The following

sections describe the survey and the participants' profiles.

5.1.1 Questionnaire Survey

A questionnaire survey was conducted online, filled out by individuals with experience with e-

commerce and online shopping. The survey was conducted on the most popular data collection

platform, MTurk, among scholars (Kull & Monahan, 2021; Ashfaq et al., 2020). According to

Hitlin (2016), most users of MTurk reside in different regions around the world. Participants

in this study came from various age groups, educational levels, cultural backgrounds, online

purchase habits, and experience statuses with AI assistants. The type of questionnaire was a

self-administered questionnaire using the Qualtrics platform. The self-administered

questionnaire is more efficient than the interviewer-administered questionnaire in terms of no

possibility of interviewer bias (Sakshaug et al., 2017).

The validated measurement scales of this study are adopted from prior literature to ensure the

reliability and validity of the survey items, as indicated in the methodology Chapter 4. These

items were adapted and edited to fit the objectives of this study. Respondents were required to

fill out the questionnaire that included both open-ended and closed-ended questions. The

questionnaire included items to measure all factors used in the conceptual research model.

Consistent with the literature on measurement scale development (Polkosky & Lewis, 2003;

Lewis & Hardzinski, 2015), the measurement scale is tested for internal reliability and

construct validity, as shown in Table 5.19. In total, 400 participants completed the survey (more

details on the data collection procedure in Chapter 4). After removing incomplete responses, a

total of 397 responses remained for data analysis. The following section describes the

respondents’ profiles of the survey.

100
5.1.2 Demographic Characteristics of Respondents

The demographic profile of the participants is presented in the subsequent table.

Table 5.1. Demographic Profile of Respondents.

Demographic Characteristics Frequency (n) Percentage (%)

Female
153 39
Gender Male
244 61
Western
293 74
Culture Eastern
104 26
18 - 25
22 5.54
25 - 35
209 52.64
35 - 45
98 24.69
Age (yrs.) 45 - 55
46 11.59
55 - 65
15 3.78
Greater than 65
7 1.76
Bachelor's degree
262 65.99
Master's degree
84 21.16
High School Degree
Degree 42 10.58
Doctorate Degree
7 1.76
Certificate IV and associates
2 0.50
$150,000 or more
4 1.01
Less than $10,000
37 9.32
$80,000 to 150,000
Income 74 18.64
$10,000 to $40,000
110 27.71
$40,000 to $80,000
172 43.32
Users
Prior experience using 312 78.59
Non-users
AI assistants 85 21.41
3-6 months ago.
135 34.01
Within the last three months.
The habit of purchasing 118 29.72
6-12 months ago.
a good online 104 26.20
More than a year ago.
28 7.05

From Table 5.1., it can be observed that the gender distribution breakdown among the

respondents indicates that 61% identified as male, while 39% identified as female. This gender

distribution demonstrates a slight majority of male participants in the study and is consistent

with the average of people who use AI chatbot technologies (Said et al., 2022; Chen et al.,

2021). Figure 5.1 illustrates the presentation of these results.

101
 Figure 5.1. Gender of participants

Considering the cultural background of respondents, the respondents were divided into two

cultures (Western and Eastern) based on the geo-location’s technology (IP address). The

responses’ geo-locations were identified in eleven countries. The location of each country was

classified into either Western or Eastern cultures. The total number of Western respondents was

293, and the respondents were from the cultures of the United States of America, the United

Kingdom, Brazil, Australia, Italy, and France. In contrast, the total number of Eastern

respondents is 104, and the respondents were from the cultures of India, China, Bangladesh,

Azerbaijan, and Turkey. This classification was performed with the acknowledgment that

several other countries can be added to each group (Hofstede, 1983). This distribution shows

that the growing use of online shopping and AI agents is a worldwide phenomenon. While the

growth varies unevenly from country to country, the Western respondents were a significant

majority of the respondents. This distribution is consistent with previous studies that found

consumers from an individualistic culture (Western) are more likely to use the Internet for e-

commerce than those from a collectivistic culture (Eastern) (Zhou et al., 2017; Hofstede, 2011).

Furthermore, Table 5.1 provides the age distribution of the respondents. The data reveals that

the age group with the most respondents was 25-45, accounting for 77.3%. This age range is

particularly significant as it represents the prime demographic for the online shopper group

(Fokina, 2022). The high representation of respondents from this age group suggests that the

study sample effectively captured individuals most likely to engage with e-commerce channels.

102
The next age group regarding respondents’ representation was above 45, comprising 15.3% of

the total, and 5% of the participants fell within the 18-25 age range. This finding indicates the

inclusion of participants from a slightly older demographic, demonstrating a diverse age

distribution within the study sample.

Moreover, 65.9% of respondents possess a bachelor's degree, with 22.92% holding a

postgraduate degree and 10.58% having a high school degree. When comparing these

proportions to the educated population aged 25 to 45, participants in this study exhibit higher

educational levels. Additionally, individuals with high school degrees have a relatively minor

representation in the sample compared to global statistics (2021). This indicates that individuals

knowledgeable about AI technologies tend to be well-educated compared to the international

population. Figure 5.2 provides a visual representation of these results.

Figure 5.2. Education level of the participants

In terms of annual household income, 43.3% of respondents reported an annual household

income range of $40,000 to $80,000, 27.7% of the respondents have an annual household

income range of $10,000 to $40,000, and 18.6% of the respondents reported an annual

household income range of $80,001 to $150,000. The percentage of respondents engaging in

online shopping aligns with the average observed among experienced online purchasers

(Hernández et al., 2011). Regarding AI assistants, 79% of respondents had utilised an AI

assistant. This high percentage indicates that the participants have the necessary experience

103
and familiarity with AI assistants, positioning them as potential users of AI assistants as an e-

commerce tool. Figure 5.3 highlights the participants' interest in obtaining specific information

or services from AI assistants. The most prominent area of interest reported by the respondents

was obtaining the product's price, with 136 respondents (44%) expressing their interest in this

aspect. Following closely, 118 respondents (38%) indicated an interest in product-related

information, while 24 respondents (8%) expressed interest in promotions. These findings

demonstrate the specific focus areas for participants when engaging with an AI assistant,

providing valuable insights into their preferences and requirements. Table 5.2 presents the

reasons provided by respondents for not interacting with the AI assistant.

Figure 5.3. Information Obtained from AI Assistants

Table 5.2. Reasons for Not Using AI Assistants by Respondents

Reasons Frequency Percentage

I prefer interaction with human agents. 38 45

I could not find the chat label easily. 13 15

I would need to spend too much time
9 11
interacting with the AI assistants.
I feel AI assistants do not match their search
8 9
preferences.
I did not need any help 7 8

I am not familiar with AI assistants. 5 6

There was no AI assistant. 3 4

I feel AI assistants do not match their purchase
2 2
preferences.

104
The top three reasons reported were a preference for human agents (45%), difficulty in locating

the chat label quickly (15%), and the need to spend too much time interacting with the AI

assistant (11%). These findings reflect the respondents’ inclination towards human interaction

and suggest that the AI assistants' ability to provide human-like responses could cater to the

preferences of participants who desire to interact with human agents. This aligns with previous

research (Gnewuch et al., 2018; Hill et al., 2015) that emphasises the importance of human-

like responses in AI assistants to address customer preferences. These findings shed light on

the participants' usage patterns, preferences, and experiences with AI assistants. The high

percentage of respondents who had used AI assistants demonstrates their familiarity with such

technology, positioning them as potential users of AI assistants in an e-commerce context. In

terms of purchasing behaviour and the recent product purchase timeframe, 34% of the

respondent’s total had purchased products within the last 3-6 months, 29.7% of the respondents

purchased a product within the last three months, 26.2% of the respondents reported their most

recent product purchase occurring 6-12 months ago and 7% of them reported making their most

recent product purchase over a year ago.

5.2 Data Examination

Before conducting PLS-SEM, a data examination was performed, encompassing multiple

essential steps. This examination included the elimination of missing records, representation

of answers, assessment of the variance of variables within constructs, and evaluation of

construct reliability and validity. This crucial process is essential as it ensures that the data is

appropriately prepared for model construction (Berger & Zhou, 2014). Since modelling

methods often have specific assumptions that data must adhere to, a thorough data examination

becomes imperative to verify whether these requirements are met. Through this step, it can be

determined if the data satisfies all the necessary requirements for PLS-SEM, enabling the

105
establishment of sets of constructs, each comprising a specific number of variables. Hence,

facilitating the formation of a set of constructs, each comprising a specific number of variables.

5.2.1 Missing Data Analysis

To successfully evaluate data, data must first be observed for missing values (Chhabra et al.,

2017). However, as outlined in Section 5.1.1, all missing data were eliminated for this study

before additional analyses. Table 5.3 to Table 5.6 presents the responses of the primary

constructs of this study. The missing data analysis was performed using descriptive statistics,

as shown in the descriptive analysis results for each construct (Table 5.7 to Table 5.12). In this

phase, the study excluded rows with missing data, specifically in cases where more than two

constructs and their corresponding items were left unanswered.

Table 5.3. Responses of Perceived Usefulness and Ease of Use

Perceived usefulness and ease of use 5-point Likert scale

Item Statement 5 4 3 2 1 Total
PU1 The AI assistant provides useful information. 85 225 68 13 6 397

PU2 The AI assistant provides sufficient content. 123 168 76 26 4 397

PU3 The AI assistant makes it easy to find the content 102 197 70 11 17 397
required.

PE1 Learning to use the AI assistant is easy for me. 105 216 64 7 5 397

PE2 The interaction with the AI assistant is clear and 146 164 58 26 3 397
understandable.

PE3 I would find it easy to use the AI assistant to 110 211 54 12 10 397
search for what I want.

Table 5.3 shows responses on a 5-point Likert scale for perceived usefulness and perceived

ease of use (PU1, PU2, PUE, PE1, PE2, and PE3) with 397 participants. Most participants rated

high (5) and (4) as strongly agree and agree’ responses for usefulness and ease of use, indicating

positive views of the AI assistants.

106
 Table 5.4. Responses of AI Assistant Capabilities

Personalisation and Interactive Communication 5-point Likert scale

Item Statement 5 4 3 2 1 Total
PER1 If I changed to another brand, the products and 62 187 107 32 9 397
services would not be as customised as I have
now.

PER2 The AI assistant offers products and services that 86 158 80 57 16 397
would be difficult for me to find.

PER3 I feel that using the AI assistant and transacting 73 197 99 17 11 397
with it may meet my personal needs.

PER 4 The AI assistant provides information about 97 192 88 17 3 397

products according to my preferences.

IC1 My interactions with the AI assistant can be more 55 200 101 29 12 397
productive than face-to-face interactions with in-
store personnel.

IC2 Using AI assistants can be more efficient than 97 165 94 34 7 397

other forms of communication.

IC3 AI assistants can save a tremendous amount of 93 203 73 16 12 397

time.

The same can be observed from Table 5.4, which shows 397 participants' 5-point Likert scale

responses for personalisation and interactive communication (PER1-4, IC1-3). Most

participants provided high ratings (5) and (4), indicating favourable perceptions of the AI

assistant's capabilities.

Table 5.5. Responses of Attitude towards the Use

Attitude 5-point Likert scale

Item Statement 5 4 3 2 1 Total

AT1 I would have positive feelings towards using the AI 81 203 90 15 8 397
assistant.

AT2 The thought of browsing a product from the AI 109 168 78 32 10 397
assistant is appealing to me.

AT3 It would be a good idea to find the right product using 95 205 64 18 15 397
the AI assistant.

Table 5.5 displays the responses of 397 participants on a 5-point Likert scale, expressing their

attitude towards utilising the AI virtual assistant (AT1, AT2, and AT3). A significant majority

107
assigned high ratings (5) and (4), revealing their positive views of the AI assistant's capabilities

and potential.

Table 5.6. Responses of Usage Intention

Usage Intention 5-point Likert scale

Item Statement
5 4 3 2 1 Total
UI1 I will use the AI assistant regularly in the future. 71 206 92 22 6 397
UI2 I will frequently use the AI assistant in the future. 115 167 73 36 6 397
UI3 I will strongly recommend others to use the AI 104 178 76 23 16 397
assistant.

A significant majority of the participants' responses were on a scale for usage intention (UI1,

U1, and U3) with high ratings (5) and (4), suggesting a strong intention to use the AI assistant

regularly and frequently in the future. Also, many participants expressed a willingness to

recommend others to use the AI assistant, reflecting a positive belief in its potential benefits.

5.2.2 Descriptive Data Analysis

Table 5.7 to Table 5.12 demonstrate the descriptive statistics for each construct of the survey

study separately. Descriptive statistics include two essential factors: standard deviations and

standard errors of the mean. The standard deviation (SD) indicates how accurate the average

value of the collected data was. Nevertheless, the standard error of the mean (SE) helps

determine how accurately a particular sample reflects the population (Dubois et al., 2015). The

SD and SE values are small across all the tested constructs of this study. Thus, the sample can

be considered a reliable reflection of the population.

Table 5.7. Descriptive Statistics for Perceived Usefulness

N Mean SD Variance Skewness Kurtosis

VAR
Statistic Statistic SE Statistic Statistic Statistic SE Statistic SE
PU1 397 3.93 0.040 0.806 0.649 -0.953 0.122 1.758 0.244
PU2 397 3.96 0.046 0.924 0.854 -0.725 0.122 0.128 0.244
PU3 397 3.90 0.048 0.960 0.921 -1.153 0.122 1.618 0.244

108
The descriptive analysis results are essential for observing the actual variance values, which

help depict the homogeneity of variables within factors. Table 5.7 displays the mean, standard

error, and standard deviation for the perceived usefulness construct, which is assessed using

three questions. The mean values for each question are consistently above 3, signifying that,

on average, participants express agreement regarding the helpfulness of using AI assistants.

Particularly, participants tend to mostly agree that utilizing AI assistants provides useful

information (mean=3.93, SD=0.806). Additionally, the table indicates that skewness scores fall

between -0.5 and -1, and kurtosis scores are less than three, suggesting deviations from a

normal distribution.

Table 5.8. Descriptive Statistics for Perceived Ease of Use

VAR N Mean SD Variance Skewness Kurtosis

Statistic Statistic SE Statistic Statistic Statistic SE Statistic SE
PE1 397 4.03 0.039 0.781 0.610 -0.915 0.122 1.835 0.244
PE2 397 4.07 0.046 0.917 0.841 -0.885 0.122 0.335 0.244
PE3 397 4.01 0.044 0.873 0.763 -1.199 0.122 2.145 0.244

The mean scores for PE1, PE2, and PE3 are 4.03, 4.07, and 4.01, respectively, as shown in

Table 5.8, indicating that, on average, participants perceive the AI assistant to be easy to use.

The standard deviations (SD) for PE1, PE2, and PE3 are 0.039, 0.046, and 0.044, respectively,

which suggest relatively low variability in responses. Skewness values ranging from -0.915 to

-1.199 and Kurtosis values below three suggest that the variables show non-normal

distribution.

Table 5.9. Descriptive Statistics for Interactive Communication

VAR N Mean SD Variance Skewness Kurtosis

Statistic Statistic SE Statistic Statistic Statistic SE Statistic SE
ICOM1 397 3.65 0.046 0.914 0.835 -0.781 0.122 0.684 0.244
ICOM2 397 3.78 0.049 0.968 0.938 -0.594 0.122 -0.072 0.244
ICOM3 397 3.88 0.046 0.916 0.839 -1.027 0.122 1.402 0.244

109
Table 5.9 presents descriptive statistics for interactive communication. Mean scores indicate

positive perceptions of the participants regarding the perceived interactive communication of

AI assistants. Low variance suggests homogeneity. Skewness (-0.781 to -1.027) and Kurtosis

(<3) show non-normal distributions.

Table 5.10. Descriptive Statistics for Personalisation

VAR N Mean SD Variance Skewness Kurtosis

Statistic Statistic SE Statistic Statistic Statistic SE Statistic SE
PERS1 397 3.66 0.046 0.915 0.837 -0.621 0.122 0.324 0.244
PERS2 397 3.61 0.055 1.097 1.204 -0.569 0.122 -0.444 0.244
PERS3 397 3.77 0.045 0.895 0.801 -0.818 0.122 1.024 0.244
PERS4 397 3.91 0.042 0.836 0.700 -0.592 0.122 0.334 0.244

The mean scores of personalisation variables are positive (3.61 to 3.91), as shown in Table

5.10. Low variance indicates homogeneity. Skewness (-0.621 to -0.818) and Kurtosis (<1.204)

suggest non-normal distributions of this factor.

Table 5.11. Descriptive Statistics for Attitude

VAR N Mean SD Variance Skewness Kurtosis

Statistic Statistic SE Statistic Statistic Statistic SE Statistic SE
AT1 397 3.84 0.043 0.860 0.740 -0.813 0.122 1.122 0.244
AT2 397 3.84 0.050 0.999 0.997 -0.778 0.122 0.212 0.244
AT3 397 3.87 0.048 0.953 0.908 -1.118 0.122 1.438 0.244

Upon observing the perceived attitude construct, which comprises three questions, it's evident

that the mean values are positive (ranging from 3.84 to 3.87). This indicates that, on average,

participants agree with the statements. Low variance suggests homogeneity. Skewness (-0.778

to -1.118) and Kurtosis (<0.997) indicate non-normal distributions.

Table 5.12. Descriptive Statistics for Usage Intentions

VAR N Mean SD Variance Skewness Kurtosis

110
 Statistic Statistic SE Statistic Statistic Statistic SE Statistic SE
UI1 397 3.79 0.043 0.852 0.726 -0.717 0.122 0.760 0.244
UI2 397 3.88 0.049 0.980 0.960 -0.727 0.122 -0.002 0.244
UI3 397 3.83 0.051 1.011 1.023 -0.943 0.122 0.705 0.244

The construct of usage intention comprises three questions. As shown in Table 5.12, presenting

the descriptive statistics for each question, it is noticeable that the mean values for each item

range from 3.79 to 3.88. This suggests that, on average, participants exhibit a willingness to

use AI assistants and recommend them to others. Skewness (-0.717 to -0.943) and Kurtosis

(<1.023) show non-normal distributions.

5.2.3 Assessment of Normality

The normality test is essential to the overall analysis of study data (Poncet et al., 2016). The

two critical factors of normality are skewness and kurtosis (Cain et al., 2017). The skewness is

a symmetry measurement; kurtosis signifies the peakiness of the data spread (Cain et al., 2017).

The acceptable kurtosis and skewness values for average data spread between -2.00 and +2.00

(Bono, 2019). Therefore, if the kurtosis is less than 3, the dataset has lighter tails than a normal

distribution (less in the tails), and if the skewness is less than -1 or greater than 1, the

distribution is highly skewed (Field 2013). Therefore, as represented in Tables 5.7 to 5.12, the

collected data for all variables are not normally distributed, based on the skewness and kurtosis

values of this research that do not fall within the recommended range. Generally, when the

sample size is more than 100, the Kolmogorov-Smirnov method is used for the normality test

(Yazici & Yolacan, 2007). If the significance value (sig) is more significant than 0.05, then the

data is stated to be normally distributed (Sormin et al., 2019). Tables 5.13 to 5.18 present the

test of Kolmogorov-Smirnov for each construct. The departure of collected data from normality

has several implications for the analysis and interpretation of results. Non-normal data can

affect the validity of parametric statistical tests such as t-tests or ANOVA, which assume

111
normality in the data distribution (Hau & Marsh, 2004). Using these tests on non-normal data

can lead to inaccurate p-values and potentially incorrect conclusions about the significance of

relationships or differences between variables. Therefore, it is essential to use appropriate

statistical methods that do not rely on normality assumptions, such as non-parametric tests like

bootstrapping techniques. These methods can provide more reliable results and accurate

representation of the data's characteristics that are not biased by the non-normal distribution of

the data.

Table 5.13. One-Sample Kolmogorov-Smirnov Test for PU

PU1 PU2 PU3

N 397 397 397
Normal Mean 3.93 3.96 3.90
Parametersa,b Std. Deviation 0.806 0.924 0.960
Most Extreme Absolute 0.314 0.251 0.296
Differences Positive 0.252 0.172 0.200
Negative -0.314 -0.251 -0.296
Test Statistic 0.314 0.251 0.296
Asymp. Sig. (2-tailed)c 0.000 0.000 0.000

For this study, the data contains 397 samples; therefore, the Kolmogorov-Smirnov method was

conducted to test for normality, as demonstrated in Tables 5.13 to 5.18. The significance values

in all tables are less than 0.05 (alpha value) for all variables of this study. Thus, the collected

data of this study are not normally distributed.

Table 5.14. One-Sample Kolmogorov-Smirnov Test for PE

PE1 PE2 PE3

N 397 397 397
Normal Mean 4.03 4.07 4.01
Parametersa,b
Std. Deviation 0.781 0.917 0.873
Most Extreme Absolute 0.293 0.251 0.306
Differences
Positive 0.251 0.162 0.225
Negative -0.293 -0.251 -0.306
Test Statistic 0.293 0.251 0.306
Asymp. Sig. (2-tailed)c 0.000 0.000 0.000

112
 Table 5.15. One-Sample Kolmogorov-Smirnov Test for ICOM

ICOM1 ICOM2 ICOM3

N 397 397 397
Normal Mean 3.65 3.78 3.88
Parametersa,b Std. Deviation 0.914 0.968 0.916
Most Extreme Absolute 0.293 0.248 0.298
Differences Positive 0.211 0.167 0.213
Negative
-0.293 -0.248 -0.298
Test Statistic 0.293 0.248 0.298
Asymp. Sig. (2-tailed)c 0.000 0.000 0.000

Table 5.16. One-Sample Kolmogorov-Smirnov Test for PERS

PERS1 PERS2 PERS3 PERS4

N 397 397 397 397

Normal Mean 3.66 3.61 3.77 3.91

Parametersa,b
Std. Deviation 0.915 1.097 0.895 0.836
Most Absolute 0.273 0.254 0.283 0.269
Extreme Positive 0.198 0.143 0.213 0.215
Differences Negative -0.273 -0.254 -0.283 -0.269
Test Statistic 0.273 0.254 0.283 0.269
Asymp. Sig. (2-tailed)c 0.000 0.000 0.000 0.000

Table 5.17. One-Sample Kolmogorov-Smirnov Test for AT

AT1 AT2 AT3

N 397 397 397
Normal Mean 3.84 3.84 3.87
Parametersa,b Std. Deviation 0.860 0.999 0.953
Most Extreme Absolute 0.289 0.261 0.308
Differences Positive 0.223 0.162 0.208
Negative -0.289 -0.261 -0.308
Test Statistic 0.289 0.261 0.308
Asymp. Sig. (2-tailed)c 0.000 0.000 0.000

Table 5.18. One-Sample Kolmogorov-Smirnov Test for UI

UI1 UI2 UI3
N 397 397 397
Normal Mean 3.79 3.88 3.83
Parametersa,b Std. Deviation 0.852 0.980 1.011
Most Extreme Absolute 0.295 0.259 0.276
Differences Positive 0.224 0.161 0.173
Negative -0.295 -0.259 -0.276

113
 Test Statistic 0.295 0.259 0.276
Asymp. Sig. (2-tailed)c 0.000 0.000 0.000

5.3 Research Model Assessment

The proposed model assessments of a measurement model for the validity and reliability of

latent constructs and a structural model for addressing the hypotheses and answering the

research questions were assessed, as discussed in Chapter 4. This section discusses the results

of the research model assessment via applying the partial least squares structural equation

modelling (PLS-SEM) method. PLS-SEM is a multivariate technique that allows unobserved

testing constructs, measurement, functional, predictive, and hypotheses of various indicators

while considering measurement errors when statistically examining data (Hair et al., 2017).

The PLS-SEM method has the critical purpose of estimating, identifying, and evaluating the

linear relationships between observed and unobserved constructs (Hair et al., 2017). Kock

(2016) estimated path coefficients for hypothesis testing via the PLS-SEM method. The

structural equation modelling method has two main parts: measurement and structural model

assessment, as shown in Figure 5.4. The measurement model represents the relationships

between the variables and the constructs. Therefore, the measurement model can determine if

the constructs are accurately measured. However, the structural model represents the

relationship between the factors only. Therefore, the structural model tests the hypothesised

relationships (Kock, 2016). The structural equation modelling method is popular in marketing

and social science studies as it is suitable for nonnormal data and supports small and large

sample sizes (Hair et al., 2019). The results of the research's measurement and structural models

are presented in the following sections.

114
 Figure 5.4. Two Main SEM Components

5.3.1 Measurement Model Assessment

The assessment of the validity and reliability of latent constructs involved the measurement

scale. The quality of this scale is crucial for establishing the credibility and accuracy of research

findings (Heale & Twycross, 2015). Consequently, scale quality is commonly determined

through concepts of reliability and validity (Heale & Twycross, 2015). Therefore, Cronbach's

alpha was used to test scale reliability, while convergent validity was measured via composite

reliability, assessing the correlation of items within the construct. Additionally, discriminant

validity was evaluated by measuring the average variance extracted and examining the

correlation of items with those of other constructs. These measures collectively contribute to

ensuring the robustness and accuracy of the research outcomes.

5.3.1.1 Internal Consistency and Scale Reliability

Table 5.19 provides the calculated scale reliability for all latent constructs employed in the

current study. This verification process ensures the validity of the indicators within the research

model through an assessment of internal consistency. Internal consistency, as defined by Tang

et al. (2014), reflects the extent to which survey responses for all constructs remain uniform

within a single measurement scale. The reliability of this measurement scale was evaluated

using Cronbach's alpha, a measure of the estimated correlation among a set of indicators or

items, as outlined by Bland and Altman (1997). Taber (2018) suggests that Cronbach's alpha

values within the range of 0.60 to 0.70 are deemed acceptable, a viewpoint supported by Kline

115
(2015) who considers a Cronbach's alpha exceeding 0.60 as satisfactory. Construct reliability

and validity results are detailed in Table 5.19.

Table 5.19. Measurement model- Construct Reliability and Validity

Construct Cronbach's alpha Composite reliability Composite reliability AVE Result

(rho_a) (rho_c)
AT 0.763 0.764 0.864 0.679 Acceptable
ICOM 0.721 0.727 0.842 0.641 Acceptable
PE 0.742 0.744 0.853 0.660 Acceptable
PERS 0.669 0.680 0.798 0.498 Acceptable
PU 0.766 0.766 0.865 0.681 Acceptable
UI 0.794 0.796 0.879 0.708 Acceptable

The satisfactory reliability suggests the presence of internal consistency among the research

constructs. The reliability at the construct level was examined, as shown in Table 5.19, which

presents Cronbach's alpha values for all constructs. The reliability coefficients (Cronbach's

alpha values) and the internal consistency of all the constructs of this study exceed the 0.6 level,

indicating sufficient construct reliability (Kline, 2015; Field, 2009).

5.3.1.2 Assessment of Scale Validity and Model Fit

Regarding validity, two validity scales are considered: convergent and discriminant validity.

Average Variance Extracts (AVE) were utilised for convergent validity with a threshold value

of 0.5 or greater. According to Fornell and Larcker (1981), if the AVE value is below 0.5 and

the composite reliability exceeds 0.6, the construct's convergent validity is still acceptable.

Table 5.19 illustrates that all constructs have AVE values very close to 0.5 or greater than 0.6,

indicating their convergent validity (Hair et al., 2019; Zaiţ & Bertea, 2011). For discriminant

validity, the Fornell-Larcker criterion and cross-loadings were employed. Table 5.20 presents

the Fornell-Larcker criterion analysis. Additionally, Table 5.20 shows that all the bold diagonal

values are more significant than the horizontal and vertical diagonal values, confirming

discriminant validity. Furthermore, Table 5.21, displaying the discriminant validity cross-

loading, demonstrates that all the self-loading values of the individual items are more

116
significant than the cross-loading of other items, reinforcing the discriminant validity of all

constructs (Fornell & Larcker, 1981).

Table 5.20. Fornell-Larcker Criterion Analysis

VAR AT ICOM PE PERS PU UI

AT 0.824
ICOM 0.766 0.800
PE 0.602 0.532 0.812
PERS 0.751 0.785 0.593 0.706
PU 0.643 0.541 0.766 0.586 0.825
UI 0.802 0.759 0.502 0.710 0.533 0.842

Table 5.21. Discriminant validity - Cross-loading

VAR AT ICOM PE PERS PU UI

AT1 0.814 0.608 0.548 0.581 0.533 0.628
AT2 0.807 0.632 0.441 0.635 0.478 0.674
AT3 0.849 0.652 0.498 0.639 0.575 0.68
ICOM1 0.552 0.798 0.337 0.588 0.326 0.573
ICOM2 0.592 0.779 0.386 0.604 0.406 0.593
ICOM3 0.683 0.824 0.533 0.684 0.545 0.649
PE1 0.461 0.394 0.806 0.424 0.601 0.378
PE2 0.468 0.41 0.797 0.477 0.638 0.379
PE3 0.535 0.488 0.834 0.539 0.626 0.463
PERS1 0.401 0.568 0.275 0.648 0.253 0.446
PERS2 0.492 0.582 0.226 0.689 0.28 0.518
PERS3 0.635 0.573 0.529 0.773 0.531 0.561
PERS4 0.553 0.515 0.58 0.707 0.529 0.473
PU2 0.529 0.467 0.641 0.502 0.797 0.453
PU3 0.561 0.484 0.614 0.522 0.849 0.47
UI1 0.712 0.635 0.456 0.604 0.461 0.849
UI2 0.653 0.636 0.414 0.618 0.453 0.831
UI3 0.658 0.645 0.394 0.572 0.432 0.845
PU1 0.501 0.388 0.64 0.427 0.829 0.396

In the examination of data from a composite model for a reflective measurement model,

Henseler et al. (2016) emphasised the necessity of impartiality in the application of structural

equation modeling techniques. The assessment of model fit indices was employed to gauge the

fidelity of the measurement model in representing the data. Consistent with prior studies, this

research computed the model fit indices, as suggested by Ramayah et al. (2017). Table 5.22

presents the outcomes of the model fit.

Table 5.22. Model Fit

Saturated model Estimated model
SRMR 0.091 0.1

117
 d_ULS 1.574 1.916
d_G 0.567 0.6
Chi-square 1393.126 1413.067
NFI 0.675 0.67

For latent variables correlation, Table 5.23 displays each construct's value of one with its self-

variables. The results indicated an acceptable level of the structural model fit, as shown in Table

5.22, signifying good factor loadings of all items into latent constructs and a well-fitted model.

Furthermore, the correlation between variables is presented via the correlation plots, as shown

in Figure 5.5.

Table 5.23. Latent Variables Correlation

AT ICOM PE PERS PU UI
AT 1 0.766 0.602 0.751 0.643 0.802
ICOM 0.766 1 0.532 0.785 0.541 0.759
PE 0.602 0.532 1 0.593 0.766 0.502
PERS 0.751 0.785 0.593 1 0.586 0.71
PU 0.643 0.541 0.766 0.586 1 0.533
UI 0.802 0.759 0.502 0.71 0.533 1

Figure 5.5. Scatter Plots for Relationship Paths

The above scatter plots, shown in Figure 5.5, visually represent the relationships between the

variables. Each relationship path has a positive correlation, indicating that as one variable

increases, the other also tends to increase. The r-square linear values, which are more than 0.3,

118
indicate that the variation in the independent variable can explain the variation in the dependent

variable. Therefore, the strong association between the variables reinforces the significance of

the observed relationships in the study.

5.3.2 Structural Model Assessment

The structural model examines the predictions about the hypotheses and the subsequent

determination of accepting or rejecting them based on the path coefficients. Figure 5.6

illustrates the conceptual framework, presenting all constructs and hypotheses. The SEM model

was calculated to investigate the relationships between the constructs for users and non-users

as the two separate datasets. The outcomes of the structural model assessment, including the

values of the path coefficients, are presented in Figure 5.7. for users and Figure 5.8. for non-

users. Additionally, Table 5.24 provides the calculations of the path coefficients in the research

model, accompanied by their corresponding significance levels.

Figure 5.6. Research Conceptual Model

Table 5.24. Path Coefficients

β Mean STD t-value p-value

β Mean STD t-value p-value
PATH (NON (NON (NON (NON (NON
(USER) (USER) (USER) (USER) (USER)
USER) USER) USER) USER) USER)

AT -> UI 0.809 0.806 0.81 0.806 0.041 0.034 19.529 23.562 0 0

119
 ICOM -> AT 0.396 0.266 0.396 0.266 0.069 0.073 5.698 3.645 0 0

PE -> AT -0.119 0.283 -0.125 0.283 0.077 0.086 1.543 3.291 0.123 0.001

PE -> PU 0.56 0.791 0.579 0.791 0.107 0.038 5.219 20.692 0 0

PERS -> AT 0.239 0.266 0.249 0.262 0.081 0.061 2.954 4.395 0.003 0

PU -> AT 0.427 0.129 0.424 0.132 0.089 0.057 4.778 2.255 0 0.024

The results indicate from Table 5.24 that usefulness (β = 0.427, p < 0.05) for non-users and (β
= 0.129, p < 0.05) for users influences the attitude towards using AI assistants for both non-
users and users. The results mention that attitude positively influences (β =0.809, P < 0.05) for
non-users and (β =0.806, P < 0.05) for users the behavioural intention of using AI assistants.
Therefore, the end-user must have a positive attitude or perception towards AI assistants
regarding their services to use this tool. Ease of use significantly influences users’ attitude (β
=0.283, P < 0.05). However, the results indicate that ease of use does not affect non-users’
attitudes (β =-0.119, P = 0.123). Perceived ease of use positively impacts perceived usefulness
for non-users (β =0.560, P < 0.05) and for users (β =0.791, P < 0.05).

Moreover, the results indicate that the AI assistant capability, which is interactive
communication, positively affects the consumer attitude towards using AI assistants for both
non-users (β = 0.396, p < 0.05) and users (β = 0.266, p < 0.05). The factor of personalisation
significantly influences consumer attitude towards AI assistants for non-users (β = 0.239, p <
0.05) and for users (β = 0.266, p < 0.05). Therefore, H1, H3, H4, H5, and H6 are supported for
users and non-users. H2 is just supported for users.

Figure 5.7. Tested Model for Users

120
 Figure 5.8. Tested Model for Non-Users

5.3.3 Multigroup Analysis

The multigroup analysis was conducted to test H7 and assess any differences between Western

and Eastern cultures regarding the relationships among the primary factors influencing the

intention to use AI assistants in e-commerce. Table 5.26 presents the results of the multigroup

analysis, comparing path differences between the two groups. For all paths, the differences are

less than 0.2, and p-values are greater than 0.5, indicating no statistical significance at the 0.05

level. Overall, the multigroup analysis suggests no significant differences in the relationships

between constructs across the Western and Eastern groups, leading to the non-support of H7.

Table 5.25. Multigroup Analysis

Difference (Western - 1-tailed (Western vs. Eastern) 2-tailed (Western vs. Eastern)
PATH
Eastern) p-value p-value

AT -> UI -0.034 0.761 0.479

ICOM -> AT -0.212 0.905 0.189

PE -> AT -0.035 0.6 0.8

PE -> PU -0.082 0.888 0.223

PERS -> AT 0.168 0.13 0.26

PU -> AT 0.146 0.142 0.283

121
The results indicate that the impact of usefulness and ease of use on attitudes toward using AI

assistants is relatively equal in Western and Eastern cultures. Furthermore, the results show that

the effect of AI capabilities on attitudes toward using AI assistants is also relatively equal in

Western and Eastern cultures, and the impact of attitude on usage intention is likewise similar

in both cultures. Therefore, there is a similarity between Western and Eastern cultures regarding

their attitudes and adoption of AI assistants, which aligns with the concept of global consumer

theory toward emerging technologies aiming to foster a sense of inclusion for global consumers

and enhance various usages (Hernani-Merino et al., 2020).

5.4 Hypothesis Testing

As elaborated in the preceding sections, subsequent to evaluating the structural model,

confirming its suitability, and examining the path coefficient, the subsequent sections delve

into the path model concerning latent constructs. This is done to assess the hypotheses and their

corresponding research questions in the present study. The results of hypothesis testing are

presented in Table 5.26. The main research question of this study is: What key factors influence

individuals' usage intention of AI assistants in e-commerce among users and non-users? The

following sections address the research question by exploring each sub-question and

hypothesis. The first sub-question is: How does perceived usefulness affect the attitudes of

users and non-users towards using AI assistants in e-commerce? The following hypothesis is

explored in this question as follow:

H1. Perceived usefulness positively influences the attitudes of users and non-users towards

using AI assistants in e-commerce.

Regarding hypothesis H1, the research models in Figure 5.7 and Figure 5.8 show that perceived

usefulness directly affects attitudes towards using AI assistants. The results show that

hypothesis H1 is accepted. The second sub-question, how does perceived ease of use affect the

122
attitudes of users and non-users towards using AI assistants in e-commerce? The following

hypothesis addresses this question as follow:

H2. Perceived ease of use positively influences the attitudes of users and non-users towards

using AI assistants in e-commerce.

Figure 5.7 and Figure 5.8 show that the relationship between the factor of the perceived ease

of use and attitude of using AI assistants is not significant for non-users but is significant for

users with path coefficients of 0.283. Accordingly, H2 is accepted for users. The third sub-

question, How does perceived ease of use affect the perceived usefulness of users and non-

users of AI assistants in e-commerce? This question is addressed in the following hypothesis:

H3. Perceived ease of use positively influences perceived usefulness among users and non-

users of AI assistants.

Table 5.24 shows that the perceived ease of use positively affects perceived usefulness and the

path significance for users and non-users. H3 is supported. The fourth sub-question: How does

interactive communication affect the attitudes of users and non-users towards using AI

assistants in e-commerce? The following hypothesis is explored in this question as follow:

H4. Interactive communication positively influences the attitudes of users and non-users

towards using AI assistants in e-commerce.

Table 5.24 shows the path significance of 0.396 for non-users and 0.266 for users. Therefore,

hypothesis H4 is accepted. The fifth sub-question: How does personalisation affect the attitudes

of users and non-users towards using AI assistants in e-commerce? It is related to the following

hypothesis.

attitudesH5. Personalisation positively influences the attitudes of users and non-users towards

using AI assistants in e-commerce.

123
Table 5.24 shows that the effect of personalisation on attitudes towards using AI assistants is

significant for non-users and users. Accordingly, H5 is accepted. The sixth sub-question: How

does attitude affect the intentions of users and non-users towards using AI assistants in e-

commerce? This question is addressed in the following hypothesis:

H6. The attitudes will positively influence the usage intentions of users and non-users toward

using AI assistants in e-commerce.

The second research question of this study: Do significant differences exist in the relations

among the primary factors influencing the intention to utilise AI assistants in e-commerce when

comparing Western and Eastern cultures? This question is addressed in the following

hypothesis:

H7. The relationships among the primary factors influencing the intention to use AI assistants

in e-commerce differ significantly between Western and Eastern cultures.

Based on the data from Table 5.24 and Table 5.25, H6 is supported, while H7 is not. The

findings indicate no significant differences between the two consumer groups. Table 5.25

reveals comparable patterns in the usage of AI assistants among Western and Eastern

consumers, supporting this conclusion. The results suggest that regional or cultural variations

between Western and Eastern consumers have a minimal impact on factors influencing their

intentions and behaviour towards AI assistants in e-commerce.

Table 5.26. Hypothesis Testing

Hypotheses Results
H1. Perceived usefulness positively influences the attitudes of users and Supported
non-users towards using AI assistants in e-commerce.
H2. Perceived ease of use positively influences the attitudes of users and Supported for users
non-users towards using AI assistants in e-commerce. Not supported for non-users
H3. Perceived ease of use positively influences perceived usefulness Supported
among users and non-users of AI assistants.
H4. Interactive communication positively influences the attitudes of users Supported
and non-users towards using AI assistants in e-commerce.

124
 H5. Personalisation positively influences the attitudes of users and non- Supported
users towards using AI assistants in e-commerce.
H6. The attitudes will positively influence the usage intentions of users and Supported
non-users toward using AI assistants in e-commerce.
H7. The relationships among the primary factors influencing the intention Not supported
to use AI assistants in e-commerce differ significantly between Western
and Eastern cultures.

5.5 Chapter Summary

This chapter describes quantitative data analysis, descriptive statistics, and evaluation of the

research framework to answer research questions and test hypotheses. The analysis of

respondents' profiles confirms the suitability of the study sample. Moreover, descriptive data

analysis reveals comprehensive insights about the collected data, showing a left-skewed

distribution without extreme outliers. The measurement scale assessment demonstrates high

construct reliability based on Cronbach's alpha values for all constructs, while scale validity

and model fit tests confirm the validity of the measurement scale. Convergent and discriminant

validity are established for the measurement scale and structural model assessments, ensuring

an acceptable level of model fit for all constructs in the study.

Subsequently, the outcomes of the structural model assessment for this study are presented.

The later sections of this chapter display the path coefficients obtained via the partial least

squares structural equation modelling (PLS-SEM) approach and multigroup analysis. The path

coefficients test supports hypotheses H1, H3, H4, H5, and H6, but not H2 and H7. Additionally,

path coefficients of indirect effects were examined to test the research model between

predictors and outcome variables. To support the results of the quantitative method and provide

more insights into the research question, machine learning and natural language processing

techniques were employed. In the upcoming chapter, the analysis of data related to the machine

learning component in this investigation will be addressed.

125
Chapter 6. Qualitative Analysis and Results

6.1 Introduction

This chapter delves into the machine learning approach adopted in this study. Phase B involves

conducting an analysis of online reviews using machine learning techniques to achieve two

primary objectives: 1) analyse consumer reviews related to the online shopping experience and

enhance the AI assistant usage experience for e-commerce, and 2) provide supportive evidence

for the research questions and gain deeper insights into consumer attitudes and experiences.

This phase employs three critical techniques for collecting and analysing online reviews:

natural language processing (NLP), topic modelling, and thematic analysis. The analysis

focuses on the attributes of the pre-defined main factors of the research model resulting from

Phase A (quantitative study) of this research. As a result, the machine learning analysis aims to

corroborate the quantitative results from the survey outlined in the previous chapter.

This chapter analyses publicly available online reviews of e-commerce channels used by the

Louis Vuitton (LV) fashion brand. The rationale behind selecting Louis Vuitton (LV) as a case

study for this phase is twofold: 1) LV is a top luxury brand in the fashion industry, and its

consumers have high expectations of the online experience, and 2) LV has started integrating

AI assistants into their shopping applications in some geographical locations. The data was

collected from three publicly available technology review platforms: Trustpilot, the Apple

Store app, and the Google Play Store.

This chapter begins by describing the review collection's aim, followed by data pre-processing

and analysis procedures. Then, the findings of the NLP analysis and the relevant statements

conducted by the thematic analysis are interpreted. Therefore, the chapter is structured as

follows. Section 6.1 provides the introduction. Section 6.2 represents an overview of the

collected online reviews. Sub-section 6.2.1 presents the outcomes of the data pre-processing.

126
As the second step before conducting the NLP analysis, sub-section 6.2.2 provides the data

annotation and vectorisation stages and their results. Section 6.3 discusses the outcomes of the

NLP analysis. Sub-section 6.3.1 demonstrates the effects of the keyword extraction from the

collected online reviews. After that, sub-section 6.3.2 presents the LDA model. Section 6.4

presents the thematic analysis and integrative interpretation of the findings. Finally, Section 6.5

summarises and concludes the chapter.

6.2 Description of Online Reviews

A Python program was developed for web scraping. The Python scraper was explicitly

designed to gather online data and implemented using Jupyter Notebook, installed through

Anaconda 3 on a Windows operating system. The programmed scraper extracted online

reviews and successfully collected a total of 997 reviews. The data collection process occurred

in March 2022. These reviews were then compiled and organised in an Excel spreadsheet for

further analysis and processing. The collected data primarily consisted of text and included

four main categories: date, rank, review, and descriptions.

6.2.1 Data Pre-processing of Online Reviews

The data pre-processing involves specific steps to clean and prepare the collected reviews for

the next stage: data analysis and interpretation. The first task of data pre-processing was data

normalisation, which included converting all the collected reviews to standard text style.

Therefore, all the text of the collected reviews was converted to lowercase, as shown in Table

6.1, which presents the results of the normalisation step. The second step of the data pre-

processing stage is removing emoticons, punctuation marks, and digits. Moreover, the pre-

processing data stage included a step of cleaning the collected text from multi-spaces and non-

English languages. Remove stop words and spaces methods were conducted as the final data

pre-processing step. Table 6.1 presents an example of the results of collected reviews after all

127
phases of the data pre-processing stage. The following section discusses the consequences of

data annotation and vectorisation methods.

Table 6.1. Data Pre-processing

6.2.2 Data Annotation and Vectorisation

After the data pre-processing and cleaning stage, data annotation was applied for the polarity

mapping of each row in all collected reviews. Table 6.2 presents an example of the results of

data annotating the collected reviews. The collected data was then converted into two sets bas

ed on the rating given to each review, categorising them as either positive or negative sentime

nt polarity. The vectorisation function was implemented to perform array operations on the co

llected reviews. The total number of positive reviews is 423, and the number of negative revi

ews is 523, as shown in Figure 6.1.

Figure 6.1. Positive and Negative Reviews

128
 Table 6.2. Data Annotation and Vectorization

6.3 Natural Language Processing Analysis

This section presents the results of the natural language processing (NLP) analysis method

applied to the collected online reviews. Researchers commonly use the NLP technique to

comprehend natural language texts' structure, meaning, and sentiments, such as articles, tweets,

and reviews (Grover et al., 2019). Through the NLP analysis, valuable insights were extracted

regarding customers' attitudes, opinions, and experiences. Moreover, various linguistic aspects,

including sentiment polarity, key phrases, and topics discussed in the reviews, were examined,

enabling a comprehensive understanding of the overall sentiment distribution among the

collected reviews and identifying specific aspects supporting the previous study's outcomes.

6.3.1 Keywords Extraction

The NLP analysis method involves a crucial step known as keyword extraction, which aims to

identify the most important words or phrases in the given dataset. One widely used technique

for this purpose is Term Frequency-Inverse Document Frequency (TF-IDF). TF-IDF is a

statistical measure that evaluates the importance of a word to a document in a collection of

documents. By assigning a weight to each word based on its frequency and the number of

129
documents it appears in, TF-IDF can determine the significance of each word in the dataset.

The higher the TF-IDF score, the more important the word is in the document.

The TF-IDF approach identifies the most critical comments in the dataset by considering the

word's frequency and rarity across the entire dataset. Therefore, the TF-IDF method was

applied for keyword extraction, effectively identifying essential words in the collected online

reviews. As shown in Figure 6.1, this method reveals frequently used keywords in both positive

and negative reviews, providing valuable insights into customer sentiments and experiences.

Figure 6.2. Frequently Used Keywords in Positive and Negative Reviews

The positive online reviews contain extracted keywords such as "great," "best," "love,"

"awesome," "fantastic," and "amazing," signifying the reasons for their positive classification.

Conversely, the negative reviews include words such as "poor," "worst," and "never," which

occurred frequently, indicating their negative sentiment. After the initial keyword extraction,

the subsequent analysis focused on identifying the top 25 most frequently occurring three-word

phrases in both positive and negative reviews, as presented in Table 6.3.

130
 Table 6.3. The Top 25 of Three-Word Phrase Extraction

Three-word combination Three-word combination

TF-IDF TF-IDF
of positive reviews of negative reviews

Among the identified phrases, "super-easy use," "easy navigate the app," and "useful awesome

loving" were indicative of positive reviews. On the other hand, terms like "poor customer

service," "horrible customer service," and "never order online" were prevalent in negative

reviews, clearly expressing customers' dissatisfaction with the corresponding service. Overall,

identifying frequently occurring three-word phrases in positive and negative reviews provides

deeper insights into the expressed sentiments and specific experiences that led to these

evaluations.

6.3.2 Topic Modelling

Various machine-learning techniques can be employed to analyse text datasets and extract

valuable insights. One such technique is Latent Dirichlet Allocation (LDA) topic modelling,

widely used in (NLP) applications to identify hidden topics in a text corpus. In the current study

context, the LDA topic modelling was utilised to analyse a dataset of online reviews and

identify their most relevant topics. The identified cases provide a comprehensive understanding

of the customers' preferences, expectations, and concerns, enabling businesses to improve their

service and meet customer needs. Additionally, the LDA topic modelling analysis reveals the

probability distribution of each topic in the positive and negative reviews. The LDA model was

then applied separately to each group to identify the main themes in the online reviews. The

131
LDA model analysis identified twenty topics within the reviews. Each topic includes a set of

ten words, specifically for positive and negative online reviews. These identified topics provide

valuable insights into the prevalent themes and concepts within the collected reviews, shedding

light on the aspects that customers frequently emphasise.

6.4. Thematic Analysis and Integrative Findings

Thematic analysis is a method used to identify patterns in qualitative data, such as customer

reviews. In the current study, thematic analysis was utilised to analyse the themes and topics

expressed in customer reviews about their online shopping experience. The objective was to

identify improvement areas based on the analysis's findings. Table 6.4 presents the thematic

analysis, revealing themes and topics in both positive and negative application reviews. By

examining these themes, developers could gain insights into the areas where customers were

satisfied or dissatisfied, using this information to enhance the AI assistants for online shopping.

Table 6.4. Thematic Analysis

The thematic analysis findings were utilised to explore all the primary constructs resulting from

the survey analysis. The results indicate that usefulness, ease of use, personalisation, and

132
interactive communication positively influence consumer attitudes. These findings align with

the evidence from the thematic analysis, providing complementary insights. Furthermore, the

thematic analysis offered additional insights, enriching the interpretation of the quantitative

results. For instance, some customers expressed concerns about the security and privacy of

their personal information when using the application, highlighting potential areas for

improvement. Developers could use this feedback to enhance AI assistants' security and

privacy features and protect customers' data. Table 6.5 presents the integrated analysis

interpretation of the mixed method.

Table 6.5. Integration Analysis

Qualitative Results (in all

Nature of Integration
Dimensions Quantitative Results dimensions captured in
/Complementary
predefined themes)
Usefulness Consumers who perceive AI Consumers appreciate the The integration of the quantitative
assistants as more useful helpful images, updated and qualitative findings suggests a
tend to have more information, and fast complementary. The quantitative
favourable attitudes towards responses to a conversation data indicates that perceived
using them in e-commerce provided during their online usefulness positively impacts
applications. shopping experience. attitudes, while the qualitative data
reinforces this finding by
highlighting the specific aspects of
helpfulness aspects based on
consumers’ experience.
Ease of use Consumers who perceive AI Online shoppers value The quantitative data showed that the
assistants as easy to use tend product searching, visual factor of ease of use is a positive
to have positive attitudes displaying, recommendation, effector towards the consumers’
towards using them. navigation, accessibility, and attitudes, and qualitative data
interactions during online supports this finding by emphasising
shopping. the key elements of ease of use based
on consumers' experience.
Personalisation If AI assistants can offer Consumers express The quantitative data showed that
personalised experiences for satisfaction with personalised personalisation positively affects the
consumers, their attitude products and suggestions consumers’ attitudes. In contrast,
towards AI assistant use is based on their needs. qualitative data supports this finding
likely to be significantly by highlighting how consumers
more positive. value receiving services or products
customised to their specific needs.
Communication AI assistants' interactive Consumers appreciate the The quantitative data showed that
communication capability experience of classy, pleasant, interactive communication
significantly impacts and efficient communication significantly influences consumers’
consumers’ attitudes towards during online conversations. attitudes, and qualitative data
AI assistant use. supports this finding by identifying
how the style of conversations leads
to customers’ satisfaction.

The reviews highlighted several key themes related to customer service, app functionality, user-

friendliness, personalisation, product quality, and interactive communication and delivery.

133
Positive reviews praised helpful and responsive customer service representatives, easy-to-use

app functionality, user-friendliness, and high product quality. In contrast, negative reviews

criticised slow response times, unhelpful customer service, bugs, crashes, glitches in the app,

difficult navigation, poor product quality, and slow and unreliable delivery. These insights

provide valuable feedback for developers to improve AI assistants' quality, stability, user

experience, product offerings, and delivery services to effectively meet customer needs and

preferences.

Table 6.5 presents the integration analysis of quantitative and qualitative results for different

dimensions of AI assistant usage. The findings show that perceived usefulness positively

impacts consumers' attitudes, with qualitative data emphasising helpful images, updated

information, and fast responses during online shopping. Similarly, ease of use positively affects

attitudes, with qualitative data highlighting key elements such as product searching, visual

displaying, and interactions. Personalisation also influences attitudes positively, with

consumers valuing customised products and suggestions. Interactive communication

significantly impacts attitudes, as consumers appreciate pleasant and efficient conversations.

Integrating quantitative and qualitative findings indicates a complementary relationship,

strengthening the overall understanding of consumers' attitudes towards AI assistant use in e-

commerce applications.

6.5 Chapter Summary

This chapter presents a comprehensive analysis of customer reviews from fashion e-commerce

platforms using natural language processing. It involves extracting keywords to identify

frequently occurring words in positive and negative reviews. Additionally, the study identifies

the most frequently occurring three-word phrases in both positive and negative reviews.

Furthermore, applying LDA topic modelling and thematic analysis provides deeper insights

into specific customer experiences, enhancing the interpretation of quantitative results.

134
Chapter 7. Discussion, Implications and Conclusion

This chapter presents discussions of the findings and recommendations of the study. Section

7.1 revises the research aim, objectives, questions, and hypotheses. Section 7.2 discusses the

analysis of the study's findings, and Section 7.3 provides the critical contributions of the

theoretical and practical implications of the study. Finally, Section 7.4 identifies the study's

limitations and provides recommendations for future research.

7.1 Research Aim and Questions

As AI continues to transform the online shopping landscape, understanding what drives

individuals' decisions to engage with AI assistants is crucial for businesses and customer

satisfaction. Recently, AI-powered assistants have gained popularity in e-commerce, offering

personalised and efficient customer experiences. However, despite their growing prominence,

there are limited studies on AI assistant adoption in e-commerce and a lack of empirical

research focusing on both users and non-users as potential users of AI assistants. Previous

studies have also provided contradictory evidence regarding some factors influencing the

intention to use AI assistants, considering a specific cultural context. With these considerations,

the primary objective of this study was to explore the fundamental factors influencing

individuals' attitudes and usage intentions toward AI assistants in e-commerce, encompassing

both users and non-users. Additionally, this study aimed to recognise cultural differences to

provide valuable insights into diverse consumer preferences. The following main research

question with its sub-questions have guided the process:

Main Research Question: What factors influence individuals' intention to use AI assistants in

e-commerce among users and non-users, and are there significant differences in these factors

impact when comparing Western and Eastern cultures?

135
To address this question, a literature review on adopting AI assistants in e-commerce was

conducted to explore recent studies on this topic. This review led to the development of the

research model along with sub-questions. Figure 7.1 presents this study’s research model and

the total number of hypotheses.

Figure 7.1. Research Structural Model

The sub-questions for the main research question, derived from the research model, were:

RQ1.1: How does perceived usefulness affect the attitudes of users and non-users towards

using AI assistants in e-commerce?

RQ1.2: How does perceived ease of use affect the attitudes of users and non-users towards

using AI assistants in e-commerce?

RQ1.3: How does perceived ease of use affect the perceived usefulness of users and non-

users of AI assistants in e-commerce?

RQ1.4: How does interactive communication affect the attitudes of users and non-users

towards using AI assistants in e-commerce?

136
 RQ1.5: How does personalisation affect the attitudes of users and non-users towards using

AI assistants in e-commerce?

RQ1.6: How does attitude affect the intentions of users and non-users towards using AI

assistants in e-commerce?

RQ1.7: Do significant differences exist in the relations among the primary factors

influencing the intention to utilise AI assistants in e-commerce when comparing Western

and Eastern cultures?

7.2 Findings

The present research findings were obtained from a mixed-method approach using quantitative

and qualitative studies. The total number of collected data for both studies is 1,397. The

discussion of the findings is presented in the following subsections, arranged around each

hypothesis and its respective results. The linkage between the research sub-questions and the

hypotheses is as follows:

RQ1.1: How does perceived usefulness affect the attitudes of users and non-users towards

using AI assistants in e-commerce?

 H1. Perceived usefulness positively influences the attitudes of users and non-users

towards using AI assistants in e-commerce.

RQ1.2: How does perceived ease of use affect the attitudes of users and non-users towards

using AI assistants in e-commerce?

 H2. Perceived ease of use positively influences the attitudes of users and non-users

towards using AI assistants in e-commerce.

RQ1.3: How does perceived ease of use affect the perceived usefulness of users and non-users

of AI assistants in e-commerce?

137
  H3. Perceived ease of use positively influences perceived usefulness among users and

non-users of AI assistants.

RQ1.4: How does interactive communication affect the attitudes of users and non-users

towards using AI assistants in e-commerce?

 H4. Interactive communication positively influences the attitudes of users and non-

users towards using AI assistants in e-commerce.

RQ1.5: How does personalisation affect the attitudes of users and non-users towards using AI

assistants in e-commerce?

 H5. Personalisation positively influences the attitudes of users and non-users towards

using AI assistants in e-commerce.

RQ1.6: How does attitude affect the intentions of users and non-users towards using AI

assistants in e-commerce?

 H6. The attitudes will positively influence the usage intentions of users and non-users

toward using AI assistants in e-commerce.

RQ1.7: Do significant differences exist in the relations among the primary factors influencing

the intention to utilise AI assistants in e-commerce when comparing Western and Eastern

cultures?

 H7. The relationships among the primary factors influencing the intention to use AI

assistants in e-commerce differ significantly between Western and Eastern cultures.

138
7.2.1 Usefulness

The path "PU -> AT" exhibits differing coefficients between user groups (β=0.12), with a

stronger impact on non-users (β=0.42), as reflected by the higher t-value for users (2.25) and

non-users (4.77) and significant p-value (users=0.02 and non-users=0.00). Thus, the hypothesis

that usefulness positively influences users and non-users towards using AI was supported.

Meanwhile, the results of another investigation underscore the noteworthy effects of factors

such as the perceived usefulness of a mobile app on users' intentions to utilise the mobile

application and their overarching attitude towards it (Hasan et al., 2021). In addition, previous

empirical research has established the impact of perceived usefulness on consumer attitudes,

as exemplified by its influence on attitudes towards the adoption of social media (Alduaij,

2019). Furthermore, another study's outcomes affirm each hypothesised relationship's

favourable and statistically significant results. In contrast, the variable of perceived usefulness

stands out as a particularly robust predictor, exerting a significant influence on individuals'

overall experiential outcomes when using chatbots (Lubbe & Ngoma, 2021). Meanwhile, a

mediating role played by the perceived usefulness of AI was also highlighted, indicating that

functional AI, as opposed to socially oriented AI, engenders a heightened sense of perceived

usefulness. This, in turn, cultivates more favourable attitudes towards AI and a heightened

perception of AI's realism. These outcomes collectively hold significant ramifications for

human-AI communication and human-machine interaction research, accentuating the

importance of understanding the differential impacts of AI types on users' perceptions and

attitudes (Kim et al., 2021).

For users, the perception that AI offers tangible benefits, aids in tasks, or enhances their overall

experiences leads to higher satisfaction, efficiency, and effectiveness in their interactions.

Similarly, the perception that AI can be functional, relevant, and practical for non-users

encourages a more favourable disposition towards AI adoption. By recognising AI's potential

139
value and advantages, non-users are more likely to overcome hesitations from unfamiliarity or

scepticism, thus fostering a more positive attitude towards AI. In both cases, the perceived

usefulness of AI serves as a catalyst, prompting users and non-users alike to perceive AI as a

valuable tool that augments capabilities and contributes positively to their experiences, thereby

fostering a receptive stance towards its adoption.

Zarouali et al. (2018) description of perceived usefulness shows that it is inherently linked to

ease of use. In this context, the perceived usefulness alludes to the perceptions consumers have

concerned productivity or task performance when using AI assistants in e-commerce. Premised

on the TAM model, the perceptions concerning ease of use tend to vary among people (Lee &

Lin, 2023; Shin, 2010). Cultural factors can be a significant influence on such perceptions. This

is aligned with the conceptualisations emerging from SET and the Theory of Reasoned Action

that social interactions, subjective norms, and influences from the social environment influence

customer satisfaction and attitudes when using AI assistants (Jiang et al., 2022; Roh et al.,

2023). The findings in the current study align with the conceptual framework by demonstrating

that positive attitudes emerging from the perceived usefulness increases the intentions to use

AI assistants in e-commerce for customers from different cultural background. Therefore,

assessing the perceived usefulness in the current study fosters an understanding of how the

effectiveness of AI assistants in meeting customer needs and expectations transcends the

cultural aspects. Also, the findings on the perceived usefulness are inherent in establishing the

value customers ascribe to AI assistants during their shopping experience. Suggestively,

positive attitudes associated with perceived usefulness imply that AI assistants are a valuable

aspect of the shopping experience.

140
7.2.2 Ease of Use

The path "PE -> AT" displays differing coefficients between non-users and users, indicating a

negative influence of Perceived ease of use on Attitude for non-users (β=-0.11) and a positive

influence for users (β=0.28), with the latter exhibiting higher t-values (users=20.69 and non-

users=5.21) and significant p-value for users=0.001 group while non-significant value for non-

users=0.12. Thus, the hypothesis was that with the favourable disposition towards ease of use,

users were positively impacted by the perceived level of simplicity in its operation, while non-

users found it challenging. The possible explanation for this can be that the users respond

positively when they perceive that interacting with the technology is straightforward and

uncomplicated. This positive response arises from their familiarity and experience with the

system, enabling them to navigate and utilise it confidently. On the other hand, non-users

lacking prior exposure to the technology may find its operation more challenging due to their

unfamiliarity. The unfamiliarity can lead to uncertainty and hesitancy in their interactions, thus

impacting their perception of ease of use. Overall, users' positive experiences with the

technology's simplicity contrast with non-users’ challenges stemming from their lack of

familiarity, underscoring the role of prior experience in shaping perceptions of ease of use.

These findings are consistent with similar studies that found a positive relationship between

usefulness perception and attitude toward using AI (Hasan et al., 2021; Ashfaq et al., 2020;

Zarouali et al., 2018). Similarly, another study investigated consumers' opinions and

purchasing intentions toward AI technology (Yin & Qiu, 2021). The study included 313

respondents (61% female) between 18 and 65 from the top ten metropolitan regions in the US.

According to the findings, perceived usefulness, simplicity of use, and performance risk all

affected customers' attitudes toward AI—positive opinions about the technology involved

buying intentions favourably (Liang et al., 2020). Furthermore, antecedent scholarly research

141
has substantiated the impact of perceived usefulness on consumer dispositions, exemplified by

its effect on attitudes towards the adoption of social media, as evidenced by Alduaij (2019).

Furthermore, the outcomes of another study affirm the positive and statistically significant

consequences of each hypothesised relationship. In contradistinction, perceived usefulness

assumes paramount importance, emerging as the preeminent predictor exerting a substantial

influence on individuals' comprehensive experiential outcomes in the context of chatbot

utilisation, as elucidated in the investigation conducted by Lubbe and Ngoma (2021).

Consumers' perceptions about the usefulness of AI platform usage significantly determine their

attitude towards AI-based e-commerce platforms. However, regardless of apparent utility or

simplicity of use, the requirement for human contact cannot be replaced by AI in particular

cultural settings (Kelly et al., 2023). In addition, another study illuminates the significant role

of ease of use in influencing consumers' disposition to trust and embrace AI technology within

online retail operations (Nagy & Hadjú, 2021). By comprehending the determinants (ease of

use) underlying customer utilisation and impediments, organisational leaders and developers

can formulate and implement these pioneering technologies more effectively, augmenting the

customer experience and elevating organisational performance (Jan et al., 2023). In contrast,

empirical findings proposed that perceived ease of use does not maintain a statistically

significant correlation with the dependent variable (Tan & Lim, 2023).

Tahar et al. (2020) described the perceived ease of use as how individuals subjectively assess

learning and using a particular technology. Zarouali et al. (2018) link it to perceived usefulness

as the practical and effortless capacity to enhance productivity or task performance when using

AI shopping assistants. This relationship between perceived ease of use and usefulness is

demonstrated in the conceptual framework as a combination of factors that influence the

customers' attitudes. According to Teo et al. (2009), the TAM model demonstrates that positive

beliefs can lead to increased intentions to adopt and continue using AI assistants because of the

142
positive attitude users have towards the technology. The insights emerging from the findings

align with the conceptual and TAM theoretical framework, considering that the beliefs and

attitudes emerging among the consumers who engage with AI assistants shape their

behavioural intentions. If an individual believes that the AI assistants are complex, there is

limited intention implying that they have a negative attitude towards the technology. The

findings on perceived ease of use help to differentiate how attitudes that emerge among users

and non-users develop and how they differ based on experience in the use of AI assistants.

Acikgoz and Vega (2022) note that the TAM model is appropriate in investigating novel

technologies, which makes findings on the perceived ease of use in the context of users and

non-users in a multicultural context significant in informing how perceived usefulness impacts

adoption and intentions to continue using the AI assistants in e-commerce.

7.2.3 Ease of Use Positively Influences Usefulness

The ease of use to usefulness pathway displays differing coefficients of 0.56 for non-users and

a notably higher 0.791 for users, both associated with highly significant t-values of 5.219 and

20.692, respectively. Thus, the hypothesis that perceived ease of use positively influences

perceived usefulness among users and non-users of AI assistants was well supported by both

groups. Hence, the positive influence of ease of use on perceived usefulness is rooted in the

cognitive benefits of reduced complexity, increased user proficiency, heightened efficiency,

and enhanced overall user experience. As users find the AI technology easy to navigate and

integrate into their tasks, their perceptions of its usefulness are positively reinforced, fostering

a more favourable disposition towards its utilisation.

Theoretically, a user's and nonuser's acceptance of a particular technology is influenced by two

noteworthy variables: the perception of AI usefulness and ease of use, notwithstanding the

assertion that the latter exerts a comparatively lesser impact, as Cha (2010) indicates. However,

143
in both (Users and Non-users) cases, the positive relationship between perceived ease of use

and usefulness can be attributed to the cognitive alignment between perceived ease of use and

the perceived usefulness of AI assistance. An interface that is easy to navigate and understand

aligns with mental models of the users and non-users, reducing perceived complexity and

increasing the likelihood of successful task completion. This alignment bolsters the perception

that the technology is effective and advantageous, consequently fostering positive attitudes and

increasing AI assistants' perceived usefulness among users and non-users. Furthermore, similar

results were observed in a study administered an online survey among a sample of 400

individuals who are residents of Jakarta and employ smartphone voice assistants. Employing

SmartPLS to assess validity, reliability, and the study's hypotheses, the outcomes substantiate

the validity of all six research hypotheses. The findings affirm the constructive effect of

perceived ease of use on perceived usefulness, signifying a statistically significant association

between the two constructs (Oktavia et al., 2023). Moreover, Noreen et al. (2023) demonstrated

a noteworthy and affirmative correlation between the intention to embrace AI and critical

factors encompassing perceived usefulness and ease of use of AI technology.

Zarouali et al. (2018) describe the perceived usefulness and ease of use as factors that determine

the practical and effortless use of AI shopping assistants that lead to increased productivity or

task performance. Based on the conceptual model adopted in the current study, the influence

that perceived usefulness and ease of use have on customers is parallel and complementary.

Therefore, the positive attitudes associated with the perceived usefulness are inherently linked

to those of the perceived ease of use. These sentiments are aligned with the TAM model, which

employs perceived usefulness and ease of use as the primary factors in the assessment of users'

attitudes and intentions to use a particular technology (Davis, 1989). The impact of perceived

ease of use on the perceived usefulness of AI assistants highlights the influence cognitive

abilities have on the attitude and intentions to use the technology. The findings show that

144
customers with adequate access to information and knowledge can effectively learn and use AI

assistants. Developing relevant technological skills leads to positive perceptions concerning

the ease of use, which subsequently creates a positive attitude and intention to continue using

because individuals feel that the technology is useful. These findings show that users have

experience and knowledge, which makes them more likely to develop a positive attitude and

intention to use AI assistants compared to non-users. In this context, the findings are important

in highlighting why the deployment of AI assistants should be coupled with adequate sharing

of information that can enable all users to develop the relevant skills to realise the benefits

associated with the technology in their shopping experience.

7.2.4 Interactive Communication

The "ICOM -> AT" path indicates positive coefficients for non-users (β=0.39) and users

(β=0.26). However, the t-values are relatively higher for non-users (5.69) than users (3.64) with

a significant p-value (0.00), implying a stronger effect of Interactive Communication on

Attitude for non-users. The results indicate that interactive communication positively affects

consumer attitudes towards AI assistants. Questions can arise about why interactive

communication has a positive effect, and the possible explanation can be the engagement,

personalisation, perceived responsiveness, and integration aspects of interactive AI interactions

collectively foster favourable attitudes by resonating with innate human psychological

inclinations and cognitive processes.

This study explicitly assessed interactive communication as a critical characteristic of AI

assistants. This finding indicates that well-designed AI assistants’ communication styles and

customised features significantly influence consumers' attitudes towards AI-based platforms.

The statistical analysis strongly supports a substantial relationship between AI assistant

characteristics and consumers' attitudes, thus confirming Hypothesis 4. These findings

145
emphasise the profound impact of well-designed conversational AI interfaces on consumers'

attitudes towards AI platforms. They underscore the importance of delivering services that

align with consumers' expectations, fostering positive attitudes and enhancing consumers'

intention to utilise AI e-service platforms (Elsholz et al., 2019). While prior research in AI

assistant design has raised concerns about the potential adverse effects of imbuing AI assistants

with high levels of human likeness (Kim et al., 2021), scholarly discourse has consistently

emphasised the importance of endowing AI assistants with specialised expertise, particularly

within the domain of e-commerce (Liew et al., 2021). Consequently, drawing from established

principles within marketing and communication literature, the strategic integration of

interactive communication is advocated to showcase the adept proficiency of AI assistants

within the context of e-commerce. Preceding empirical investigations have substantiated that

interactive communication is a robust predictor of attitudes toward online communicators (Lee

et al., 2015). Moreover, this study found that interactive communication is crucial for e-

commerce applications. For example, a consumer mentioned that good experiences with

customer service applications include classy, pleasant, and efficient conversation. These results

are consistent with the literature (Ahmed et al., 2022; Svikhnushina et al., 2021; Yang et al.,

2019; Cheng & Jin, 2019), which highlights that analysing consumer reviews is a significant

way to improve the technology to meet consumers' needs and preferences. Interactive

communication in e-commerce AI applications is crucial for creating engaging, personalised,

and satisfying customer experiences. Facilitating effective information exchange,

customisation, and support contributes to increased customer engagement, loyalty, and,

ultimately, the success of e-commerce businesses. The strength of the result is not surprising

since AI virtual assistants have a wide range of benefits to business firms and consumers. AI

assistants enable online consumers to prompt any information related to products. Consumers

146
were able to recognise the benefits of AI assistants, and this may explain why interactive

communication is a significant determinant of attitude.

Interactive communication alludes to the AI shopping assistant's capacity to facilitate two-way

communication for a personalised experience (Chung et al. 2020; Srinivasan et al. 2002).

Effective interactive communication entails a bidirectional exchange of information in real-

time (Abeele, 2007). According to Kim et al. (2023), AI assistance is more effective when they

facilitate real-time communication with the users to ensure timely response and feedback that

can accelerate meeting the customer needs and enhance their shopping experience. These

insights are aligned with TAM and the conceptual framework because effective communication

influences the attitudes and intentions to use AI assistants by enhancing the perceived

usefulness (Ikumoro & Jawad, 2019). Furthermore, the perceived ease of use determines

successful interactive communication between the customers and the AI assistant requires.

Suggestively, individuals who find the system difficult to use are less likely to engage in

interactive communication. Insights emerging from the influence interactive communication

has on attitudes towards AI assistants facilitate an understanding of how people from different

backgrounds relate to the anthropomorphic features of AI assistants. It also highlights how

languages and other cultural aspects associated with interactive communication between

companies and customers are integrated into the AI shopping assistant experience. Such

considerations are critical in enhancing the quality of services offered by the AI assistant and

taking a personalised approach to customer satisfaction (Choi et al., 2020; Dwivedi et al.,

2022).

7.2.5 Personalisation

The "PERS -> AT" path indicates positive coefficients for non-users (β=0.23) and users

(β=0.26). While the t-values are relatively lower for non-users (2.95) than users (4.39) with a

147
significant p-value (Users=0.00 and non-users=0.003), implying a stronger effect of

personalisation on Attitude for non-users. Thus, the findings suggest that personalisation in AI

assistants favours consumers' disposition towards adopting AI assistants. Meanwhile,

incorporating personalisation attributes into AI assistants cultivates a harmonious relationship

between consumers and technology. Through tailored experiences, trust-building, enhanced

efficiency, and alignment with prevailing cultural norms, personalisation emerges as a pivotal

driver in fostering consumers' favourable disposition towards adopting AI assistants.

Contemporary advancements in electronic commerce AI platforms are distinctly centred on

providing tailored configurational alternatives, aiming to cater to distinct individual

preferences, as elucidated by Wu et al. (2020). For example, the luxury brand Gucci has

implemented a chatbot mechanism, which serves as a conduit for dispensing bespoke

promotional communications to targeted patrons who express a keen interest in personalised

merchandise, as detailed in the study by Chung et al. (2020). Furthermore, the empirical

exploration herein focuses on a specific demographic subset: female consumers engaged in

fashion retail within the United Kingdom subjected to personalised promotional overtures. The

outcomes emerging from this case study demarcate a pronounced inclination among these

participants towards seeking economic concessions for items of interest, concomitant with an

aspiration to elevate their in-store experiential encounters. However, it is discerned that these

consumers exhibit a discernible aversion towards instances of disruptive interventions and

generic promotional gestures. Concurrently, a conspicuous propensity towards exercising

autonomous control surfaces alongside endeavours to regulate the disclosure of sensitive

personal information and ameliorate the calibre of recommendations procured. Of noteworthy

significance, comprehensive analysis casts light upon inherent incongruities that permeate

customers' anticipations concerning personalised engagements, specifically necessitating the

monitoring and utilising of geographical coordinates (Canhoto et al., 2023).

148
The comprehensive analysis of consumer reviews regarding the Louis Vuitton website and

mobile applications through NLP had captivating findings. The research affirms that

usefulness, ease of use, interactive communication, and personalisation can enhance

consumers' attitudes and satisfaction. Notably, a satisfied consumer revealed the application’s

significant benefits and informative nature, and an unsatisfied consumer lamented its lack of

usefulness and confusing interface. This finding indicates that a well-designed AI assistant and

how the service is valuable and easy to use has the potential to determine consumers' attitudes

about the usage of AI assistants. These results are consistent with the literature (Hasan et al.,

2021; Ashfaq et al., 2020; Zarouali et al., 2018).

Personalisation alludes to the AI shopping assistant's capacity to tailor services with the

objective of meeting the specific needs and preferences of each customer (Chung et al., 2020;

Srinivasan et al., 2002). The personalisation of AI shopping assistants' services is critical,

considering that each consumer is unique (Hsu et al., 2021). A personalised approach entails

offering the right services at the right time and to the right person (Chandra et al., 2022). The

insights from the current study affirm those of Kaaniche et al. (2020), who demonstrate that

customers are attracted to AI solutions that foster the personalisation of services. The findings

align with the conceptual framework that personalisation contributes to the positive attitudes

users and non-users develop towards AI assistants in e-commerce. Additionally, personalisation

is tied to the TAM factors of perceived ease of use and usefulness (Kashive et al., 2018).

Personalised AI assistant services imply that individuals can customise their interactions with

the system, which enhances the ease of use. The high-quality services fostered by

personalisation reflect the perceived usefulness. These findings are important to demonstrate

how a personalised approach creates value for the customer and the e-commerce companies

using AI shopping assistants. Personalisation allows the AI assistants to develop constructive

relationships with customers from different cultural backgrounds and enhances focus on their

149
specific needs and interests, which enhances their satisfaction. Therefore, personalisation can

enhance how AI shopping assistants are used to strategically market products by focusing on

particular customer characteristics.

7.2.6 The Role of Attitude in the Intention to Use AI Assistants

The path "AT - > UI" demonstrates high coefficients for both non-users (β=0.80) and users

(β=0.80), with highly significant t-values (non-users=19.52 and users=23.56) with significant

p-value (<0.005), suggesting a strong and consistent impact of Attitude on User intension for

both groups. Thus, the results mention that consumers' attitude positively influences the

behavioural intention of using AI assistants as an e-commerce tool. This linkage is rooted in

consumer behaviour's psychological and cognitive dimensions, elucidating the intrinsic

interplay between attitude and behavioural intention in the context of AI-driven e-commerce

tools. Primarily, a positive attitude signifies consumers' favourable perceptions, beliefs, and

emotions towards AI assistants in the e-commerce milieu. Perceptions of usefulness, ease of

use, trustworthiness, and personalisation of the AI assistant shape this disposition. When

consumers view AI assistants as valuable aids that enhance shopping experiences, provide

tailored recommendations, and streamline decision-making processes, they are more inclined

to engage with these tools.

Furthermore, a positive attitude generates psychological comfort and reduces perceived risk

among consumers. As they perceive AI assistants as competent, reliable, and capable of

delivering satisfactory outcomes, apprehensions regarding technological glitches, inaccurate

suggestions, or privacy breaches are assuaged. This psychological comfort increases the

likelihood of consumers embracing AI assistants as integral components of their e-commerce

journey.

150
The results of Hypothesis H6 are consistent with previous studies (Kasilingam, 2020; Zarouali

et al., 2018). Moreover, it reveals that a positive attitude leads to accepting or using an

application or any new technology, which in this study is AI (Mullins & Cronan, 2021). Hence,

attitudes represent individuals' overall evaluation, feelings, and beliefs about a particular

technology. Consumers who develop positive attitudes toward AI assistants are likelier to

perceive them as valuable, enjoyable, and beneficial. Positive attitudes create a sense of

perceived usefulness and ease of use, which are important determinants of consumer intention

to adopt and utilise AI assistants (Jo, 2022). On the other hand, negative attitudes can create

fear or distrust, leading to a decreased intention to use AI assistants (Hornung & Smolnik,

2022). Therefore, understanding and influencing consumer attitudes through effective

communication, education, and addressing concerns play a crucial role in promoting the

adoption and acceptance of AI assistants in the consumer market. Meanwhile, when individuals

have a positive behavioural intention toward AI assistants, it suggests that they are inclined to

adopt and utilise the technology in their daily lives. This intention often translates into actual

usage behaviour, as individuals tend to follow through on their intentions and act accordingly.

Nonetheless, studying behavioural intention provides valuable insights into consumers'

adoption and utilisation patterns, aiding in developing and improving AI assistants to better

cater to users' needs and preferences. Henceforth, it is discerned that the behavioural intention

shall ascertain the subsequent utilisation conduct (Davis, 1989). This notion finds empirical

substantiation in the work of Purwanto and Loisa (2020), who observed a constructive and

statistically substantial association between mobile banking systems and the eventual

manifestation of utilisation conduct about banking systems. Further, a congruent observation

was articulated by Tandi Lwoga and Questier (2014), wherein the behavioural intent underlying

the engagement with communication systems emerged as a reliable predictor of subsequent

effective usage behaviour.

151
Moon and Kim (2001) describe attitude as the feelings experienced by an individual when

using AI shopping assistants. According to Jain and Weiten (2020), attitude is a psychological

construct that reflects the beliefs and feelings individuals relate to particular products or

services. Therefore, attitude is associated with behavioural intentions and decision-making.

Zarouali et al. (2018) describe usage intentions as the willingness to use and continue using the

AI shopping assistants. A positive attitude towards the use of AI shopping assistants emerges

from the customers' fulfilment with the services. This aligns with TAM in that the positive

attitude is an outcome of psychological conceptualisations regarding the perceived ease of use

and usefulness. As highlighted in the conceptual framework, attitude is a construct of the

cumulative impact the TAM factors, interactive communication, and personalisation have on

the customers' experience (Pei & Zhenxiang, 2006). Although TAM does not focus on cultural

influences, the inclusion of interactive communication and personalisation in the conceptual

framework affirms that cultural differences can influence the attitude and intention to adopt

and continue using AI assistants. The findings highlight the subjective nature of attitude and

intentions as psychological and cognitive constructs. Therefore, individuals from different

cultural backgrounds and experiences might express different attitudes and intentions to use AI

assistants.

7.2.7 Comparison of Western and Eastern Consumers

The results present the differences in the mean values (Western - Eastern) for various paths in

a multi-group analysis of attitudes and intentions related to AI usage. The p-values for both

one-tailed and two-tailed tests are reported to assess the statistical significance of these

differences. The findings reveal no statistically significant differences between Western and

Eastern consumers' attitudes and intentions towards using AI assistants across the examined

paths. Specifically, for paths AT -> UI, ICOM -> AT, PE -> AT, PE -> PU, PERS -> AT, and

PU -> AT, the differences in means are not substantial enough to warrant statistical

152
significance, as evidenced by the p-values exceeding conventional significance thresholds (p >

0.05). This suggests that the observed variations between Western and Eastern consumers'

responses are likely due to random fluctuations or inherent similarities between the two groups

in the context of AI assistant attitudes and intentions rather than meaningful cultural or regional

distinctions.

In Western cultural contexts, distinguished by their propensity for individualism (Kim et al.,

2022; Kim, 1994) and emphasis on autonomous decision-making, an environment conducive

to swift technology assimilation tends to prevail. This trend is driven by congruence between

adopting technological innovations and notions of self-expression and individual efficacy

(Leidner & Kayworth, 2006). Conversely, a more circumspect stance towards integrating

technology is often observed within Eastern cultural milieus, characterised by a prominent

emphasis on collectivism and the pursuit of social harmony (Triandis, 2015). In such settings,

stability and adherence to tradition hold elevated significance, manifesting as a discernible

predilection for measured and deliberate technology incorporation, as underscored by the

insights of Erez et al. (1993). Consequently, Western and Eastern consumers exhibit similarities

in accepting and adopting AI assistants. A question can arise as to why and how both cultures

have similar attitudes and intentions. There may be different reasons. Firstly, advancements in

technology, such as the widespread availability of smartphones and the Internet, have

facilitated the dissemination of information and global trends. As a result, consumers across

different cultures have been exposed to similar experiences and influences, leading to shared

beliefs and attitudes towards emerging technologies like AI assistants. Secondly, the

globalisation of markets and the interconnectedness of economies have contributed to the

diffusion of cultural values and consumer behaviours. Western and Eastern cultures have

experienced increased cultural exchange and interaction, which has led to adopting similar

beliefs and preferences, including using AI assistants. Moreover, the functionality and benefits

153
offered by AI assistants are often universal and independent of cultural differences. AI

assistants provide users convenience, efficiency, and personalised experiences regardless of

their cultural background. These shared advantages contribute to the similarity in beliefs

regarding their utilisation among Western and Eastern consumers. Furthermore, the results of

Gillespie et al. (2023) reveal a consistent trend indicating reduced trust, increased ambivalence,

and less favourable perceptions of AI in Western nations. Additionally, individuals in Singapore

exhibit relatively lower levels of trust, ambivalence, and positive attitudes towards AI, albeit

to a lesser degree (Gillespie et al., 2023). The findings in the current study show that the

primary factors lack a significant influence on customer attitudes and intentions to use AI

assistants in e-commerce. These insights contradict the sentiment expressed by Jiang et al.

(2022) based on SET that social interactions tend to influence how individuals interact with AI

shopping, considering the lack of a significant difference in the attitudes and intentions

emerging in the cross-cultural context. Nevertheless, the findings affirm that the primary

factors influencing the intention to use AI assistants prevail across cultures. The insights are

inherent in demonstrating that the technological features and alignment of the AI assistants

with the customer needs and preferences supersede cultural dimensions in e-commerce.

7.2.8 Mixed-Method Findings

This section discusses the mixed method findings of quantitative and qualitative studies of this

research. Firstly, consumers who have an increased perception of AI assistants' usefulness

exhibit a correspondingly favourable disposition toward their integration within e-commerce

applications (Balakrishnan & Dwivedi, 2021). This alignment is substantiated by qualitative

insights, wherein consumers manifest appreciation for the salient attributes of these AI entities,

including the provision of illustrative imagery, real-time updates, and expeditious

responsiveness during online shopping interactions (Xiong, 2022). Evident from the synthesis

of quantitative and qualitative facets is a harmonious convergence, wherein the quantitative

154
evidence corroborates the affirmative impact of perceived usefulness on attitudes. In contrast,

the qualitative revelations amplify this effect by elucidating the intricate utility dimensions

illuminated by consumers' experiential encounters.

The findings from this study affirm the perception that the perceived ease of use and usefulness

have a positive correlation with the altitude and intention to adopt and continue using AI

assistants in e-commerce. Integration of the qualitative and quantitative findings demonstrates

that cultural differences that exist among Western and Eastern cultures have no significant

impact on the perceived ease of use and usefulness. Therefore, cultural differences have no

impact on the attitude and intention to adopt and use AI shopping assistants. However, the

personalisation of AI assistant services is essential in enhancing customer satisfaction and

creating value for the company and customers in their shopping experience.

The dimension of ease of use unfurls a comparable narrative, as consumers attributing inherent

ease to using AI assistants manifest an inherently positive stance towards their adoption

(Bawack & Desveaud, 2022). This alignment finds reinforcement in the qualitative domain,

where online shoppers extol the virtues of streamlined product searches, visual presentations,

personalised recommendations, navigational fluidity, accessibility enhancements, and

interactive engagements that embellish their virtual shopping sojourns. The quantitative

assessment underscores this synergy by affirming the affirmative influence of ease of use on

consumer attitudes, and the qualitative expositions validate this trajectory by accentuating the

pivotal constituents of user-friendliness through the lens of consumers' lived encounters.

Personalisation, another cardinal dimension, emanates as a pivotal determinant, positing that

AI assistants' proclivity to furnish bespoke experiences begets a correspondingly elevated

proclivity of consumers toward their utilisation (Kumar et al., 2019). This contention is

buttressed by qualitative accounts wherein consumers express contentment with personalised

155
product offerings and pertinent suggestions grounded in their unique requisites. This qualitative

semblance resonates harmoniously with the quantitative substrate, wherein it becomes palpable

that personalisation accrues as a salient contributor to fostering favourable consumer attitudes.

The qualitative strands further support this standpoint by accentuating the conspicuous

resonance engendered by the discerning alignment of AI-enabled personalisation and

consumers' bespoke expectations.

The findings on personalisation in the current study align with the conceptualisation that a

tailored approach to interacting with customers leads to higher satisfaction and positive

attitudes and intention to continue using AI assistants. Additionally, the study demonstrates that

personalisation is valued by customers from all cultures, which implies that it is a factor that

should be integrated into the AI assistants adopted in e-commerce. Such considerations can

contribute to enhanced marketing capabilities and a driver for competitive advantage in the

dynamic business environment characterised by a highly diverse customer base.

Communication, the final dimension under scrutiny, accentuates AI assistants' interactive

conversational adeptness in shaping consumers' predisposition toward their incorporation (Ho,

2021). As discerned through the qualitative lens, consumers evince distinct gratification from

sophisticated, affable, and efficient communication throughout their virtual interactions. The

quantitative landscape substantiates this panorama, delineating that interactive communication

has substantive sway over consumers' attitudes. The qualitative dimension dovetails elegantly

with this purview, unravelling the intricate interplay between communication styles and

customer satisfaction, converging with the quantitative findings in a harmonious confluence.

Quantitative and qualitative strands interlace in the tapestry of these dimensions, imbuing the

study's findings with robust and multi-dimensional validation. The findings in the current study

show a consensus that interactive communication is inherent in developing altitude and

intention to use AI shopping assistants. In the cross-cultural context, interactive communication

156
enhances the perceptions about the ease of use and usefulness of AI assistants. These insights

show that interactive communication is considered essential by all consumers. Therefore,

enhancing the interactive communication capabilities of the AI assistants can contribute to the

positive attitudes and intention to use among all customers. The communication should be

categorised by alignment of the language with the customers' cultural background for all users

to engage with the AI assistants and the companies effectively gain valuable feedback that can

be used for continuous improvement of the technologies and offerings.

7.4 Contribution and Implication for this Study

7.4.1 Theoretical Implications

In technological adoption and human-computer interaction, the interplay between usefulness,

ease of use, interactive communication, personalisation with AI, attitude towards AI usage, and

behavioural intention to use AI assistants reveals a complex web of theoretical implications

(Chen et al., 2022). The finding that perceived usefulness positively influences users' and non-

users' attitudes towards AI usage has significant theoretical implications. This finding is

consistent with TAM and related behavioural theories, affirming the crucial role of perceived

usefulness in shaping individuals' attitudes toward adopting new technologies. It underscores

the universality of this construct, suggesting that perceptions of a technology's practical

benefits not only influence current users' attitudes but also impact how attitudes develop among

those unfamiliar with the technology. This suggests that efforts to enhance perceived usefulness

through research or highlighting utilitarian advantages could effectively foster positive

attitudes among existing users and mitigate skepticism or resistance among non-users. These

insights contribute to a deeper understanding of technology adoption dynamics.

Furthermore, the finding that perceived ease of use positively influences users' attitudes

towards using AI in e-commerce holds significant theoretical implications. This suggests that

157
AI systems' perceived simplicity and user-friendliness play a pivotal role in shaping

individuals' attitudes, transcending the distinction between those familiar with the technology

and those not. This emphasises the critical role of developing intuitive and accessible AI

interfaces to improve attitudes and adoption across diverse user groups. It underscores the

potential to reduce entry barriers and enhance overall acceptance of AI-driven e-commerce

platforms (Balakrishnan & Dwivedi, 2024). In addition, theoretical implications arise when

exploring the relationship between usefulness and ease of use and consumer attitudes toward

AI assistants, shedding light on how these factors interact and influence one another. Here,

some of these theoretical implications are discussed. First, this study makes a valuable

contribution to the existing body of research on AI technology. Second, this research builds

upon existing studies on technology acceptance by exploring novel external factors that

enhance the effectiveness of AI assistants, thereby contributing to the advancement of AI

literature (Istiqomah & Alfansi, 2024). Moreover, it is central to this intricate framework

regarding AI assistants' usefulness and ease of use, which act as foundational pillars shaping

their perceptions and subsequent behaviours (De Cicco et al., 2020). Perceived usefulness,

reflecting the extent to which individuals believe AI assistants enhance their efficiency and

effectiveness, can be seen as a driving force behind the willingness to engage with such

technology (Zhang et al., 2021). Similarly, perceived ease of use, encompassing the perceived

usefulness of integrating AI assistants into daily routines, directly influences users' attitudes

and adoption decisions (Uzir et al., 2023).

Moreover, AI assistant characteristics amplify users' engagement and satisfaction, particularly

in interactive communication and personalisation (Alkhateeb et al., 2023). The degree to which

AI assistants facilitate seamless, human-like interactions plays a vital role in mitigating the

psychological gap between users and machines. By employing natural language processing and

context comprehension, AI assistants foster meaningful and fluid dialogues, enhancing the

158
overall user experience. Furthermore, personalisation, tailoring interactions and

recommendations based on individual preferences and behaviour history, adds another layer of

appeal (Dwivedi et al., 2021). This adaptation fosters a sense of AI assistants being personalised

companions, aligning with users' unique needs and desires, consequently solidifying positive

attitudes and behavioural intentions.

Additionally, attitude, a pivotal psychological construct, mediates behavioural intentions

(Acikgoz et al., 2023). Favourable attitudes stemming from perceived usefulness, ease of use,

interactive communication, and personalisation are precursors to users' propensity to integrate

AI assistants into their lives (Ashrafi & Easmin, 2023). Furthermore, the attitude-behaviour

relationship is manifested in the behavioural intention to use AI assistants. Individuals’ intent

to engage with AI technology is influenced by their perception of its value, compatibility with

their needs, and the anticipation of a gratifying experience (de Andrés-Sánchez & Gené-Albesa,

2023). The intention signifies the users' inclination to embrace AI assistants and reflects their

proactive approach towards incorporating them as indispensable tools. This dynamic interplay

reinforces those individual attitudes, fortified by perceived utility and ease, which are pivotal

determinants of AI adoption trajectories.

Finally, the lack of support for the hypothesised difference between Western and Eastern

consumers' attitudes and intentions towards using AI assistants bears essential theoretical

implications. While prior research has often emphasised cultural variations in technology

acceptance, the present model's failure to uncover such disparities suggests a need for re-

evaluation. Western societies, marked by their individualistic orientation (Kim et al., 1994) and

emphasis on self-reliance, commonly cultivate a milieu conducive to swift technology

assimilation, given its consonance with concepts of self-actualisation and pragmatic efficacy

(Leidner & Kayworth, 2006). Conversely, Eastern cultures, notable for their emphasis on

collectivism and communal harmony (Triandis, 2015), often exhibit a cautious attitude towards

159
technological integration (Erez et al., 1993). This outcome challenges assumptions that cultural

factors inherently lead to divergent perceptions of AI technology, highlighting the significance

of individual-level variables and context-specific influences. These results underscore the

complex interplay of psychological, socioeconomic, and situational factors in shaping attitudes

and intentions towards AI across diverse consumer groups, urging researchers to adopt a more

nuanced and holistic approach when examining cross-cultural technology adoption patterns.

The revelations emerging from this study demonstrate the need to expand the TAM model to

cover the diverse aspects of technology adoption in the modern day. Although TAM is valuable

in assessing the primary factors of perceived ease of use and usefulness in shaping the attitude

and intention to use, the model needs to expound on the underlying factors that influence

decision-making. Integrating a cultural dimension can enhance how TAM is employed as a

theoretical basis for investigating a diverse population of technology users. A cultural

dimension can help to establish the differences among people and the factors that align their

interests in a particular technology with their characteristics. Suggestively, the TAM can be

improved by highlighting how culture is essential in shaping the psychological and cognitive

factors that drive attitude and intention to use a particular technology.

In conclusion, converging AI assistant attributes, attitudes, and behavioural intentions establish

a multifaceted paradigm that shapes the landscape of AI technology assimilation. Recognising

the intricate links between these elements is indispensable for designing effective strategies to

encourage AI adoption. Developers must cultivate intuitive interfaces that enhance perceived

ease of use while integrating interactive communication and personalisation, ensuring positive

experiences. Simultaneously, fostering positive initial experiences and streamlined onboarding

processes can cultivate favourable attitudes that steer individuals towards embracing AI

technology. As AI continues its transformative march, a nuanced understanding of these

160
theoretical implications will be instrumental in navigating the complex terrain of human-AI

interaction.

7.4.2 Practical Implications

The study's findings regarding AI’s usefulness and ease of use, coupled with AI assistant

characteristics such as interactive communication and personalisation, have significant

practical implications. By considering the practical implications outlined below, businesses can

better navigate cultural diversity and tailor their AI assistant offerings to effectively engage and

meet the needs of diverse consumer markets.

Firstly, businesses should focus on enhancing AI assistants' functionality, aligning them with

users' perceptions of usefulness and ease of use to drive adoption. In addition, marketing

managers have the opportunity to enrich customers' experience through the introduction of

advanced AI assistants that can effectively provide customer support. The research findings

demonstrate that the attributes of AI assistants, particularly interactive communication,

significantly impact consumers' attitudes towards these AI assistants. The finding implies that

consumers are likely to use and adapt this AI application regularly, and managers can gain

advantages from understanding their customers’ chat preferences and integrating various

elements of communication style design, such as language, context, empathy, and friendliness.

Managers must recognise the significance of aligning communication styles with customer

preferences to maximise the application's utilisation and ensure its success in the market.

Understanding consumers' chatting styles can influence user experience design to embed

personalised chat options to match customer requirements or personalities. Implementing

personalised conversation options offers consumers greater control over their interactions,

empowering them to modify language preferences, adjust their tone of voice, and receive

tailored recommendations. By incorporating these features, AI assistants enhance the perceived

value for users, resulting in an improved overall experience. This customisation enables brands

161
to establish stronger customer connections, enhancing online customer experience and

satisfaction. Furthermore, these adaptations contribute to a more seamless and meaningful

brand-consumer interaction, fostering increased loyalty and brand affinity. Consequently,

integrating personalised conversation options benefits consumers by giving them a sense of

agency. It serves as a valuable tool for brands seeking to optimise their customer experiences

in the digital realm.

Second, this study also provides insights for AI developers; for instance, AI assistant

developers can enhance their products by prioritising prompt and efficient responses to

customer inquiries while also implementing filtering mechanisms to deliver more relevant and

personalised content. This improvement is likely to have a more significant impact on the long-

term viability of an information system. Developers need to recognise and capitalise on such

opportunities to optimise user experiences and achieve better outcomes in AI development.

Moreover, changing attitudes towards AI assistants will profoundly impact consumers’

inclination to utilise AI assistants for communicating with a brand through e-commerce

platforms. This positive and transformative change is critical because firms across nearly every

business sector (e.g., retailing, manufacturing, healthcare, and financial) progressively allocate

more resources to their technology investments. The imperative drives these organisations to

accomplish various objectives (Grewal et al., 2020) and for financial savings. Additionally,

consumers may interact more with companies by embracing AI assistants as a communication

medium, resulting in improved customer experiences and more convenience. This positive

trend is a critical opportunity for organisations to harness new technology and successfully

communicate with their target audience, promoting growth and competitive advantage in the

market. Furthermore, AI-based assistants may offer a cost-effective way to increase and

seamlessly scale up AI interactions with customers across multiple platforms (Grewal et al.,

2020). The successful implementation of chatbots and virtual assistants may increase

162
communication effectiveness and improve customer support productivity by integrating AI

capabilities (Chong et al., 2021; Clark, 2020). Furthermore, the lack of supported differences

in attitudes and intentions towards using AI assistants between Western and Eastern consumers

suggests significant practical implications. Organisations developing AI technologies and

services can adopt a more standardised approach in designing and marketing AI assistants,

focusing on universal user needs and preferences rather than tailoring extensively to regional

differences. This can lead to cost savings in development and marketing efforts, allowing for

more efficient resource allocation. However, it is crucial to remain vigilant and continue

monitoring potential regional shifts in attitudes and preferences, as cultural, socioeconomic,

and technological factors may evolve, potentially altering the dynamics between Western and

Eastern consumer segments in the future.

The insights emerging from the current study demonstrate the need for a strategic approach to

designing and implementing AI technologies in e-commerce that involves gathering adequate

data on the target customer base. Although there is no significant statistical difference between

Western and Eastern cultures in the primary factors that influence attitudes and intentions to

adopt AI shopping assistants, the relevance of interactive communication and personalisation

demonstrates the need to focus on understanding the customer. Therefore, technology

developers and the management of e-commerce businesses require input from customers in the

development of AI solutions to enhance the capacity to communicate. For instance,

understanding the use of language among the target customers can improve the flow of

information, facilitating valuable feedback for the company while improving the customer

experience by meeting their specific needs and preferences. The findings show that there are

opportunities to increase the societal acceptance of AI technologies among e-commerce

consumers. Non-users can be transformed into users by targeting the psychological and

cognitive factors that influence the attitudes and intentions to use a particular technology. Such

163
achievements entail focusing on the technology features that enhance the perceived ease of use

and usefulness. This can be achieved by offering the relevant information to enhance the

customer skills and knowledge concerning AI assistants and developing them to respond to the

diverse needs of experienced and new users. Leveraging the large volumes of data available in

the organisational and third-party platforms can improve the AI assistants by integrating

cultural sensitivity in responding to customer needs and preferences. However, ethical

considerations should be upheld when using customer data from within and outside the

company, considering the impact data breaches and privacy violations can have on customers

and the companies involved.

7.5 Conclusion

Technology now provides online assistance in e-commerce through AI assistants for global

consumers. This research undertook an interdisciplinary approach by conducting two distinct

studies. Firstly, an empirical survey questionnaire was employed to examine the relationships

among the constructs of the proposed model. Next, the second study used ML and NLP

techniques to analyse customer reviews and achieve the objective of eliciting supportive

insights. The survey data analysis employs various statistical techniques, including primary

descriptive analyses, multivariate analysis, and PLS-SEM utilising SPSS and SmartPLS

(Version 4) software programs. Through these analyses, the survey validates an empirical

model that effectively captured the interrelationships among the constructs of the conceptual

model and formulated a hypothesis. Subsequently, the second stage of the research focuses on

conducting the NLP analysis, aligning with the study's objectives. The study findings indicate

that the usefulness, interactive communication, and personalisation were well supported by the

users and non-users of AI assistants, and these factors positively influenced the attitudes, while

ease of use was not supported by non-users. In addition, attitude positively impacts usage

intentions to continue using AI assistants among users and non-users. Moreover, there were no

164
significant differences exist in the relationships among the primary factors influencing the

intention to utilise AI assistants in e-commerce when comparing Western and Eastern cultural

groups. Furthermore, these findings emphasise the importance of e-commerce users carefully

selecting a suitable characteristic for AI assistants during development to ensure customer

satisfaction. In conclusion, this research contributes to the understanding of AI assistant usage.

It provides practical insights for developing AI virtual assistants, providing inspiration for

future researchers in this emerging field. In addition, the collaborative efforts of AI developers,

researchers, and users can redefine the boundaries of human-computer interaction and lead us

into an era where AI assistants are seamlessly integrated into the fabric of daily existence. As

the potential of AI continues to unfold, understanding and adapting to these dynamics will be

pivotal in charting the course of technology adoption and its impact on society.

7.6 Limitations and Direction for Future Research

Although this thesis makes a significant contribution to AI assistant adoption and its usage in

e-commerce, some limitations should be noted for future research. The following points

summarise these limitations and provide recommendations for future research.

 There was a limitation regarding the focus of this study on specific factors that limited

the ability to explore other factors of AI capabilities that could be a significant

influencer on the attitudes towards AI assistants. Future studies could explore the

impact of other AI assistants’ capabilities, like proactive assistance and advanced

problem-solving, that might influence consumers' beliefs and attitudes towards using

AI assistants. Given the focus of this study on the perspectives of users and non-users,

however, external factors of consumers’ demographics, such as age and gender, may

impact the adoption of AI assistants. Future studies could investigate how a specific

group of age or gender responds to the AI assistant interaction and explore their

differences and similarities towards their intention to use this technology.

165
 There was a limitation regarding the sampling and sample size. This study has

variations in sample sizes between the quantitative and qualitative components, which

could affect the balance of insights derived from each component. Additionally, the

focus on Louis Vuitton e-commerce applications in the qualitative study might limit the

generalizability of findings to a broader range of e-commerce platforms. Future studies

could include a comparable sample size and consider a wider selection of e-commerce

platforms to ensure a more balanced integration of insights and enhance the

generalizability of findings. Also, a limitation of the quantitative study in relation to

dividing the sample into Western and Eastern cultures based on the regions of the

responses. Future research could explore more country-specific studies to deepen

understanding of cultural influences on AI assistant use. The data from the two

countries can show how people from different places think and act differently. It helps

compare the users' differences that affect their attitudes and actions. But to understand

global trends better for using AI technologies, including more countries would be

helpful. The use of IP address in sampling lacked accuracy and led to assumptions that

participants' location reflected their cultural background, ignoring that they might be

using VPN proxies or might have travelled or moved to their current locations. Future

studies should focus on the experience and attitude that emerge from users, factoring in

the diversity in demographic backgrounds to establish how social factors, norms, and

beliefs affect their perceptions towards AI assistants. Therefore, future studies could

purposefully select a diverse set of countries from both Western and Eastern regions

and use alternative sampling techniques that ensure accuracy in the characteristics of

the participants, considering that the IP address approach used in this study did not

guarantee that individuals were from their respective cultures.

166
 Another limitation of the study was the focus on extending the TAM model, limiting

the ability to adopt other models and explore other external factors that may have also

been significant for fostering AI assistant adoption in e-commerce. Future studies may

explore alternative research models, such as a combination of the TRA and cultural

dimensions theories, to delve into the cultural factors influencing the usage behaviour

of AI assistants. Further research could also focus on media influence, peer opinions,

or marketing campaigns promoting AI assistant usage.

 From a research context point of view, in this study, the main domain is e-commerce.

Further investigations in AI applications have the potential to expand upon discoveries

by examining additional factors that influence the intention to utilise AI assistants in

various contexts. For instance, exploring the impact of such predictors in education or

healthcare can enhance the understanding of this phenomenon.

 There was also a limitation regarding the two examples of AI assistant interaction

simulations that were presented to participants, which limited the ability to obtain in-

depth perceptions regarding the usage of AI assistants. Future studies can explore

incorporating participants' real-time interaction with AI assistants, which would

provide valuable insights into the evolving interests of consumers during such

interactions. Furthermore, it will also help identify specific pain points and outcomes

in live chats across varying timeframes. Interviewing 10-15 customers should be

conducted to give deeper explanations that match the online review data. This approach

will provide a better understanding of customer experiences and opinions, allowing for

the identification of deep insights that complement the findings of the quantitative data.

These interviews can also provide detailed feedback on user satisfaction, areas needing

improvement, and suggestions for making AI assistants better.

167
7.7 Chapter Summary

This chapter offers an in-depth analysis of the research findings and provides recommendations

based on those findings. It begins by revisiting the research aim and the research questions.

Following that, it presents a discussion of the study's findings. The theoretical and practical

implications of the study's findings are discussed. Finally, this chapter concluded by identifying

the study's limitations and providing recommendations for future research.

168
References

Al-Adwan, A. S., Li, N., Al-Adwan, A., Abbasi, G. A., Albelbisi, N. A., & Habibi, A. (2023). Extending
the technology acceptance model (TAM) to Predict University Students’ intentions to use metaverse-
based learning platforms. Education and Information Technologies, 28(11), 15381-15413.
Abdalla, M. M., Oliveira, L. G. L., Azevedo, C. E. F., & Gonzalez, R. K. (2018). Quality in qualitative
organizational research: Types of triangulation as a methodological alternative. Administração: ensinoe
pesquisa, 19(1).
AbuShawar, B., & Atwell, E. (2015). ALICE chatbot: Trials and outputs. Computación y
Sistemas, 19(4), 625-632.
Acikgoz, F., Perez‐Vega, R., Okumus, F., & Stylos, N. (2023). Consumer engagement with AI‐powered
voice assistants: A behavioural reasoning perspective. Psychology & Marketing.
Adiwardana, D., Luong, M. T., So, D. R., Hall, J., Fiedel, N., Thoppilan, R., ... & Le, Q. V. (2020).
Towards a human-like open-domain chatbot. arXiv preprint arXiv:2001.09977.
Adam, M., Wessel, M., & Benlian, A. (2021). AI-based chatbots in customer service and their effects
on user compliance. Electronic Markets, 31(2), 427-445.
Aguinis, H., Villamor, I., & Ramani, R. S. (2021). MTurk research: Review and
recommendations. Journal of Management, 47(4), 823-837.
Aguinis, H., Villamor, I., & Ramani, R. S. (2021). MTurk Research: Review and Recommendations.
Journal of Management, 47(4), 823–837.
Ajzen, I. (2020). The theory of planned behaviour: Frequently asked questions. Human Behavior and
Emerging Technologies, 2(4), 314-324.
Ajzen, I., and Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. Englewood
Cliffs, NJ: Prentice-Hall, Inc.
Ajisoko, P. (2020). The use of Duolingo apps to improve English vocabulary learning. International
Journal of Emerging Technologies in Learning (iJET), 15(7), 149-155.
Ahmed, A., Aziz, S., Khalifa, M., Shah, U., Hassan, A., Abd-Alrazaq, A., & Househ, M. (2022).
Thematic analysis on user reviews for depression and anxiety chatbot apps: Machine learning
approach. JMIR Formative Research, 6(3), e27654.
Akhtar, M., Neidhardt, J., & Werthner, H. (2019). The potential of chatbots: analysis of chatbot
conversations. In 2019 IEEE 21st Conference on Business Informatics (CBI) (Vol. 1, pp. 397-404).
IEEE.
Akour, I., Alnazzawi, N., Alshurideh, M., Almaiah, M. A., Al Kurdi, B., Alfaisal, R. M., & Salloum, S.
(2022). A Conceptual Model for Investigating the Effect of Privacy Concerns on E-Commerce
Adoption: A Study on United Arab Emirates Consumers. Electronics, 11(22), 3648.
Ali, I. (2020). The COVID-19 pandemic: Making sense of rumour and fear: Op-ed. medical
anthropology, 39(5), 376-379.
Alduaij, M. (2019). ‘Employing the technology acceptance model to explore the trends of social media
adoption and its effect on perceived usefulness and ease of use’. Journal of Information Technology
Management, 11(2), 129–143.
Aldboush, H. H., & Ferdous, M. (2023). Building Trust in Fintech: An Analysis of Ethical and Privacy
Considerations in the Intersection of Big Data, AI, and Customer Trust. International Journal of
Financial Studies, 11(3), 90.

169
Alnefaie, A., Gupta, D., Bhuyan, M. H., Razzak, I., Gupta, P., & Prasad, M. (2020). End-to-end analysis
for text detection and recognition in natural scene images. In 2020 International Joint Conference on
Neural Networks (IJCNN) (pp. 1-8). IEEE.
Al-Adwan, A. S., Li, N., Al-Adwan, A., Abbasi, G. A., Albelbisi, N. A., & Habibi, A. (2023). Extending
the technology acceptance model (TAM) to Predict University Students’ intentions to use metaverse-
based learning platforms. Education and Information Technologies, 1-33.
Alnefaie, A., Singh, S., Kocaballi, A. B., & Prasad, M. (2021). Factors Influencing Artificial Intelligence
Conversational Agents Usage in the E-commerce Field: A Systematic Literature Review.
Altinok, D. (2018). An ontology-based dialogue management system for banking and finance dialogue
systems—arXiv preprint arXiv:1804.04838.
Alves, T., Natálio, J., Henriques-Calado, J., & Gama, S. (2020). Incorporating personality in user
interface design: A review. Personality and Individual Differences, 155, 109709.
Alharahsheh, H. H., & Pius, A. (2020). A review of key paradigms: Positivism VS
interpretivism. Global Academic Journal of Humanities and Social Sciences, 2(3), 39-43.
Alkhateeb, M., Duné, E., & Akriem, E. (2023). The Impact of AI On Internal Communication Within
an Organization: A Critical Examination of AI Adoption.
Ameen, N., Tarhini, A., Reppel, A., & Anand, A. (2021). Customer experiences in the age of artificial
intelligence. Computers in Human Behavior, 114, 106548.
Ammenwerth, E. (2019). Technology acceptance models in health informatics: TAM and UTAUT. Stud
Health Technol Inform, 263, 64-71.
Amblee, N., & Bui, T. (2011). Harnessing the influence of social proof in online shopping: The effect
of electronic word of mouth on sales of digital microproducts. International journal of electronic
commerce, 16(2), 91-114.
Antwi, S. K., & Hamza, K. (2015). Qualitative and quantitative research paradigms in business
research: A philosophical reflection. European journal of business and management, 7(3), 217-225.
Ansari, A., & Mela, C. F. (2003). E-customization. Journal of marketing research, 40(2), 131-145.
Araujo, T. (2018). Living up to the chatbot hype: The influence of anthropomorphic design cues and
communicative agency framing on conversational agent and company perceptions. Computers in
Human Behaviour, pp. 85, 183 –189.
Araujo, T., Helberger, N., Kruikemeier, S., & De Vreese, C. H. (2020). In AI, we trust? Perceptions
about automated decision-making by artificial intelligence. AI & society, 35, 611-623.
Arthur, R. (2017). Louis Vuitton Becomes Latest Luxury Brand to Launch a Chatbot.
Argal, A., Gupta, S., Modi, A., Pandey, P., Shim, S., & Choo, C. (2018). Intelligent travel chatbot for
predictive recommendation in echo platform. In 2018 IEEE 8th Annual Computing and Communication
Workshop and Conference (CCWC) (pp. 176-183). IEEE.
Aron, J. (2011). How innovative is Apple's new voice assistant, Siri?
Ashrafi, D. M., & Easmin, R. (2023). Okay, Google, Good to Talk to You… Examining the
Determinants Affecting Users’ Behavioral Intention for Adopting Voice Assistants: Does Technology
Self-Efficacy Matter? International Journal of Innovation and Technology Management, 20(02),
2350004.
Ashfaq, M., Yun, J., Yu, S., & Loureiro, S. M. C. (2020). I, Chatbot: Modeling the determinants of users’
satisfaction and continuance intention of AI-powered service agents. Telematics and Informatics, 54,
101473.

170
Asenahabi, B. M. (2019). Basics of research design: A guide to selecting appropriate research
design. International Journal of Contemporary Applied Research, 6(5), 76-89.
Bahja, M. (2020). Natural language processing applications in business. E-Business-higher education
and intelligence applications.
Baker, E., Al‐Gahtani, S. S., and Hubona, G. S. (2007). The effects of gender and age on new technology
implementation in a developing country: Testing the theory of planned behaviour (TPB). Information
Technology & People, 20(4), 352-375.
Balakrishnan, J., & Dwivedi, Y. K. (2024). Conversational commerce: entering the next stage of AI-
powered digital assistants. Annals of Operations Research, 333(2), 653-687.
Balakrishnan, J., & Dwivedi, Y. K. (2021). Conversational commerce: entering the next stage of AI-
powered digital assistants. Annals of Operations Research, 1-35.
Bavaresco, R., Silveira, D., Reis, E., Barbosa, J., Righi, R., Costa, C., ... & Moreira, C. (2020).
Conversational agents in business: A systematic literature review and future research
directions. Computer Science Review, 36, 100239.
Bawack, R., & Desveaud, K. (2022). Consumer Adoption of Artificial Intelligence: A Review of
Theories and Antecedents.
Bawack, R. E., Wamba, S. F., Carillo, K. D. A., & Akter, S. (2022). Artificial intelligence in E-
Commerce: a bibliometric study and literature review. Electronic markets, 32(1), 297-338.
Belli, G. (2008). Nonexperimental Quantitative Research. In: Lapan, S. D. & Quartaroli, M. T. (eds.)
Research Essentials: An Introduction to Designs and Practices. Wiley.

Berger, V. W., & Zhou, Y. (2014). Kolmogorov–smirnov test: Overview. Wiley statsref: Statistics
reference online.

Bhattacherjee, A. (2001). Understanding information systems continuance: An expectation-

confirmation model. MIS Quarterly, 351-370.
Bhawiyuga, A., Fauzi, M. A., Pramukantoro, E. S., & Yahya, W. (2017). Design of an e-commerce chat
robot for automatically answering customer questions. In 2017 International Conference on Sustainable
Information Engineering and Technology (SIET) (pp. 159-162). IEEE.
Billion, J., & Van den Abeele, D. (2007, June). ICOM: A Communication Framework for Interoperable
European Railways. In 2007 7th International Conference on ITS Telecommunications (pp. 1-6). IEEE.
Bland, J. M., & Altman, D. G. (1997). Statistics notes: Cronbach's alpha. BMJ, 314(7080), 572.

Bono, R., Arnau, J., Alarcón, R., & Blanca, M. J. (2019). Bias, precision, and accuracy of skewness and
kurtosis estimators for frequently used continuous distributions. Symmetry, 12(1), 19.
Brandtzaeg, P. B., & Følstad, A. (2018). Chatbots: changing user needs and
motivations. interactions, 25(5), 38-43.
Bryman, A. & Bell, E. (2015). Business Research Method’s, 4th edition, OUP Oxford.
Brill, T. M., Munoz, L., & Miller, R. J. (2022). Siri, Alexa, and other digital assistants: a study of
customer satisfaction with artificial intelligence applications. In The Role of Smart Technologies in
Decision Making (pp. 35-70). Routledge.
Burnette, C. Blair, Jessica L. Luzier, Brooke L. Bennett, Chantel M. Weisenmuller, Patrick Kerr, Shelby
Martin, Jillian Keener, and Lisa Calderwood. "Concerns and recommendations for using Amazon
MTurk for eating disorder research." International Journal of Eating Disorders 55, no. 2 (2022): 263-
272.

171
Cain, M. K., Zhang, Z., & Yuan, K. H. (2017). Univariate and multivariate skewness and kurtosis for
measuring nonnormality: Prevalence, influence, and estimation. Behaviour research methods, 49,
1716-1735.
Canhoto, A. I., Keegan, B. J., & Ryzhikh, M. (2023). Snakes and ladders: unpacking the personalisation-
privacy paradox in the context of AI-enabled personalisation in the physical retail
environment. Information Systems Frontiers, 1-20.
Casula, M., Rangarajan, N., & Shields, P. (2021). The potential of working hypotheses for deductive
exploratory research. Quality & Quantity, 55(5), 1703-1725.
Cavana, R., Delahaye, B., & Sekeran, U. (2001). Applied business research: Qualitative and
quantitative methods. John Wiley & Sons.
Cecez-Kecmanovic, D. (2005). Basic assumptions of the critical research perspectives in information
systems. Handbook of critical information systems research: Theory and application, (19-46).
Çelik, H. E., & Yilmaz, V. (2011). Extending the technology acceptance model for adopting e-shopping
by consumers in Turkey. Journal of Electronic Commerce Research, 12(2), 152.
Cha, J. (2010). Factors affecting the frequency and amount of social networking site use: Motivations,
perceptions, and privacy concerns. First Monday.
Cherrier, H. (2007). Ethical consumption practices: co‐production of self‐expression and social
recognition. Journal of Consumer Behaviour: An International Research Review, 6(5), 321-335.
Chandra, S., Verma, S., Lim, W. M., Kumar, S., & Donthu, N. (2022). Personalisation in personalised
marketing: Trends and ways forward. Psychology & Marketing, 39(8), 1529-1562.
Chandler, J., & Shapiro, D. (2016). Conducting clinical research using crowdsourced convenience
samples. Annual Review of Clinical Psychology, 12(1), 53–81.
Chalmers, A. F. (1999). What is this thing called Science? (Vol. Third Edition). Open University Press
Chaves, A. P., & Gerosa, M. A. (2021). How should my chatbot interact? A survey on social
characteristics in human–chatbot interaction design. International Journal of Human–Computer
Interaction, 37(8), 729-758.
Chen, H., Chan-Olmsted, S., Kim, J., & Mayor Sanabria, I. (2022). Consumers’ perception of artificial
intelligence applications in marketing communication. Qualitative Market Research: An International
Journal, 25(1), 125-142.
Chen, J. S., Le, T. T. Y., & Florence, D. (2021). Usability and responsiveness of artificial intelligence
chatbot on online customer experience in e-retailing. International Journal of Retail & Distribution
Management, 49(11), 1512–1531.
Chawla, N., & Kumar, B. (2022). E-commerce and consumer protection in India: the emerging
trend. Journal of Business Ethics, 180(2), 581-604.
Cheng, Y., & Jiang, H. (2022). Customer–brand relationship in the era of artificial intelligence:
understanding the role of chatbot marketing efforts. Journal of Product & Brand Management, 31(2),
252-264.

Cheng, M., & Jin, X. (2019). What do Airbnb users care about? An analysis of online review
comments. International Journal of Hospitality Management, 76, 58-70.

Chhabra, G., Vashisht, V., & Ranjan, J. (2017). A comparison of multiple imputation methods for data
with missing values. Indian Journal of Science and Technology, 10(19), 1-7.

172
Chi, T. (2018). Understanding Chinese consumer adoption of apparel mobile commerce: An extended
TAM approach. Journal of Retailing and Consumer Services, 44, 274-284.

Choudhury, A., Elkefi, S., & Tounsi, A. (2024). Exploring factors influencing user perspective of
ChatGPT as a technology that assists in healthcare decision making: A cross-sectional survey
study. PLoS One, 19(3), e0296151.

Choi, J., Lee, H. J., Sajjad, F., & Lee, H. (2014). The influence of national culture on the attitude towards
mobile recommender systems. Technological Forecasting and Social Change, 86, 65-79.
Choi, Y., Choi, M., Oh, M., & Kim, S. (2020). Service robots in hotels: understanding the service quality
perceptions of human-robot interaction. Journal of Hospitality Marketing & Management, 29(6), 613-
635.
Chi, O. H., Chi, C. G., Gursoy, D., & Nunkoo, R. (2023). Customers’ acceptance of artificially
intelligent service robots: The influence of trust and culture. International Journal of Information
Management, 70, 102623.
Chi, O. H., Chi, C. G., Gursoy, D., & Nunkoo, R. (2023). Customer’s acceptance of artificially
intelligent service robots: The influence of trust and culture. International Journal of Information
Management, 70, 102623.
Chong, T., Yu, T., Keeling, D. I., & de Ruyter, K. (2021). AI chatbots on the service’s frontline
addressing the challenges and opportunities of the agency. Journal of Retailing and Consumer Services,
63, 102735.
Chung, M., Ko, E., Joung, H., & Kim, S. J. (2020). Chatbot e-service and customer satisfaction
regarding luxury brands. Journal of Business Research, 117, 587-595.
Clark, S. 5 Ways Chatbots Improve Employee Experience. 2020.
Clarizia, F., Colace, F., Lombardi, M., Pascale, F., & Santaniello, D. (2018). Chatbot: An education
support system for students. In Cyberspace Safety and Security: 10th International Symposium, CSS
2018, Amalfi, Italy, October 29–31, 2018, Proceedings 10 (pp. 291-302). Springer International
Publishing.
Clark, L., Pantidi, N., Cooney, O., Doyle, P., Garaialde, D., Edwards, J., ... & Cowan, B. R. (2019,
May). What makes a good conversation? Challenges in designing truly conversational agents.
In Proceedings of the 2019 CHI conference on human factors in computing systems (pp. 1-12).
Collins, C., Dennehy, D., Conboy, K., & Mikalef, P. (2021). Artificial intelligence in information
systems research: A systematic literature review and research agenda. International Journal of
Information Management, 60, 102383.
Cornali, F., & Tirocchi, S. (2012). Globalization, education, information and communication
technologies: what relationships and reciprocal influences?. Procedia-Social and Behavioral
Sciences, 47, 2060-2069.
Conboy, K., Fitzgerald, G., & Mathiassen, L. (2012). Qualitative methods research in information
systems: motivations, themes, and contributions. European Journal of Information Systems, 21(2), 113-
118.
Carcary, M. (2009). The research audit trial—enhancing trustworthiness in qualitative
inquiry. Electronic journal of business research methods, 7(1), pp11-24.
Dahiya, M. (2017). A tool of conversation: Chatbot. International journal of computer sciences and
engineering, 5(5), 158-161.
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information
technology. MIS Quarterly, 319-340.

173
Daowd, A., Hasan, R., Eldabi, T., Rafi-ul-Shan, P. M., Cao, D., & Kasemsarn, N. (2021). Factors
affecting eWOM credibility, information adoption and purchase intention on Generation Y: a case from
Thailand. Journal of Enterprise Information Management, 34(3), 838-859.

Dawadi, S., Shrestha, S., & Giri, R. A. (2021). Mixed-methods research: A discussion on its types,
challenges, and criticisms. Journal of Practical Studies in Education, 2(2), 25-36.

Davenport, T., Guha, A., Grewal, D., & Bressgott, T. (2020). How artificial intelligence will change the
future of marketing. Journal of the Academy of Marketing Science, 48, 24-42.

De Villiers, M. R. (2005, July). Three approaches as pillars for interpretive information systems
research: development research, action research and grounded theory. In Proceedings of the 2005
annual research conference of the South African Institute of Computer Scientists and Information
Technologists on IT research in developing countries (pp. 142-151).

De Cicco, R., Silva, S. C., & Alparone, F. R. (2020). Millennials' attitude toward chatbots: an
experimental study in a social relationship perspective. International Journal of Retail & Distribution
Management, 48(11), 1213-1233.

de Andrés-Sánchez, J., & Gené-Albesa, J. (2023). Explaining Policyholders’ Chatbot Acceptance with
a Unified Technology Acceptance and Use of Technology-Based Model. Journal of Theoretical and
Applied Electronic Commerce Research, 18(3), 1217-1237.

de Kervenoael, R., Hasan, R., Schwob, A., & Goh, E. (2020). Leveraging human-robot interaction in
hospitality services: Incorporating the role of perceived value, empathy, and information sharing into
visitors’ intentions to use social robots. Tourism Management, 78, 104042.

Ding, S., & Saunders, R. A. (2006). Talking up China: An analysis of China’s rising cultural power and
global promotion of the Chinese language. East Asia, 23(2), 3-33.

Drigas, A., Papanastasiou, G., & Skianis, C. (2023). The School of the Future: The Role of Digital
Technologies, Metacognition and Emotional Intelligence. International Journal of Emerging
Technologies in Learning (Online), 18(9), 65.

Duffett, R. (2020). The YouTube marketing communication effect on cognitive, affective and
behavioural attitudes among Generation Z consumers. Sustainability, 12(12), 5075.

Duijst, D. (2017). Can we improve the user experience of chatbots with personalisation? Master's thesis.
University of Amsterdam.
Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs, C., Crick, T., ... & Williams, M. D.
(2021). Artificial Intelligence (AI): Multidisciplinary perspectives on emerging challenges,
opportunities, and agenda for research, practice and policy. International Journal of Information
Management, 57, 101994.
Elad, B. (2023). Online Shoppers Statistics by Country and Demographics. Enterprise Apps Today.
Elsholz, E., Chamberlain, J., and Kruschwitz, U. (2019). ‘Exploring language style in chatbots to
increase perceived product value and user engagement’. Paper presented at the Proceedings of the 2019
Conference on Human Information Interaction and Retrieval.
Erez, M., & Earley, P. C. (1993). Culture, self-identity, and work. Oxford University Press, USA.
Espinoza, J., Crown, K., & Kulkarni, O. (2020). A guide to chatbots for COVID-19 screening at
pediatric health care facilities. JMIR public health and surveillance, 6(2), e18808.

174
Eren, B. A. (2021). Determinants of customer satisfaction in chatbot use: evidence from a banking
application in Turkey. International Journal of Bank Marketing, 39(2), 294-311.
Fan, S., Radford, J., & Fabian, D. (2016). A mixed-method research to investigate the adoption of
mobile devices and Web2. 0 technologies among medical students and educators. BMC medical
informatics and decision making, 16(1), 1-8.
Fakis, A., Hilliam, R., Stoneley, H., & Townend, M. (2014). Quantitative analysis of qualitative
information from interviews: A systematic literature review. Journal of Mixed Methods Research, 8(2),
139-161.
Famili, A., Shen, W. M., Weber, R., & Simoudis, E. (1997). Data preprocessing and intelligent data
analysis. Intelligent Data Analysis, 1(1), 3–23. https://doi.org/10.3233/IDA-1997-1102.
Faqih, K. M. (2022). Internet shopping in the Covid-19 era: Investigating the role of perceived risk,
anxiety, gender, culture, and trust in the consumers’ purchasing behaviour from a developing country
context. Technology in Society, 70, 101992.
Faqih, K. M. (2016). An empirical analysis of factors predicting the behavioral intention to adopt
Internet shopping technology among non-shoppers in a developing country context: Does gender
matter?. Journal of retailing and consumer services, 30, 140-164.
Fokina, M. (2022). Online shopping statistics: E-commerce trends for 2022.
Tidio. https://www.tidio.com/blog/online-shopping-statistics/
Foroudi, P., Gupta, S., Sivarajah, U., & Broderick, A. (2018). Investigating the effects of smart
technology on customer dynamics and customer experience. Computers in Human Behavior, 80, 271-
282.
Fernandes, A. (2020). Luxury Chatbots Helping Combat Slump in Sales During Pandemic. Retrieved
from https://verloop.io/blog/how-luxury-chatbots-are-helping/.
Ferrario, A., & Nägelin, M. (2020). The art of natural language processing: classical, modern and
contemporary approaches to text document classification. Modern and Contemporary Approaches to
Text Document Classification.
Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods
designs—principles and practices. Health services research, 48(6pt2), 2134-2156.
Fielding, N. G. (2012). Triangulation and mixed methods designs: Data integration with new research
technologies. Journal of mixed methods research, 6(2), 124-136.
Filieri, R., & Mariani, M. (2021). The role of cultural values in consumers' evaluation of online review
helpfulness: a big data approach. International Marketing Review, 38(6), 1267-1288.
Fishbein, M. (1979). A theory of reasoned action: some applications and implications.

Field, A. (2009), Discovering Statistics Using SPSS. Third Edition, SAGE Publications Ltd.

Field, A. (2013). Discovering statistics using IBM SPSS statistics. Sage.

Frank, B., Herbas-Torrico, B., & Schvaneveldt, S. J. (2021). The AI-extended consumer: technology,
consumer, and country differences in the formation of demand for AI-empowered consumer
products. Technological Forecasting and Social Change, 172, 121018.
Foroughi, B., Iranmanesh, M., & Hyun, S. (2019). Understanding the determinants of mobile banking
continuance usage intention. Journal of Enterprise Information Management.
Fornell, C., & Larcker, D. F. (1981). Structural equation models with unobservable variables and
measurement error: Algebra and statistics.

175
Følstad, A., & Brandtzaeg, P. B. (2020). Users' experiences with chatbots: findings from a questionnaire
study. Quality and User Experience, 5(1), 3.
Forkus, S. R., Contractor, A. A., Goncharenko, S., Goldstein, S. C., & Weiss, N. H. (2022). Online
crowdsourcing to study trauma and mental health symptoms in military populations: A case for
Amazon’s Mechanical Turk (MTurk) platform. Psychological trauma: theory, research, practice, and
policy.
Galetsi, P., Katsaliaki, K., & Kumar, S. (2023). Exploring benefits and ethical challenges in the rise of
mHealth (mobile healthcare) technology for the common good: An analysis of mobile applications for
health specialists. Technovation, 121, 102598.
Ghazi, A. N., Petersen, K., Reddy, S. S. V. R., & Nekkanti, H. (2018). Survey research in software
engineering: Problems and mitigation strategies. IEEE Access, 7, 24703-24718.
Gjurković, M., & Šnajder, J. (2018). Reddit: A gold mine for personality prediction. In Proceedings of
the second workshop on computational modelling of people’s opinions, personality, and emotions in
social media (pp. 87-97).
Gao, Y., & Liu, H. (2022). Artificial intelligence-enabled personalisation in interactive marketing: a
customer journey perspective. Journal of Research in Interactive Marketing, (ahead-of-print), 1-18.
Gillespie, N., Lockey, S., Curtis, C., Pool, J., & Akbari, A. (2023). Trust in Artificial Intelligence: A
global study.
Gnewuch, U., Morana, S., Adam, M., & Maedche, A. (2018). Faster is not always better: understanding
the effect of dynamic response delays in human-chatbot interaction.
Goertzen, M. J. (2017). Introduction to quantitative research and data. Library Technology
Reports, 53(4), 12-18.
Go, E., & Sundar, S. S. (2019). Humanising chatbots: The effects of visual, identity and conversational
cues on humanness perceptions. Computers in Human Behavior, 97, 304-316.
Grewal, D., Hulland, J., Kopalle, P. K., & Karahanna, E. (2020). The future of technology and
marketing: A multidisciplinary perspective. Journal of the Academy of Marketing Science, 48(1), 1-8.

Grover, P., Kar, A. K., & Ilavarasan, P. V. (2019). Impact of corporate social responsibility on
reputation—Insights from tweets on sustainable development goals by CEOs. International journal of
information management, 48, 39-52.
Greene, J. C., Caracelli, V. J., & Graham, W. F. (1989). Toward a conceptual framework for mixed-
method evaluation designs. Educational evaluation and policy analysis, 11(3), 255-274.
Goduka, N. (2012). From positivism to indigenous science: A reflection on world views, paradigms and
philosophical assumptions. Africa Insight, 41(4), 123-138.
Gümüş, N., & Çark, Ö. (2021). The Effect of Customers’ Attitudes Towards Chatbots on their
Experience and Behavioural Intention in Turkey. Interdisciplinary Description of Complex Systems:
INDECS, 19(3), 420-436.
Gupta, S., Borkar, D., De Mello, C., & Patil, S. (2015). An e-commerce website-based
chatbot. International Journal of Computer Science and Information Technologies, 6(2), 1483-1485.
Guetterman, T. C., & Fetters, M. D. (2018). Two methodological approaches to the integration of mixed
methods and case study designs: A systematic review. American Behavioral Scientist, 62(7), 900-918.
Guo, W., & Luo, Q. (2023). Investigating the impact of intelligent personal assistants on the purchase
intentions of Generation Z consumers: The moderating role of brand credibility. Journal of Retailing
and Consumer Services, 73, 103353.

176
Guzman, A. L. (2016). Making AI safe for humans: A conversation with Siri. In Socialbots and their
friends (pp. 85-101). Routledge.
Guzman, A. L., & Lewis, S. C. (2020). Artificial intelligence and communication: A human–machine
communication research agenda. New media & society, 22(1), 70-86.
Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and how to report the results
of PLS-SEM. European Business Review, 31(1), 2-24.
Hair, J., Hollingsworth, C. L., Randolph, A. B., & Chong, A. Y. L. (2017). An updated and expanded
assessment of PLS-SEM in information systems research. Industrial management & data
systems, 117(3), 442-458.
Harlap, A., Cui, H., Dai, W., Wei, J., Ganger, G. R., Gibbons, P. B., ... & Xing, E. P. (2016, October).
Addressing the straggler problem for iterative convergent parallel ML. In Proceedings of the seventh
ACM symposium on cloud computing (pp. 98-111).
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2013). Partial least squares structural equation modelling:
Rigorous applications, better results and higher acceptance. Long range planning, 46(1-2), 1-12.

Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (2010). Analysis using dependence
techniques. Multivariate Data Analysis, 7th ed. New Jersey: Prentice Hall, 154-176.Haleem, A., Javaid,
M., Qadri, M. A., Singh, R. P., & Suman, R. (2022). Artificial intelligence (AI) applications for
marketing: A literature-based study. International Journal of Intelligent Networks.

Hasan, M. R., Maliha, M., & Arifuzzaman, M. (2019). Sentiment analysis with NLP on Twitter data. In 2019
International Conference on Computer, Communication, Chemical, Materials and Electronic Engineering
(IC4ME2) (pp. 1-4). IEEE.
Hasan, A. A. T. (2022). Determinants of intentions to use the Foodpanda mobile application in
Bangladesh: the role of attitude and fear of COVID-19. South Asian Journal of Marketing, 4(1), 17-32.

Hasan, I., Ahmed, S. P., Ahmed, S. U., & Yousuf, T. B. (2021). Factors influencing users' willingness
for online messaging services: a developing country perspective. International Journal of Mobile
Communications, 19(1), 75-98.

Hassanein, K., & Head, M. (2007). ‘Manipulating perceived social presence through the web interface and its
impact on attitude towards online shopping’. International journal of human-computer studies, 65(8), 689-708.
Harasis, A. A., Qureshi, M. I., & Rasli, A. (2018). Development of research continuous usage intention
of e-commerce. A systematic review of literature from 2009 to 2015. International Journal of
Engineering & Technology, 7(2.29), 73–78.
Harrigan, M., Feddema, K., Wang, S., Harrigan, P., & Diot, E. (2021). How trust leads to online
purchase intention founded in perceived usefulness and peer communication. Journal of Consumer
Behaviour, 20(5), 1297-1312.
Han, M. C. (2021). The impact of anthropomorphism on consumers’ purchase decision in chatbot
commerce. Journal of Internet Commerce, 20(1), 46-65.
Hau, K. T., & Marsh, H. W. (2004). The use of item parcels in structural equation modelling: Non‐
normal data and small sample sizes. British Journal of Mathematical and Statistical Psychology, 57(2),
327-351.
Heale, R., & Twycross, A. (2015). Validity and reliability in quantitative studies. Evidence-based
nursing, 18(3), 66-67.
Hernani-Merino, M., Lazo, J. G. L., López, A. T., Mazzon, J. A., & López-Tafur, G. (2020). An
international market segmentation model based on susceptibility to global consumer culture. Cross
Cultural & Strategic Management, 28(1), 108-128.

177
Hernández, B., Jiménez, J., & José Martín, M. (2011). Age, gender and income: Do they really moderate
online shopping behaviour? Online information review, 35(1), 113-133.
Hesham, F., Riadh, H., & Sihem, N. K. (2021). What have we learned about the effects of the COVID-
19 pandemic on consumer behaviour? Sustainability, 13(8), 4304.
Henseler, J., Ringle, C. M., & Sarstedt, M. (2016). Testing measurement invariance of composites using
partial least squares. International marketing review, 33(3), 405-431.
Hitchcock, J. H., & Nastasi, B. K. (2024). Advancing Cross-Cultural Research Through Mixed Methods
Inquiry. Journal of Mixed Methods Research, 15586898241256207.
Hien, H. T., Cuong, P. N., Nam, L. N. H., Nhung, H. L. T. K., & Thang, L. D. (2018). Intelligent
assistants in higher-education environments: the FIT-EBot, a chatbot for administrative and learning
support. In Proceedings of the 9th International Symposium on Information and Communication
Technology (pp. 69-76).
Hitlin, P. (2016) Research in the Crowdsourcing Age, a Case Study.
Holden, M.T. & Lynch, P. (2004). Choosing the appropriate methodology: Understanding research
philosophy. The marketing review, 4(4), pp.397-409.
Hofstede, G. (1983). National cultures in four dimensions: A research-based theory of cultural
differences among nations. International studies of management & organization, 13(1-2), 46-74.
Hofstede, G. (2011). Dimensionalizing cultures: The Hofstede model in context. Online readings in
psychology and culture, 2(1), 8.
Hong, Y. Y., Wan, C., No, S., & Chiu, C. Y. (2007). Multicultural identities. Handbook of cultural
psychology, 323-345.
Ho, S. P. S., & Chow, M. Y. C. (2023). The role of artificial intelligence in consumers’ brand preference
for retail banks in Hong Kong. Journal of Financial Services Marketing, 1-14.
Ho, R. C. (2021). Chatbot for online customer service: customer engagement in the era of artificial
intelligence. In Impact of globalisation and advanced technologies on online business models (pp. 16-
31). IGI Global.
Hornung, O., & Smolnik, S. (2022). AI is invading the workplace: negative emotions towards the
organisational use of personal virtual assistants. Electronic Markets, 1-16.
Hornbæk, K., & Hertzum, M. (2017). Technology acceptance and user experience: A review of the
experiential component in HCI. ACM Transactions on Computer-Human Interaction (TOCHI), 24(5),
1-30.
Hsu, S. H. Y., Tsou, H. T., & Chen, J. S. (2021). “Yes, we do. Why not use augmented reality?” customer
responses to experiential presentations of AR-based applications. Journal of Retailing and Consumer
Services, 62, 102649.
Hsieh, J. Y., and Liao, P. W. (2011). Antecedents and moderators of online shopping behaviour in
undergraduate students. Social Behavior and Personality: An international journal, 39(9), 1271-1280.
Huda, M. N. (2023). Analysis the Critical Factors of M-government Service Acceptance: An Integrating
Theoretical Model between TAM and ECM. Policy & Governance Review, 7(2), 109-124.
Huang, M. H., & Rust, R. T. (2021). A strategic framework for artificial intelligence in marketing.
Journal of the Academy of Marketing Science, 49, 30-50.
Huang, C. Y., Yang, M. C., Huang, C. Y., Chen, Y. J., Wu, M. L., & Chen, K. W. (2018). A chatbot-
supported smart wireless interactive healthcare system for weight control and health promotion. In 2018

178
IEEE International Conference on Industrial Engineering and Engineering Management (IEEM) (pp.
1791-1795).
Hussain, S., Ameri Sianaki, O., & Ababneh, N. (2019). A survey on conversational agents/chatbot
classification and design techniques. In Web, Artificial Intelligence and Network Applications:
Proceedings of the Workshops of the 33rd International Conference on Advanced Information
Networking and Applications (WAINA-2019) 33 (pp. 946-956). Springer International Publishing.
Ikumoro, A. O., & Jawad, M. S. (2019). Intention to use intelligent conversational agents in e-commerce
among Malaysian SMEs: an integrated conceptual framework based on tri-theories including unified
theory of acceptance, use of technology (UTAUT), and TOE. International Journal of Academic
Research in Business and Social Sciences, 9(11), 205-235.
Insider. (2020). Chatbot Market in 2021: Stats, trends, and Companies in the Growing AI chatbot
industry. Retrieved from https://www.businessinsider.com/chatbot-market-stats-trends.
Istiqomah, P., & Alfansi, L. (2024). Navigating Style: Exploring the Influence of Perceived Benefit and
Perceived Ease of Use on Attitude Towards Use in AI-Enhanced Fashion E-Commerce. Journal of
Entrepreneurship and Business, 5(1), 1-14.
Ivankova, N. V., Creswell, J. W., & Stick, S. L. (2006). Using mixed-methods sequential explanatory
design: From theory to practice. Field methods, 18(1), 3-20.
Jain, S. P., & Weiten, T. J. (2020). Consumer psychology of implicit theories: A review and
agenda. Consumer Psychology Review, 3(1), 60-75.
Jain, S., Basu, S., Dwivedi, Y. K., & Kaur, S. (2022). Interactive voice assistants–Does brand credibility
assuage privacy risks? Journal of Business Research, 139, 701-717.
Jafari, A., Aly, M., & Doherty, A. M. (2022). An analytical review of market system dynamics in
consumer culture theory research: Insights from the sociology of markets. Journal of Business
Research, p. 139, 1261–1274.
Jantsch, L. B., & Neves, E. T. (2022). “Talking table” as a data integration strategy in mixed methods
research. Escola Anna Nery, 27.
Jannach, D., Manzoor, A., Cai, W., & Chen, L. (2021). A survey on conversational recommender
systems. ACM Computing Surveys (CSUR), 54(5), 1-36.
Jan, I. U., Ji, S., & Kim, C. (2023). What (de) motivates customers to use AI-powered conversational
agents for shopping? The extended behavioural reasoning perspective. Journal of Retailing and
Consumer Services, 75, 103440.
Jecker, N. S., & Nakazawa, E. (2022). Bridging East-West Differences in Ethics Guidance for AI and
Robotics. AI, 3(3), 764-777.
Jelavic, M., & Ogilvie, K. (2010). Knowledge management views in Eastern and Western cultures: an
integrative analysis. Journal of Knowledge Globalization, 3(2), 51-69.
Jervis, R. (2006). Understanding beliefs. Political Psychology, 27(5), 641–663.
Jeong, S. G., Hur, H. J., & Choo, H. J. (2020). The Effect of Fashion Shopping Chatbot Characteristics
on Service Acceptance Intention-Focusing on Anthropomorphism and Personalization. Journal of the
Korean Society of Clothing and Textiles, 44 (4), 573–593.
Jiang, H., Cheng, Y., Yang, J., & Gao, S. (2022). AI-powered chatbot communication with customers:
Dialogic interactions, satisfaction, engagement, and customer behaviour. Computers in Human
Behavior, 134, 107329.
Jovic, D. (2020). Fascinating Chatbot Statistics. Retrieved from https://www.smallbizgenius.net/by-the-
numbers/chatbot-statistics.

179
Jo, H. (2022). Continuance intention to use artificial intelligence personal assistant: type, gender, and
use experience. Heliyon, 8(9).
Kaaniche, N., Laurent, M., & Belguith, S. (2020). Privacy-enhancing technologies for solving the
privacy-personalization paradox: Taxonomy and survey. Journal of Network and Computer
Applications, 171, 102807.
Kashive, N., Powale, L., & Kashive, K. (2020). Understanding user perception toward artificial
intelligence (AI) enabled e-learning. The International Journal of Information and Learning
Technology, 38(1), 1-19.
Kalia, A. K., Telang, P. R., Xiao, J., & Vukovic, M. (2017). Quark: a methodology to transform people-
driven processes into chatbot services. In Service-Oriented Computing: 15th International Conference,
ICSOC 2017, Malaga, Spain, November 13–16, 2017, Proceedings (pp. 53-61). Springer International
Publishing.
Kasilingam, D. L. (2020). Understanding the attitude and intention to use smartphone chatbots for
shopping. Technology in Society, 62, 101280.
Kamalia, D., Djajadinata, M., Gunawan, F. H., & Gunadi, W. (2022). The role of hedonic motivation
and FOMO on the Impulsivity of e-commerce users during COVID-19 Pandemics in Indonesia.
In Proceedings of the International Conference on Industrial Engineering and Operations Management
Istanbul.
Kaushik, V., & Walsh, C. A. (2019). Pragmatism as a research paradigm and its implications for social
work research. Social sciences, 8(9), 255.
Kautish, P., Purohit, S., Filieri, R., & Dwivedi, Y. K. (2023). Examining the role of consumer
motivations to use voice assistants for fashion shopping: The mediating role of awe experience and
eWOM. Technological Forecasting and Social Change, 190, 122407.
Kaplan, B., & Duchon, D. (1988). Combining qualitative and quantitative methods in information
systems research: a case study. MIS Quarterly, 571-586.
Kaplan, B., & Maxwell, J. A. (2005). Qualitative research methods for evaluating computer information
systems. In Evaluating the organisational impact of healthcare information systems (pp. 30-55).
Keith, M. G., & McKay, A. S. (2024). Too Anecdotal to Be True? Mechanical Turk Is Not All Bots
and Bad Data: Response to Webb and Tangney (2022). Perspectives on Psychological Science,
17456916241234328.
Kelly, S., Kaye, S. A., & Oviedo-Trespalacios, O. (2022). What factors contribute to the acceptance of
artificial intelligence? A systematic review. Telematics and Informatics, 101925.
Kelly, A. E., & Palaniappan, S. (2023). Using a technology acceptance model to determine factors
influencing continued usage of mobile money service transactions in Ghana. Journal of Innovation and
Entrepreneurship, 12(1), 34.
Kerrigan, M. R. (2014). A framework for understanding community collegesorganisationalal capacity
for data use: A convergent parallel mixed methods study. Journal of mixed methods research, 8(4), 341-
362.
Kerlyl, A., Hall, P., & Bull, S. (2006). Bringing chatbots into education: Towards natural language
negotiation of open learner models. In International conference on innovative techniques and
applications of artificial intelligence (pp. 179-192).
Kelley, K., Clark, B., Brown, V., & Sitzia, J. (2003). Good practice in the conduct and reporting of
survey research. International Journal for Quality in Health Care, 15(3), 261-266.

180
Khan, S., & Rabbani, M. R. (2021). Artificial intelligence and NLP-based chatbot for Islamic banking
and finance. International Journal of Information Retrieval Research (IJIRR), 11(3), 65-77.
Khanpour, H., Guntakandla, N., & Nielsen, R. (2016). Dialogue act classification in domain-
independent conversations using a deep recurrent neural network. In Proceedings of Coling 2016, the
26th International Conference on Computational Linguistics: Technical papers (pp. 2012-2021).
Khan, E., Sperotto, A., van der Ham, J., & van Rijswijk-Deij, R. (2023). Stranger VPNs: Investigating
the Geo-Unblocking Capabilities of Commercial VPN Providers. In International Conference on
Passive and Active Network Measurement (pp. 46-68). Cham: Springer Nature Switzerland.
Kiger, M. E., & Varpio, L. (2020). Thematic analysis of qualitative data: AMEE Guide No. 131. Medical
Teacher, 42(8), 846–854. https://doi.org/10.1080/0142159X.2020.1755030.
Kim, J., & Gambino, A. (2016). Do we trust the crowd or information system? Effects of personalisation
and bandwagon cues on users' attitudes and behavioural intentions toward a restaurant recommendation
website. Computers in Human Behavior, 65, 369-379.
Kim, M., Ko, H., Choi, J., & Lee, J. (2023). FlumeRide: Interactive Space Where Artists and Fans Meet-
and-Greet Using Video Calls. IEEE Access, 11, 31594-31606.
Kim, J., Merrill Jr., K., & Collins, C. (2021). AI as a friend or assistant: The mediating role of perceived
usefulness in social AI vs. functional AI. Telematics and Informatics, 64, 101694.
Kim, Y. J., Toh, S. M., and Baik, S. (2022). Culture creation and change: Making sense of the past to
inform future research agendas. Journal of Management.
Kim, M. S. (1994). Cross-cultural comparisons of the perceived importance of conversational
constraints. Human communication research, 21(1), 128-151.
Kim, M., & Chang, B. (2020). The effect of service quality on the reuse intention of a chatbot: Focusing
on user satisfaction, reliability, and Immersion. International Journal of Contents, 16(4), 1-15.
Klaus, P., & Zaichkowsky, J. L. (2022). The convenience of shopping via voice AI: Introducing
AIDM. Journal of Retailing and Consumer Services, 65, 102490.
Kline, R. B. (2023). Principles and practice of structural equation modelling. Guilford publications.
Kock, N. (2016). Hypothesis testing with confidence intervals and P values in PLS-SEM. International
Journal of e-Collaboration (IJeC), 12(3), 1-6.
Kocielnik, R., Amershi, S., & Bennett, P. N. (2019, May). Will you accept an imperfect AI? Exploring
designs for adjusting end-user expectations of AI systems. In Proceedings of the 2019 CHI Conference
on Human Factors in Computing Systems (pp. 1-14).
Kowatsch, T., Nißen, M., Shih, C. H. I., Rüegger, D., Volland, D., Filler, A., ... & Farpour-Lambert, N.
(2017). Text-based healthcare chatbots supporting patient and health professional teams: preliminary
results of a randomised controlled trial on childhood obesity. Persuasive Embodied Agents for Behavior
Change (PEACH2017).
Koć-Januchta, M. M., Schönborn, K. J., Roehrig, C., Chaudhri, V. K., Tibell, L. A., & Heller, H. C.
(2022). Connecting concepts helps put main ideas together”: cognitive load and usability in learning
biology with an AI-enriched textbook. International Journal of Educational Technology in Higher
Education, 19(1), 11.
Kocaballi, A. B., Berkovsky, S., Quiroz, J. C., Laranjo, L., Tong, H. L., Rezazadegan, D., ... & Coiera,
E. (2019a). The personalisation of conversational agents in health care: a systematic review. Journal of
medical Internet research, 21(11), e15360.
Kocaballi, A. B., Laranjo, L., & Coiera, E. (2019b). Understanding and measuring user experience in
conversational interfaces. Interacting with Computers, 31(2), 192-207.

181
Kull, A. J., Romero, M., & Monahan, L. (2021). How may I help you? Driving brand engagement
through the warmth of an initial chatbot message. Journal of Business Research, 135, 840-850.
Kumar, V., Rajan, B., Venkatesan, R., & Lecinski, J. (2019). Understanding the role of artificial
intelligence in personalised engagement marketing. California Management Review, 61(4), 135-155.
Kreutzer, R. T., & Sirrenberg, M. (2020). Understanding artificial intelligence. Berlin, Germany:
Springer International Publishing.
Kronemann, B., Kizgin, H., Rana, N., & K. Dwivedi, Y. (2023). How AI encourages consumers to share
their secrets? The role of anthropomorphism, personalisation, and privacy concerns and avenues for
future research. Spanish Journal of Marketing-ESIC, 27(1), 2-19.
Ladkoom, K., and Thanasopon, B. (2020). Factors Influencing Reuse Intention of e-Payment in
Thailand: A Case Study of PromptPay—paper presented at the ICEIS (1).
Laranjo, L., Dunn, A. G., Tong, H. L., Kocaballi, A. B., Chen, J., Bashir, R., ... & Coiera, E. (2018).
Conversational agents in healthcare: a systematic review. Journal of the American Medical Informatics
Association, 25(9), 1248-1258.
Laroche, M., Ueltschy, L. C., Abe, S., Cleveland, M., & Yannopoulos, P. P. (2004). Service quality
perceptions and customer satisfaction: evaluating the role of culture. Journal of international
marketing, 12(3), 58-85.
Lee, K. Y., Sheehan, L., Lee, K., & Chang, Y. (2021). The continuation and recommendation intention
of artificial intelligence-based voice assistant systems (AIVAS): the influence of personal traits. Internet
Research, 31(5), 1899-1939.
Lin, J. S. C., & Hsieh, P. L. (2007). The influence of technology readiness on satisfaction and behavioral
intentions toward self-service technologies. computers in Human Behavior, 23(3), 1597-1615.
Lee, H., Park, N., and Hwang, Y. (2015). A new dimension of the digital divide: Exploring the
relationship between broadband connection, smartphone use and communication competence.
Telematics and Informatics, 32(1), 45-56.
Lee, M., & Ko, E. (2019). The Effect of Customized Chatbot Services on Brand Loyalty toward Luxury
Brands. In 2019 Global Fashion Management Conference at Paris (pp. 828-828).
Lee, J. C., & Lin, R. (2023). The continuous usage of artificial intelligence (AI)-powered mobile fitness
applications: the goal-setting theory perspective. Industrial Management & Data Systems, 123(6), 1840-
1860.
Lewis, J. R., and Hardzinski, M. L. (2015). Investigating the psychometric properties of the Speech
User Interface Service Quality questionnaire. International Journal of Speech Technology, 18 (3), 479-
487.
Leidner, D. E., & Kayworth, T. (2006). A review of culture in information systems research: Toward a
theory of information technology culture conflict. MIS Quarterly, 357-399.
Li, M., & Wang, R. (2023). Chatbots in e-commerce: The effect of chatbot language style on customers’
continuance usage intention and attitude toward the brand. Journal of Retailing and Consumer
Services, 71, 103209.
Lim, V., Rooksby, M., & Cross, E. S. (2021). Social robots on a global stage: establishing a role for
culture during human–robot interaction. International Journal of Social Robotics, 13(6), 1307-1333.
Lai, P. C. (2017). The literature review of technology adoption models and theories for the novelty
technology. JISTEM-Journal of Information Systems and Technology Management, 14, 21-38.

182
Lai, C., Wang, Q., Li, X., & Hu, X. (2016). The influence of individual-espoused cultural values on
self-directed use of technology for language learning beyond the classroom. Computers in Human
Behavior, 62, 676-688.
Li, K. (2023). Determinants of College Students’ Actual Use of AI-Based Systems: An Extension of the
Technology Acceptance Model. Sustainability, 15(6), 5221.
Li, C., Pan, R., Xin, H., & Deng, Z. (2020). Research on artificial intelligence customer service on
consumer attitude and its impact during online shopping. In Journal of Physics: Conference Series (Vol.
1575, No. 1, p. 012192).
Liang, Y., Lee, S. H., & Workman, J. E. (2020). Implementation of artificial intelligence in fashion: Are
consumers ready? Clothing and Textiles Research Journal, 38(1), 3-18.
Liao, C., Palvia, P., & Chen, J. L. (2009). Information technology adoption behaviour life cycle: Toward
a Technology Continuance Theory (TCT). International Journal of Information Management, 29(4),
309-320.
Liao, H., Proctor, R. W., & Salvendy, G. (2008). Content preparation for cross-cultural e-commerce: a
review and a model. Behaviour & Information Technology, 27(1), 43-61.
Libai, B., Bart, Y., Gensler, S., Hofacker, C. F., Kaplan, A., Kötterheinrich, K., & Kroll, E. B. (2020). A
brave new world? On AI and the management of customer relationships. Journal of Interactive
Marketing, 51(1), 44-56.
Liew, T. W., Tan, S. M., Tee, J., and Goh, G. G. G. (2021). The effects of designing conversational
commerce chatbots with expertise cues. In 2021 14th International Conference on Human System
Interaction (HSI) (pp. 1-6). IEEE.
López, G., Quesada, L., & Guerrero, L. A. (2018). Alexa vs. Siri vs. Cortana vs. Google Assistant: A
comparison of speech-based natural user interfaces. In Advances in Human Factors and Systems
Interaction: Proceedings of the AHFE 2017 International Conference on Human Factors and Systems
Interaction, July 17− 21, 2017, The Westin Bonaventure Hotel, Los Angeles, California, USA 8 (pp.
241-250). Springer International Publishing.
Luo, X., Tong, S., Fang, Z., & Qu, Z. (2019). Frontiers: Machines vs. humans: The impact of artificial
intelligence chatbot disclosure on customer purchases. Marketing Science, 38(6), 937-947.
Lubbe, I., & Ngoma, N. (2021). Useful chatbot experience provides technological satisfaction: An
emerging market perspective. South African Journal of Information Management, 23(1), 1-8.
Lv, J., Jiang, X., & Jiang, A. (2022). Application of Virtual Reality Technology Based on Artificial
Intelligence in Sports Skill Training. Wireless Communications and Mobile Computing, 2022.
Mackenzie, N., & Knipe, S. (2006). Research dilemmas: Paradigms, methods and methodology. Issues
in educational research, 16(2), 193-205.
Madden, T. J., Ellen, P. S., & Ajzen, I. (1992). A comparison of the theory of planned behaviour and the
theory of reasoned action. Personality and Social Psychology Bulletin, 18(1), 3-9.
Maedche, A., Legner, C., Benlian, A., Berger, B., Gimpel, H., Hess, T., ... & Söllner, M. (2019). AI-
based digital assistants: Opportunities, threats, and research perspectives. Business & Information
Systems Engineering, 61, 535-544.
Malodia, S., Kaur, P., Ractham, P., Sakashita, M., & Dhir, A. (2022). Why do people avoid and postpone
the use of voice assistants for transactional purposes? A perspective from decision avoidance
theory. Journal of Business Research, 146, 605-618.

183
Mantelero, A. (2022). The social and ethical components in AI systems design and management. In
Beyond Data: Human Rights, Ethical and Social Impact Assessment in AI (pp. 93-137). The Hague:
TMC Asser Press.
Mariani, M. M., Hashemi, N., & Wirtz, J. (2023). Artificial intelligence empowered conversational
agents: A systematic literature review and research agenda. Journal of Business Research, 161, 113838.
Maier, D., Waldherr, A., Miltner, P., Jähnichen, P., & Pfetsch, B. (2018). Exploring issues in a networked
public sphere: Combining hyperlink network analysis and topic modelling. Social Science Computer
Review, 36(1), 3-20.
Malodia, S., Islam, N., Kaur, P., & Dhir, A. (2021). Why do people use artificial intelligence (AI)-
enabled voice assistants? IEEE Transactions on Engineering Management.
Marinkovic, V., and Kalinic, Z. (2017). Antecedents of customer satisfaction in mobile commerce:
Exploring the moderating effect of customisation. Online Information Review.

Martin, A., Nateqi, J., Gruarin, S., Munsch, N., Abdarahmane, I., Zobel, M., & Knapp, B. (2020). An
artificial intelligence-based first-line defence against COVID-19: digitally screening citizens for risks
via a chatbot. Scientific reports, 10(1), 19012.
Maryanto, R. H., & Kaihatu, T. S. (2021). Customer loyalty is an impact of perceived usefulness to grab
users, mediated by customer satisfaction and moderated by perceived ease of use. Binus Business
Review, 12(1), 31-39.
McLean, G., Osei-Frimpong, K., & Barhorst, J. (2021). Alexa, do voice assistants influence consumer
brand engagement?–Examining the role of AI-powered voice assistants in influencing consumer brand
engagement. Journal of Business Research, 124, 312-328.
Mekni, M. (2021). An artificial intelligence based virtual assistant using conversational
agents. Journal of Software Engineering and Applications, 14(9), 455-473.
Merhi, M. I. (2021). Multi-country analysis of e-commerce adoption: The impact of national culture
and economic development. Pacific Asia Journal of the Association for Information Systems, 13(3), 4.
Milhorat, P., Lala, D., Inoue, K., Zhao, T., Ishida, M., Takanashi, K., ... & Kawahara, T. (2019). A
conversational dialogue manager for the humanoid robot ERICA. In Advanced Social Interaction with
Agents: 8th International Workshop on Spoken Dialog Systems (pp. 119-131). Springer International
Publishing.
Mimoun, M. S. B., & Poncin, I. (2015). A valued agent: How ECAs affect website customers'
satisfaction and behaviours. Journal of Retailing and Consumer Services, 26, 70-82.
Miner, A. S., Laranjo, L., & Kocaballi, A. B. (2020). Chatbots in the fight against the COVID-19
pandemic. NPJ digital medicine, 3(1), 65.
Mikalef, P., & Gupta, M. (2021). Artificial intelligence capability: Conceptualization, measurement
calibration, and empirical study on its impact on organisational creativity and firm
performance. Information & Management, 58(3), 103434.
Mollick, J., Cutshall, R., Changchit, C., & Pham, L. (2023). Contemporary Mobile Commerce:
Determinants of Its Adoption. Journal of Theoretical and Applied Electronic Commerce
Research, 18(1), 501-523.
Mokhlis, S. (2009). Religious differences in some selected aspects of consumer behaviour: a Malaysian
study. The Journal of International Management Studies, 4(1), 67-76.
Morgan, D. L. (2014). Pragmatism as a paradigm for social research. Qualitative Inquiry, 20(8), 1045-
1053.

184
Moran-Ellis, J., Alexander, V. D., Cronin, A., Dickinson, M., Fielding, J., Sleney, J., & Thomas, H.
(2006). Triangulation and integration: processes, claims and implications. Qualitative research, 6(1),
45-59.
Moriuchi, E., Landers, V. M., Colton, D., & Hair, N. (2021). Engagement with chatbots versus
augmented reality interactive technology in e-commerce. Journal of Strategic Marketing, 29(5), 375-
389.
Mou, J., Cohen, J., Dou, Y., & Zhang, B. (2020). International buyers’ repurchase intentions in a Chinese
cross-border e-commerce platform: A valence framework perspective. Internet Research, 30(2), 403-
437.
Moon, J. W., & Kim, Y. G. (2001). Extending the TAM for a World-Wide-Web context. Information
& Management, 38 (4), 217–230.

Moon, K., & Blackman, D. (2014). A guide to understanding social science research for natural
scientists. Conservation biology, 28(5), 1167-1177.

Mukumbang, F. C. (2023). Retroductive theorizing: a contribution of critical realism to mixed methods

research. Journal of Mixed Methods Research, 17(1), 93-114.
Mullins, J. K., & Cronan, T. P. (2021). Enterprise systems knowledge, beliefs, and attitude: A model of
informed technology acceptance. International Journal of Information Management, 59, 102348.

Nam, K., Dutt, C. S., Chathoth, P., Daghfous, A., & Khan, M. S. (2021). The adoption of artificial
intelligence and robotics in the hotel industry: prospects and challenges. Electronic Markets, 31, 553-
574.
Naeini, F. H. (2012). Usage Pattern, Perceived Usefulness and Ease of Use of Computer Games among
Malaysian Elementary School Students. Research Journal of Applied Sciences, Engineering and
Technology, 4(23), 5285-5297.
Nagy, S., & Hajdú, N. (2021). Consumer acceptance of the use of artificial intelligence in online
shopping: Evidence from Hungary. Amfiteatru Economic, 23(56), 155-173.
Nicolescu, L., & Tudorache, M. T. (2022). Human-computer interaction in customer service: the
experience with AI chatbots—a systematic literature review. Electronics, 11(10), 1579.
Nikolenko, S. I., Koltcov, S., & Koltsova, O. (2017). Topic modelling for qualitative studies. Journal
of Information Science, 43(1), 88-102.
Nofirda, F. A., & Ikram, M. (2023, June). The Use of Artificial Intelligence on Indonesia Online
Shopping Application in Relation to Customer Acceptance. In Ninth Padang International Conference
on Economics Education, Economics, Business and Management, Accounting and Entrepreneurship
(PICEEBA 2022) (pp. 642-651).
Noreen, U., Shafique, A., Ahmed, Z., & Ashfaq, M. (2023). Banking 4.0: Artificial intelligence (AI) in
banking industry & consumer’s perspective. Sustainability, 15(4), 3682.
Nordheim, C. B., Følstad, A., & Bjørkli, C. A. (2019). An initial model of trust in chatbots for customer
service—findings from a questionnaire study. The Journal of Interacting with Computers, 31(3), 317-
335.

Norman, D. (2013). The Design of Everyday Things: Revised and Expanded Edition. Basic books.

Nuruzzaman, M., & Hussain, O. K. (2018). A survey on chatbot implementation in the customer service
industry through deep neural networks. In 2018 IEEE 15th International Conference on e-Business
Engineering (ICEBE) (pp. 54-61). IEEE.

185
Nunn, N. (2012). Culture and the historical process. Economic History of Developing Regions, 27(sup
1), 108-126.
Oh, K. J., Lee, D., Ko, B., & Choi, H. J. (2017). A chatbot for psychiatric counselling in mental
healthcare service based on emotional dialogue analysis and sentence generation. In 2017 18th IEEE
International Conference on Mobile Data Management (MDM) (pp. 371-375). IEEE.
Oguntosin, V., & Olomo, A. (2021). Development of an e-commerce chatbot for a university shopping
mall. Applied Computational Intelligence and Soft Computing, 2021, 1-14.
Ohk, K., Park, S.-B., and Hong, J.-W. (2015). The influence of perceived usefulness, perceived ease of
use, interactivity, and ease of navigation on satisfaction in mobile applications. Advanced Science and
Technology Letters, 84(2015), 88-92.
Oktavia, T., Agung, C. A., Thalib, D. I., & Rahmanda, Y. D. (2023). An Empirical Study on the Factors
Influencing the Attitude and Behaviors Towards Smartphone Voice Assistant Usage. Journal of System
and Management Sciences, 13(3), 481-504.
Okuda, T., & Shoda, S. (2018). AI-based chatbot service for the financial industry. Fujitsu Scientific
and Technical Journal, 54(2), 4-8.
Onofrei, G., Filieri, R., & Kennedy, L. (2022). Social media interactions, purchase intention, and
behavioural engagement: The mediating role of source and content factors. Journal of Business
Research, 142, 100-112.
Othman, S., Steen, M., & Fleet, J. (2020). A sequential explanatory mixed methods study design: An
example of how to integrate data in a midwifery research project. Journal of Nursing Education and
Practice, 11(2), 75-89.
Palmatier, R. W., Houston, M. B., & Hulland, J. (2018). Review articles: purpose, process, and
structure. Journal of the Academy of Marketing Science, 46, 1-5.
Pantano, E., & Di Pietro, L. (2012). Understanding consumer acceptance of technology-based
innovations in retailing. Journal of technology management & innovation, 7(4), 1-19.
Park, E. S., & Park, M. S. (2020). Factors of the technology acceptance model for construction
IT. Applied Sciences, 10(22), 8299.
Pereira, J., & Díaz, O. (2018). A quality analysis of Facebook Messenger’s most popular chatbots.
In Proceedings of the 33rd annual ACM symposium on applied computing (pp. 2144-2150).
Peña-García, N., Gil-Saura, I., Rodríguez-Orejuela, A., & Siqueira-Junior, J. R. (2020). Purchase
intention and purchase behaviour online: A cross-cultural approach. Heliyon, 6(6).
Petter, S. C., & Gallivan, M. J. (2004). Toward a framework for classifying and guiding mixed method
research in information systems. In 37th Annual Hawaii International Conference on System Sciences,
2004. Proceedings of the (pp. 10-pp). IEEE.
Perifanis, N. A., & Kitsios, F. (2023). Investigating the influence of artificial intelligence on business
value in the digital era of strategy: A literature review. Information, 14(2), 85.
Petersson, A. H., Pawar, S., & Fagerstrøm, A. (2023). Investigating the factors of customer experiences
using real-life text-based banking chatbot: a qualitative study in Norway. Procedia Computer Science,
219, 697-704.
Pei, Z., & Zhenxiang, Z. (2006). A framework for personalized service website based on TAM. In 2006
International Conference on Service Systems and Service Management (Vol. 2, pp. 1598-1603). IEEE.
Pitardi, V., & Marriott, H. R. (2021). Alexa, she's not human, but… Unveiling the drivers of consumers'
trust in voice‐based artificial intelligence. Psychology & Marketing, 38(4), 626-642.

186
Pillai, R., Sivathanu, B., & Dwivedi, Y. K. (2020). Shopping intention at AI-powered automated retail
stores (AIPARS). Journal of Retailing and Consumer Services, 57, 102207.

Pinsky, L. (1951). Do machines think about machines thinking? The Turing Test: Verbal Behavior as
the Hallmark of Intelligence, 143.

Polit, D. F., & Beck, C. T. (2010). Generalization in quantitative and qualitative research: Myths and
strategies. International journal of nursing studies, 47(11), 1451-1458.

Polkosky, M. D., & Lewis, J. R. (2003). Expanding the MOS: Development and psychometric
evaluation of the MOS-R and MOS-X. International Journal of Speech Technology, 6(2), 161-182.

Poncet, A., Courvoisier, D. S., Combescure, C., & Perneger, T. V. (2016). Normality and sample size
do not matter in selecting an appropriate statistical test for two-group comparisons. Methodology.

Prasad, V. A., & Ranjith, R. (2020). Intelligent chatbot for lab security and automation. In 2020, the
11th International Conference on Computing, Communication and Networking Technologies (ICT) (pp.
1-4). IEEE.

Purwanto, E., & Loisa, J. (2020). The intention and use behaviour of the mobile banking system in
Indonesia: UTAUT Model. Technology Reports of Kansai University, 62(06), 2757-2767.

Rana, M., & Arora, N. (2022). How does social media advertising persuade? An investigation of the
moderation effects of corporate reputation, privacy concerns and intrusiveness. Journal of Global
Marketing, 35(3), 248-267.
Rakhra, M., Gopinadh, G., Addepalli, N. S., Singh, G., Aliraja, S., Reddy, V. S. G., & Reddy, M. N.
(2021). E-commerce assistance with a smart chatbot using artificial intelligence. In 2021 2nd
International Conference on Intelligent Engineering and Management (ICIEM) (pp. 144-148). IEEE.
Radziwill, N. M., & Benton, M. C. (2017). Evaluating the quality of chatbots and intelligent
conversational agents. arXiv preprint arXiv:1704.04579.
Ramayah, T., Yeap, J. A., Ahmad, N. H., Halim, H. A., & Rahman, S. A. (2017). Testing a confirmatory
model of Facebook usage in SmartPLS using consistent PLS. International Journal of Business and
Innovation, 3(2), 1-14.
Rawlings, C. M. (2020). Cognitive authority and the constraint of attitude change in groups. American
Sociological Review, 85(6), 992-1021.
Ranoliya, B. R., Raghuwanshi, N., & Singh, S. (2017). Chatbot for university-related FAQs. In 2017
International Conference on Advances in Computing, Communications, and Informatics (ICACCI) (pp.
1525-1530). IEEE.
Rese, A., Ganster, L., & Baier, D. (2020). Chatbots in retailers’ customer communication: How to
measure their acceptance. Journal of Retailing and Consumer Services, 56, 102176.
Rezaei, S., & Valaei, N. (2017). Branding in a multichannel retail environment: Online stores vs app
stores and the effect of product type. Information Technology & People, 30(4), 853-886.
Rhee, C. E., & Choi, J. (2020). Effects of personalisation and social role in voice shopping: An
experimental study on product recommendation by a conversational voice agent. Computers in Human
Behavior, 109, 106359.
Rindfleisch, A., Malter, A. J., Ganesan, S., & Moorman, C. (2008). Cross-sectional versus longitudinal
survey research: Concepts, findings, and guidelines. Journal of marketing research, 45(3), 261-279.

187
Ringle, C. M., Sarstedt, M., Mitchell, R., & Gudergan, S. P. (2020). Partial least squares structural
equation modelling in HRM research. The International Journal of Human Resource
Management, 31(12), 1617-1643.
Riquelme, L. F., and Rosas, J. (2014). Multicultural perspectives: The road to cultural
competence. Language development: Foundations, processes, and clinical applications, 255-256.
Roy, R., & Naidoo, V. (2021). Enhancing chatbot effectiveness: The role of anthropomorphic
conversational styles and time orientation. Journal of Business Research, 126, 23-34.
Roh, T., Park, B. I., & Xiao, S. S. (2023). Adoption of AI-enabled Robo-advisors in Fintech:
Simultaneous Employment of UTAUT and the Theory of Reasoned Action. Journal of Electronic
Commerce Research, 24(1), 29-47.
Rodgers, W., Murray, J. M., Stefanidis, A., Degbey, W. Y., & Tarba, S. Y. (2023). An artificial
intelligence algorithmic approach to ethical decision-making in human resource management
processes. Human Resource Management Review, 33(1), 100925.
Ruan, Y., & Mezei, J. (2022). When do AI chatbots lead to higher customer satisfaction than human
frontline employees in online shopping assistance? Considering product attribute type. Journal of
Retailing and Consumer Services, 68, 103059.
Saniuk, S., Grabowska, S., & Gajdzik, B. (2020). Personalisation of products in the Industry 4.0 concept
and its impact on achieving a higher level of sustainable consumption. Energies, 13(22), 5895.
Sakshaug, J. W., Hülle, S., Schmucker, A., & Liebig, S. (2017, August). Exploring the effects of
interviewer-and self-administered survey modes on record linkage consent rates and bias. In Survey
Research Methods (Vol. 11, No. 2, pp. 171-188).
Sarkar, S., Chauhan, S., & Khare, A. (2020). A meta-analysis of antecedents and consequences of trust
in mobile commerce. International Journal of Information Management, 50, 286-301.
Sarica, S., & Luo, J. (2021). Stopwords in technical language processing. Plos one, 16(8), e0254937.
Said, M., Ramayah, T., & Al Salihi, S. M. R. (2022). Modelling Chatbots Adoption for Online Shopping
Amidst the Covid-19 Pandemic. Global Business and Management Research, 14(4s), 329-338.
Sánchez-Díaz, X., Ayala-Bastidas, G., Fonseca-Ortiz, P., & Garrido, L. (2018). A knowledge-based
methodology for building a conversational chatbot as an intelligent tutor. In Advances in Computational
Intelligence: 17th Mexican International Conference on Artificial Intelligence, MICAI 2018,
Guadalajara, Mexico, October 22–27, 2018, Proceedings, Part II 17 (pp. 165-175). Springer
International Publishing.
Song, R., Moon, S., Chen, H., & Houston, M. B. (2018). When marketing strategy meets culture: The
role of culture in product evaluations. Journal of the Academy of Marketing Science, 46, 384-402.
Straub, D., Loch, K., Evaristo, R., Karahanna, E., & Srite, M. (2002). Toward a theory-based
measurement of culture. Journal of Global Information Management (JGIM), 10(1), 13-23.
Sano, A. V. D., Imanuel, T. D., Calista, M. I., Nindito, H., & Condrobimo, A. R. (2018, September).
The application of AGNES algorithm to optimise knowledge base for tourism chatbot. In 2018
International Conference on Information Management and Technology (ICIMTech) (pp. 65-68). IEEE.
Saunders, M., Lewis, P., & Thornhill, A. (2009). Research methods for business students. Pearson
education.
Sciarelli, M., Prisco, A., Gheith, M. H., & Muto, V. (2022). Factors affecting the adoption of blockchain
technology in innovative Italian companies: an extended TAM approach. Journal of Strategy and
Management, 15(3), 495-507.

188
Schanke, S., Burtch, G., & Ray, G. (2021). Estimating the impact of “humanizing” customer service
chatbots. Information Systems Research, 32(3), 736-751.
Selamat, M. A., & Windasari, N. A. (2021). Chatbot for SMEs: Integrating customer and business owner
perspectives. Technology in Society, 66, 101685.
Serban, I. V., Lowe, R., Henderson, P., Charlin, L., & Pineau, J. (2015). A survey of available corpora
for building data-driven dialogue systems. arXiv preprint arXiv:1512.05742.
Serrano-Malebran, J., & Arenas-Gaitan, J. (2021). When does personalisation work on social media? A
posteriori segmentation of consumers. Multimedia Tools and Applications, 80, 36509-36528.
Setiaji, B., & Wibowo, F. W. (2016). Chatbot using knowledge in database: human-to-machine
conversation modelling. In 2016, 7th International Conference on Intelligent Systems, Modelling, and
Simulation (ISMS) (pp. 72-77). IEEE.
Septianto, F., Kemper, J. A., & Choi, J. J. (2020). The power of beauty? The interactive effects of awe
and online reviews on purchase intentions. Journal of Retailing and Consumer Services, 54, 102066.
Sfenrianto, S., & Vivensius, G. (2020). Analysis of factors influencing customer experience of e-
commerce users in Indonesia through the application of Chatbot technology. J. Theor. Appl. Inf.
Technol, 98(7), 953-962.
Sharma, A., & Jhamb, D. (2020). Changing consumer behaviours towards online shopping-an impact
of COVID-19. Academy of Marketing Studies Journal, 24(3), 1-10.
Shannon-Baker, P. (2016). Making paradigms meaningful in mixed methods research. Journal of mixed
methods research, 10(4), 319-334.
Shiau, W. L., Yuan, Y., Pu, X., Ray, S., & Chen, C. C. (2020). Understanding fintech continuance:
perspectives from self-efficacy and ECT-IS theories. Industrial Management & Data Systems, 120(9),
1659-1689.
Shin, D. H. (2010). Analysis of online social networks: A cross‐national study. Online Information
Review, 34(3), 473-495.
Shneiderman, B. (2020). Bridging the gap between ethics and practice: guidelines for reliable, safe, and
trustworthy human-centred AI systems. ACM Transactions on Interactive Intelligent Systems
(TiiS), 10(4), 1-31.
Sheehan, B., Jin, H. S., and Gottlieb, U. (2020). Customer service chatbots: Anthropomorphism and
adoption. Journal of Business Research, 115, 14-24.

Sidi, F., Panahy, P. H. S., Affendey, L. S., Jabar, M. A., Ibrahim, H., & Mustapha, A. (2012, March).
Data quality: A survey of data quality dimensions. In 2012 International Conference on Information
Retrieval & Knowledge Management (pp. 300-304). IEEE.

Simpson, S. H. (2015). Creating a data analysis plan: What to consider when choosing statistics for a
study. The Canadian journal of hospital pharmacy, 68(4), 311.

Singh, C., Dash, M. K., Sahu, R., & Kumar, A. (2024). Investigating the acceptance intentions of online
shopping assistants in E-commerce interactions: Mediating role of trust and effects of consumer
demographics. Heliyon, 10(3).
Sindermann, C., Yang, H., Elhai, J. D., Yang, S., Quan, L., Li, M., & Montag, C. (2022). Acceptance
and fear of Artificial Intelligence: Associations with personality in a German and a Chinese sample.
Discover Psychology, 2(1), 8.

189
Siebra, C., Correia, W., Penha, M., Macedo, J., Quintino, J., Anjos, M., ... & Santos, A. L. (2018). Virtual
assistants for mobile interaction: A review from the accessibility perspective. In Proceedings of the 30th
Australian Conference on Computer-Human Interaction (pp. 568-571).
Srivastava, M., & Kaul, D. (2016). Exploring the link between customer experience–loyalty–consumer
spend. Journal of Retailing and Consumer Services, 31, 277-286.
Srite, M., & Karahanna, E. (2006). The role of espoused national cultural values in technology
acceptance. MIS Quarterly, 679-704.
Skjerve, H., Braaum, L. E., Goth, U. S., & Sørensen, A. (2023). Using Simulations to Help Public
Health Students Overcome Language Barriers for Better Health Outcomes. International Journal of
Environmental Research and Public Health, 20(13), 6259.
Sormin, E., Julianti, K., Nadeak, B., & Naibaho, L. (2019). Use of construction inquiry learning model
to improve the interest of learning students grade XI SMA Angkasa 2 in colloid materials. PEOPLE
International Journal of Social Sciences, 5(2), 908-917.
Sobti, N. (2019). Impact of demonetisation on the diffusion of mobile payment service in India:
Antecedents of behavioural intention and adoption using extended UTAUT model. Journal of Advances
in Management Research, 16(4), 472-497.
Song, X., Yang, S., Huang, Z., & Huang, T. (2019). The application of artificial intelligence in electronic
commerce. In Journal of Physics: Conference Series (Vol. 1302, No. 3, p. 032030). IOP Publishing.
Srinivasan, S. S., Anderson, R., and Ponnavolu, K. (2002). Customer loyalty in e-commerce: an
exploration of its antecedents and consequences. Journal of Retailing, 78 (1), 41-50.
Subedi, D. (2016). Explanatory sequential mixed method design as the third research community of
knowledge claim. American Journal of Educational Research, 4(7), 570-577.
Sudaryanto, M. R., Hendrawan, M. A., & Andrian, T. (2023). The Effect of Technology Readiness,
Digital Competence, Perceived Usefulness, and Ease of Use on Accounting Students Artificial
Intelligence Technology Adoption. In E3S Web of Conferences (Vol. 388). EDP Sciences.
Suhel, S. F., Shukla, V. K., Vyas, S., & Mishra, V. P. (2020, June). Conversation to automation in banking
through chatbot using artificial machine intelligence language. In 2020 8th International Conference
on Reliability, Infocom Technologies and Optimisation (trends and future directions) (ICRITO) (pp.
611-618). IEEE.
Suresh, T. P., Yong, P. L., Chyi, Y. S., & Musa, R. (2023). Connecting with Generation Z: Consumer
Acceptance of the Use of Artificial Intelligence in Online Shopping. Journal of Entrepreneurship and
Business (JEB), 11(1), 56-68.
Sutherland, I., & Kiatkawsin, K. (2020). Determinants of the guest experience in Airbnb: a topic
modelling approach using LDA. Sustainability, 12(8), 3402.
Svikhnushina, E., Placinta, A., & Pu, P. (2021). User expectations of conversational chatbots based on
online reviews. In Designing Interactive Systems Conference 2021 (pp. 1481-1491).
Synnott, J., Harkin, M., Horgan, B., McKeown, A., Hamilton, D., McAllister, D., ... & Nugent, C.
(2020). The digital skills, experiences and attitudes of the Northern Ireland social care workforce toward
technology for learning and development: Survey study. JMIR medical education, 6(2), e15936.
Syarova, L. (2022). Chatbot usage in e-retailing and the effect on customer satisfaction.
Taber, K. S. (2018). The use of Cronbach’s alpha when developing and reporting research instruments
in science education. Research in science education, 48, 1273-1296.

190
Tabet, S., Bhogaraju, P., & Ash, D. (2000). Using XML as a Language Interface for AI Applications.
In Pacific Rim International Conference on Artificial Intelligence (pp. 103-110). Berlin, Heidelberg:
Springer Berlin Heidelberg.
Tahar, A., Riyadh, H. A., Sofyani, H., & Purnomo, W. E. (2020). Perceived ease of use, perceived
usefulness, perceived security and intention to use e-filing: The role of technology readiness. The
Journal of Asian Finance, Economics and Business (JAFEB), 7(9), 537-547.
Tan, P. K., & Lim, C. M. (2023). Factors That Affect User Satisfaction of Using E-Commerce Chatbot:
A Study on Generation Z. International Journal of Business and Technology Management, 5(1), 292-
303.
Tandi Lwoga, E., & Questier, F. (2014). Faculty adoption and usage behaviour of open access scholarly
communication in health science universities. New Library World, 115(3/4), 116-139.
Tang, W. P., Chan, C. W., So, W. K., & Leung, D. Y. (2014). Web-based interventions for caregivers of
cancer patients: a review of the literature. Asia-Pacific journal of oncology nursing, 1(1), 9-15.
Teo, T., & Huang, F. (2019). Investigating the influence of individually espoused cultural values on
teachers’ intentions to use educational technologies in Chinese universities. Interactive Learning
Environments, 27 (5-6), pp. 813-829.
Teo, T., Lee, C. B., Chai, C. S., & Wong, S. L. (2009). Assessing the intention to use technology among
pre-service teachers in Singapore and Malaysia: A multigroup invariance analysis of the Technology
Acceptance Model (TAM). Computers & Education, 53(3), 1000-1009.
Teodorescu, D., Aivaz, K. A., Vancea, D. P. C., Condrea, E., Dragan, C., & Olteanu, A. C. (2023).
Consumer Trust in AI Algorithms Used in E-Commerce: A Case Study of College Students at a
Romanian Public University. Sustainability, 15(15), 11925.
Toader, D. C., Boca, G., Toader, R., Măcelaru, M., Toader, C., Ighian, D., & Rădulescu, A. T. (2019).
The effect of social presence and chatbot errors on trust. Sustainability, 12(1), 256.
Triandis, H. C. (2015). Raised in a collectivist culture, one may become an individualist. In Working at
the Interface of Cultures (pp. 38-46). Routledge.
Tsai, W. H. S., Liu, Y., & Chuan, C. H. (2021). How chatbots' social presence communication enhances
consumer engagement: the mediating role of parasocial interaction and dialogue. Journal of Research
in Interactive Marketing, 15(3), 460-482.
Uprichard, E., & Dawney, L. (2019). Data diffraction: Challenging data integration in mixed methods
research. Journal of mixed methods research, 13(1), 19-32.
Uzir, M. U. H., Bukari, Z., Al Halbusi, H., Hock, R. L. T., Wahab, S. N., Rasul, T., ... & Eneizan, B.
(2023). Applied artificial intelligence: Acceptance-intention-purchase and satisfaction on smartwatch
usage in a Ghanaian context. Heliyon.
Uysal, E., Alavi, S., & Bezençon, V. (2022). Trojan horse or useful helper? A relationship perspective
on artificial intelligence assistants with humanlike features. Journal of the Academy of Marketing
Science, 50(6), 1153-1175.
Vaidyam, A. N., Wisniewski, H., Halamka, J. D., Kashavan, M. S., & Torous, J. B. (2019). Chatbots
and conversational agents in mental health: a review of the psychiatric landscape. The Canadian
Journal of Psychiatry, 64(7), 456-464.
Van den Broeck, E., Zarouali, B., & Poels, K. (2019). Chatbot advertising effectiveness: When does the
message get through? Computers in Human Behavior, 98, 150-157.
Venkatesh, V., Speier-Pero, C., & Schuetz, S. (2022). Why do people shop online? A comprehensive
framework of consumers’ online shopping intentions and behaviours. Information Technology &
People, 35(5), 1590-1620.

191
Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information
technology: Toward a unified view. MIS Quarterly, 27(3), 425–478.
Venkatesh, V., Brown, S. A., & Bala, H. (2013). Bridging the qualitative-quantitative divide: Guidelines
for conducting mixed methods research in information systems. MIS Quarterly, 21-54.
Wang, C., Ahmad, S. F., Ayassrah, A. Y. B. A., Awwad, E. M., Irshad, M., Ali, Y. A., ... & Han, H.
(2023a). An empirical evaluation of technology acceptance model for Artificial Intelligence in E-
commerce. Heliyon.
Wang, S., Menon, S., Long, T., Henderson, K., Li, D., Crowston, K., ... & Chilton, L. B. (2023b).
ReelFramer: Co-creating News Reels on social media with Generative AI. arXiv preprint
arXiv:2304.09653.
Weizenbaum, J. (1966). ELIZA—a computer program for the study of natural language communication
between man and machine. Communications of the ACM, 9(1), 36-45.
Whitley, E., & Ball, J. (2002). Statistics review 4: sample size calculations. Critical care, 6 (4), 1–7.

Wilkinson, D., Alkan, Ö., Liao, Q. V., Mattetti, M., Vejsbjerg, I., Knijnenburg, B. P., & Daly, E. (2021).
Why or why not? The effect of justification styles on chatbot recommendations. ACM Transactions on
Information Systems (TOIS), 39(4), 1-21.
Woolley, C. M. (2009). Meeting the mixed methods challenge of integration in a sociological study of
structure and agency. Journal of Mixed Methods Research, 3(1), 7-25.
Wu, F. C., Hong, O. N.-J., Trappey, A. J., & Trappey, C. V. (2020). VR-enabled chatbot system
supporting transformer mass-customization services. Paper presented at the 27th ISTE International
Conference on Transdisciplinary Engineering, TE 2020.

Xie, Q., Tan, D., Zhu, T., Zhang, Q., Xiao, S., Wang, J., ... & Yi, P. (2019). Chatbot application on
cryptocurrency. In 2019 IEEE Conference on Computational Intelligence for Financial Engineering &
Economics (CIFEr) (pp. 1-8).
Xiao, L., Guo, F., Yu, F., & Liu, S. (2019). The effects of online shopping context cues on consumers’
purchase intention for cross-border E-Commerce sustainability. Sustainability, 11(10), 2777.
Xie, K. L., Zhang, Z., & Zhang, Z. (2014). The business value of online consumer reviews and
management response to hotel performance. International Journal of Hospitality Management, 43, 1-
12.
Xiong, Y., Shi, Y., Pu, Q., & Liu, N. (2024). More trust or more risk? User acceptance of artificial
intelligence virtual assistant. Human Factors and Ergonomics in Manufacturing & Service
Industries, 34(3), 190-205.
Xiong, Y. (2022). The impact of artificial intelligence and digital economy consumer online shopping
behaviour on market changes. Discrete Dynamics in Nature and Society, 2022.
Yamaguchi, H., Mozgovoy, M., & Danielewicz-Betz, A. (2018). A Chatbot Based On AIML Rules
Extracted From Twitter Dialogues. In FedCSIS (Communication Papers) (pp. 37-42).
Yang, B., Liu, Y., Liang, Y., & Tang, M. (2019). Exploiting user experience from online customer
reviews for product design. International Journal of Information Management, 46, 173-186.
Yazici, B., & Yolacan, S. (2007). A comparison of various tests of normality. Journal of Statistical
Computation and Simulation, 77(2), 175-183.
Yen, Y. S. (2023). Channel integration affects usage intention in food delivery platform services: the
mediating effect of perceived value. Asia Pacific Journal of Marketing and Logistics, 35(1), 54-73.

192
Yin, J., & Qiu, X. (2021). AI technology and online purchase intention: Structural equation model based
on perceived value. Sustainability, 13(10), 5671.
Yi, M. Y., Fiedler, K. D., & Park, J. S. (2006). Understanding the role of individual innovativeness in
the acceptance of IT‐based innovations: Comparative analyses of models and measures. Decision
sciences, 37(3), 393-426.
Youn, S. Y., & Lee, K. H. (2019). Proposing value-based technology acceptance model: Testing on paid
mobile media service. Fashion and textiles, 6(1), 1-16.
Zaiţ, A., & Bertea, P. S. P. E. (2011). Methods for testing discriminant validity. Management &
Marketing Journal, 9(2), 217-224.

Zarouali, B., Van den Broeck, E., Walrave, M., & Poels, K. (2018). Predicting consumer responses to
a chatbot on Facebook. Cyberpsychology, Behavior, and Social Networking, 21(8), 491-497.

Zhang, S., Meng, Z., Chen, B., Yang, X., & Zhao, X. (2021). Motivation, Social Emotion, and the
Acceptance of Artificial Intelligence Virtual Assistants—Trust-Based Mediating Effects. Frontiers in
Psychology, 12, 728495.

Zhang, D., Pee, L. G., & Cui, L. (2021). Artificial intelligence in E-commerce fulfillment: A case study
of resource orchestration at Alibaba’s Smart Warehouse. International Journal of Information
Management, 57, 102304.

Zhang, K. Z., Zhao, S. J., Cheung, C. M., & Lee, M. K. (2014). Examining the influence of online
reviews on consumers' decision-making: A heuristic–systematic model. Decision Support Systems, 67,
78-89.

Zierau, N., Hausch, M., Bruhin, O., & Söllner, M. (2020). Towards Developing Trust-Supporting
Design Features for AI-Based Chatbots in Customer Service. In ICIS (Vol. 2020, pp. 1-9).

Zimmermann, R., Mora, D., Cirqueira, D., Helfert, M., Bezbradica, M., Werth, D., ... & Auinger, A.
(2023). Enhancing brick-and-mortar store shopping experience with an augmented reality shopping
assistant application using personalised recommendations and explainable artificial
intelligence. Journal of Research in Interactive Marketing, 17(2), 273-298.

Zhou, L., Dai, L., & Zhang, D. (2007). Online shopping acceptance model critical survey of consumer
factors in online shopping. Journal of Electronic Commerce Research, 8(1).

193
Appendices

Appendix A. Factors Questions of the Survey Questionnaire

194
195
Appendix B. AI Assistant Interaction Simulations

196