International Journal of Scientific Research in Computer Science, Engineering and Information Technology

ISSN : 2456-3307 (www.ijsrcseit.com)

doi : https://doi.org/10.32628/CSEIT20667
Artificial Intelligence Chatbot Adoption Framework for Real-Time Customer
Care Support in Kenya
Geoffrey Nyongesa, 2Kelvin Omieno, 3Daniel Otanga
1

1,3Department of Information Technology, Masinde Muliro University of Science and Technology, Kakamega Kenya

2Department of Information Technology and Informatics, Kaimosi Friends University College, Kaimosi, Kenya

Correspondence mail: [email protected]

ABSTRACT

Article Info In today’s society, most if not all sectors digitize and automate in order to
Volume 6, Issue 6 become more efficient. The increasing availability and sophistication of
Page Number: 100-117 software technologies disrupt labor markets by making workers redundant.
Publication Issue : Within this context, a significant change is companies’ introduction of chatbots
November-December-2020 as a supplement to human customer support is vital. Chatbots are computer
programs that interact with humans through natural language. The purpose of
chatbots is to simulate a human conversation in response to natural language
input through text or voice. There seems to be no proper guidelines for
adoption of chatbots for provision of customer care support services in
telecommunication industry in Kenya. The aim of this research was to develop
a framework for adoption of artificially intelligent chatbot application in
telecommunication industry. This was achieved through determination of the
status of implementation of chatbots in Kenya and identification of key metrics
that served as indicators for chatbot adoption. The metrics were identified
through review of previous technology adoption frameworks and models. The
study adopted mixed methods where qualitative and quantitative data was
collected using interview schedules and questionnaires respectively. Content
analysis was used in analyzing qualitative data. Quantitative data was analyzed
descriptively and results were presented using tables. The target population
included experts in the field of AI in two telecommunication firms in Kenya.
The study sample was drawn, using Delphi technique, from the two
telecommunication firms. The descriptive and principal component analysis
were utilized. The results of this research study will be crucial to all
Article History telecommunication firms in guiding them on the most effective and efficient
Accepted : 10 Nov 2020 way of adopting AI chatbot application for customer support services.
Published : 23 Nov 2020 Keywords : Artificial Intelligence, Chatbot, Customer Experience.

Copyright: © the author(s), publisher and licensee Technoscience Academy. This is an open-access article distributed
under the terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-
commercial use, distribution, and reproduction in any medium, provided the original work is properly cited
100
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

I. INTRODUCTION organization or brand [11]. This encompasses new

trends ranging from chatbots [13], powered by
Technology has changed how people, organizations advanced AI applications to deliver on-demand and
and firms function nowadays. This development will refined customer service and monitoring feedback
have extensive global impact on performance, from customers in real-time [8]. Furthermore,
resilience and competitiveness for industries. The customers are now performing their everyday
technologies in future at the convergence of activities with the support of technology [14]. This
intelligent machines, data and people will create includes; augmented reality systems in retailing,
opportunities for improvement based on new healthcare and education which enable users to
business models, transformed behavior and well- understand their environment in a completely new
informed decision making. These intelligent way. These trends sequentially translate into new
technologies including, artificial intelligence (AI), suggestions for both customers and managers. The
virtual reality (VR), augmented reality (AR), text rapid advancements in technology in today’s services
mining, Internet of Things (IoT), lead to enhanced demand relooking at science frameworks that are in
predictive capabilities, improved experiences, and existence, into lenses by which academia can assist
real time actionable insights. Whether we look at managers detect and find solutions to real-world
Business-to-Consumer (B2C) or Business-to-Business problems in real time while maintaining customer-
(B2B) settings, big data, information technology, centricity.
artificial intelligence, business intelligence &
analytics, new technology delivery platforms are However, researchers can assist customers self-detect
emerging [1][2] [3][4]. and find solutions to problems which will benefit
customers everyday lives. For scientists, these
In addition, there is increase in the use of service developments need questions to be asked on, how to
technologies which has led to increased research use new technology to develop business value,
attention due to problems in service marketing. measure and improve customer experience [15][16]
These research attentions are being done in and for questions to be re asked, for instance,
disciplines such as information systems and computer technology readiness [17], engagement and big data
science [5][6][7]. This shows the need for [4], quantifying e-service quality [18]. When data or
multidisciplinary service research [8] while information is not managed properly, it results in
preserving a vigorous service-theoretic focus. This is negative customer experience which leads to low
specially the case in service analytics [9], an area customer satisfaction. This deficiency takes place in
ranging from Business-to-Consumer Customer several situations because customer experience (CX)
Relationship Management metrics, to advanced is entire compendium of touch-points during an
Business-to-Business IoT metrics and information interaction between the customer and a brand, firm
technology infrastructure, both hardware and or business [19], which spans from advertising,
software supporting collection, measurement and customer care, trustworthiness and service or product
reporting of cloud-based software infrastructure [10]. features [20].
Service analytics are emerging with the rise in
artificially intelligent agents such as IBM Watson [12]. Therefore, it is important for industries to structure
Customer experience in the modern service context is customer experience (CX) across these various
concerned primarily with measurement and interactions with customers. Nonetheless, many
integration of customer interactions with industries fail to consider their services and products

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

101
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

as an all-round experience, some firms focus on social some limited level of adoption of chatbot in some
media and websites, others concentrate on face-to- organizations such as telecommunications, the
face or physical retail services or on-the-phone adoption process itself is haphazard. Therefore, there
customer service interaction [21]. Allen [22], did a is need for a well formulated guide to support full
research study to find out if both customers and adoption of chatbots. This paper presents artificial
businesses perceive same CX quality. It was intelligence chatbot adoption framework for real-
discovered that even-though 80% of firms considered time customer care support which serves as a guide
they offered a great CX, only eight percent of their for adoption of AI chatbot application in
clients expressed they got satisfied by experience they telecommunication industry.
received. The study recognized discrepancy in the
way customers and firms perceived the quality of CX. II. Basic concepts of artificial intelligence
Customer satisfaction and perceived quality of service
are two demonstrated success metrics for customer Computers and data science have been deployed to
service providers [23] [24]. However, how can we image recognition, process natural language, learn
measure customer satisfaction? and draw from past experience. The use of the
techniques of computer science to perform these
Kim [25] identified, value added services to enhance tasks is collectively known as Artificial Intelligence
convenience and enjoyment, and customer support, (AI). The application of AI today abounds:
to quickly process customer requests. These are two diagnostics in healthcare, translation of languages,
critical metrics for customer satisfaction. Customers self-driving cars, and in financial services for risk
making a phone call to firms for customer care profiling and portfolio management. Machine
service support results in low satisfaction, not learning is characterized by the little or lack of
mentioning long waits before talking to available human intervention with the enablement of
human agents [26][27]. GetHuman.com was created algorithms [28].
in 2007 to provide shortcuts for customers to reduce
long waits while calling customer care support 2.1 Machine Learning
numbers in big firms. Time Warner Cable Spectrum
Machine learning is a basic concept of artificial
(TWCS) had a bad reputation in customer service
intelligence since the field begun [29]. It is the
support. They decided to use online chat support that
application of AI which provides applications with
was available 24 seven to address poor customer
ability to learn and become better via computer
service issues. The effort and time customers require
algorithms [30] [31]. There are three types of
to find solutions to their problems and access to
machine learning algorithms: Unsupervised
information is a factor to be considered in successful
learning is defined as the ability of machines to get
CX.
patterns in an input stream without the need for
There is proliferation of technologies and there are
labelling the inputs by a human. Supervised
many emerging technologies. The demand for
learning consists of numerical regression and
prompt and frequent feedback and support for
classification. The human is required to label input
customers in various organizations gives a
data. Classification determines which category
competitive advantage to the organization. The
something belongs in, which occurs after an
alternative to human customer care support is
algorithm analyzes various examples of things (data)
embracing the Artificial intelligence systems such as
from several different categories. Regression results
Chatbot applications. Despite the fact that there is

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

102
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

in a function which describes causal relationships Modern statistical Natural Language Processing
between input and output, and predict how the approaches combine all these techniques and often
output data ought to change as input data changes attain a paragraph or page level accuracy, but they do
[31]. not have semantic understanding needed to
categorize isolated sentences in the right way. A part
Both regression and classifier learners are observed as
from problems in encoding semantic common-sense
function approximators, striving to learn unknown
knowledge, the available semantic Natural Language
functions. For instance, spam classifier is regarded as
Processing scales imperfectly to be practical in
a function which learns and maps texts of an email to
business enterprise applications. Above semantic
either spam or not spam. The theory of
NLP, the main aim of NLP is to consolidate an
computational learning assesses learners
understanding of common-sense reasoning [37]. The
via computational complexity by sample complexity
chatbot application adopted should read and
or other concepts of optimization [32].
understand English language. As an example, if a
With reinforcement learning [33], the agents are
human says he or she has a running nose and
rewarded for good responses and penalized for bad
smelling feet, the chatbot application should
responses. The agents use the sequence of reward and
understand just like human. The chatbot application
penalties to form strategies to operate in their
should not escalate conversation by saying, “The feet
problem space. With machine learning algorithms,
should run and the nose, smell.” Therefore, NLP can
AI chatbot application can learn from past interaction
be used for recognition.
with customers and customize responses to specific
individuals and services through prediction. 2.3 Natural Language Understanding (NLU)

2.2 Natural Language Processing (NLP) NLU is a part of NLP in AI that deals with the ability
to read text, process it and understand the meaning.
NLP [34], provides machines with ability to read
Natural Language Understanding is regarded as AI-
and comprehend human natural language. A
hard problem [38]. NLU involves disassembling of the
sufficient and powerful NLP application would
input, then parsing the input into data that can be
facilitate natural language user interfaces and
analyzed by the analysts or machine learning experts.
acquirement of knowledge from written sources, for
This part is more complex than the reverse process of
example, Newswire text. Applications of NLP
converting data to language because of the
include, mining of texts, answering questions,
unexpected occurrences of symbols or words or
information retrieval [35], and machine translation
phrases or unexpected features in any input. Human
[36]. Many present approaches use frequencies of
beings have a tendency to use abbreviations, or some
occurrences of words to construct syntactical
slang word that might not even be in the language.
representation of text. Some searching strategies like
The Natural Language Understanding algorithm
keyword spotting are common and scalable but are
should be able to tackle such difficulties and carry
not effective, for example, if one has a query to
out the conversion smoothly [38].
search for a, "dog", the search may only match
documents having the word "dog" but may miss out 2.4 Human Computer Interaction (HCI)
on the word, "poodle". Lexical affinity uses word
HCI deals with the interactions and relationships
occurrence like, "accident" to evaluate the
between computers and humans. HCI is more than
sentiments of documents.
user interfaces and “screen-deep'' [39]. Human
Computer interaction is a field covering different

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

103
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

areas [40]. In the early years of the history of HCI, seconds later. In addition, Cole [43] propose
the focus was on user interfaces, especially GUIs, including other nonspeech noises as well, for
using windows, menus, icons, mice, to develop usable example, filled pauses, laughter, inhalation and
applications. As user interface problems could be coughing. A more robust approach can be attained by
understood, the main HCI concerns shifted from using the prosody of one, his or her voice to
interface to address observations as expressed by D. determine the proper strategy [42]. To this end, the
Engelbart, who said that, “If the ease of use was the system must analyze the rate, pitch and amplitude of
only permitted criteria, people would have stuck to the voice. In addition to detecting the characteristic
tricycles, and not try bicycles.” Present HCI research behavior of a non-speech moment, this could be put
goals [41] deals with shared understanding, tasks, in use for responding appropriately to a certain
justifications, explanations, and not just interfaces. emotion like a robotic companion for the elderly or
chatbot application for customer care.
The new key challenges improve how people utilize
computers to think, work, communicate, critique, 2.6 Chatbot Technologies and Implementation.
learn, argue, explain, debate, decide, observe,
Chatbot applications are computer applications
calculate, design and simulate. The chatbot
which interact with people via natural language [44].
application should have a simple interface.
The main reason for humans to use chatbots is
2.5 Natural Language Processing as part of Human productivity, meaning quicker answer with less effort.
Computer Interaction Moreover, chatbots have been implemented with a
variety of reasons, such as provide information,
Developing a system that can both understand
emotional support, social support, entertainment or
natural language and respond appropriately is not
link users to other humans or machines [45].
enough if we pursue true HCI. The system will also
Customer service is a domain where chatbots have
need a strategy as to how it can sustain a fluid
achieved strong and growing interest [46]. The
conversation. It could harm the flow of the
renewed interest in chatbots is also partly driven by
conversation, for instance, if the responses would
developments within E-Commerce and E-service to
interrupt the collocutor or if he or she has to wait too
include Natural Language Interfaces [47]. In Norway,
long for the responses. Another vital part of the
there is a change in how customers are offered
dialog strategy is the acknowledgement of the
assistance. Chatbots are gradually becoming a regular
collocutor and the uttered sentences by giving an
function in customer service platforms in banks,
appropriate back-channel feedback. According to the
insurance, consulting and industry.
paper of Cantrell [42], the lack of this feedback can
cause a human collocutor to become unsure of
The humanlike conversation of chatbots gives
whether or not he or she is being understood. To
customers the opportunity to type questions, and in
enable a dialog strategy to function properly, it
return get meaningful answers to those questions in
should be implemented in an early stage. There, it
everyday language [48]. Chatbots can thereby be used
must analyze the flow of the utterances to detect any
to handle many routine queries which basically make
characteristic behavior.
up most service requests [46]. Furthermore, chatbots
Cantrell [42] suggest to focus on non-speech noise, never require vacation, get grumpy or tired.
for instant pauses. They note that the timing of the According to Brynjolfsson and McAfee (2017) [49],
feedback is not even that important and can occur the initial step of listening and understanding will be
between immediately after the utterance to a few the hardest part of automating customer service.

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

104
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

within 2025, including live telephone and online

Chatbots have also been proven useful in other conversations.
domains. Within the health domain, chatbots have
been evaluated favorably in comparison to search The current and predicted impact of chatbots and AI-
engines and information lines in answering applications in customer service illustrates the fast-
adolescent’s questions to alcohol, drugs and sex [48]. growing change that makes it essential to gain
A study in the educational domain found that knowledge about how chatbots are used and
students were overwhelmingly positive to use perceived by users. In spite of the early optimism
chatbots as a means for learning and practice of a concerning the launch of chatbot technologies by
foreign language [50]. Recently, the use of chatbots Facebook and Microsoft, theorists noted user
have been discussed in the domain of human adoption of chatbot applications to be less substantial
resources services, as a way of recruiting candidates than hoped [56]. The potential of chatbots have
for jobs, supposedly facilitating the recruitment seemingly not yet been realized as expected by the
process [51]. Although chatbots are the object of a technology industry. One reason might be that many
recent surge of interest, chatbot research and chatbots currently on the market have failed to fulfil
development dates back to the 1960’s [52]. The user’s needs due to a relatively high frequency of
renewed interest is largely attributed to two meaningless responses, unclear purposes or
developments. The developments within machine insufficient usability [57]. Lack of user focus by
learning and artificial intelligence (AI) made chatbot developers seems to be prominent.
applications easier to train and implement as a result
of strengthened capabilities in identifying user 2.7 Empirical studies
intents and sentiments, and improved Natural
Previous research studies focused mainly on web-
Language Processing [53].
based chatbot applications that were accessed
through a computer or a laptop. From the ’60s
Furthermore, chatbots have gained renewed interest
onwards, chatbot applications were originally used
as a result of changes in the popularity and
for entertainment using basic keyword match
availability of messaging platforms. This channel
techniques to respond to user inputs [58]. Since then,
allows businesses reach out to their target audience
research studies have been carried out on Natural
anytime anywhere via platform such as WhatsApp,
Language interfaces and texts, and various
Slack, Facebook Messenger or WeChat [54]. As they
architectures for chatbot applications have been
grow numerous on messaging platforms and
developed [58]. Research by Mott [59] states that
implemented as virtual assistants by some
chatbot software may be used in businesses enterprise
technological companies, the use of chatbots is slowly
to help in customer service, help desk, technical
becoming part of people’s daily lives [50]. Chatbot
support and guided selling. Their research study
technology has been incorporated into different
concentrated on application of web-based virtual
online environments like E-Commerce, delivery
agents and technical problems regarding the design
services and Daily News. Lately, there has been a
and deployment on massive scale. They stated that
substantial growth in developing chatbots for
conversational virtual agents need a good language
marketing and customer service [54]. Servion [55],
processing capability. For the agents to be deployed
recently predicted 95% of all customer support
on a large scale, they should be reliable, secure, and
service interactions to be handled by AI-applications
interoperable within existing IT infrastructure.

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

105
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

In addition to studies by Mott [58][59], they stated 2.8.1 Chatbot Implementation Globally.
that besides commerce, chatbot applications can be Recently, the industry and political institutions have
used for language learning, entertainment, and in expressed strong and growing interest in AI, with AI
education as also shown by Kerly [60], who studied research and application programs rapidly growing
the way chatbot applications may be used as a globally. The industry is focused on future uses of
negotiation in education as in aid for students. Artificial Intelligence. As stated by a report [68] that
Additionally, the virtual agents may be utilized in was provided by Venture Capital Corporation
healthcare industry as expressed in research study by Insights (VCCI) in United States in July of 2016,
Bickmore [61], who presented a virtual health Google, Intel, Microsoft, Apple, Twitter and other IT
counselor. Several research studies discussed giants have acquired approximately 140 financial
recommendation virtual agents [62]; programs to firms in Artificial Intelligence field since 2011.
assist in recommending consumers when making
decisions while shopping online. Shopping bots are In the first half of 2016, investments in AI exceeded
also well researched. They are bots designed to help that discovered in 2015, and about 200 artificial
customers in comparing and shopping products intelligence related firms raised above $1.4 billion in
online [63][64]. Atwell [58] reviewed the importance stock market. A significant number of mergers along
of chatbot application in multiple domains like with flood of capital, are expediting integration of
education and E-Commerce and inferred that chatbot artificial intelligence technologies in applications,
programs should not replace nor imitate a thus increasing the fast evolution of related economy.
conversation with humans. They believed that For instance, Google created an uproar after
chatbot application should be designed as a tool to purchasing the neural network firm called
help people. Using natural language should ease HCI. DNNresearch ultra-expensively in 2013 [69].

A recent research study [65], described an approach Deep learning is the latest technology in the industry
on implementing a web-based AI chatbot application and has enabled Google enhance precision of picture
which could operate as digital virtual assistant in searching. Deep learning is the main technology
scheduling meetings. However, the interaction linked with Unmanned Ground Vehicles (UGV) and
happened through e-mail messages. There are Google Glass [70]. Google has been developing
research studies [66][67], which proposed measures applications from mobile first toward AI first. The
to assess chatbot applications. As these AI integration of artificial intelligence with firms
applications expands to the messenger interfaces, we demands created significant shift in the mode of
need a different research technique to measure their service delivery. For instance, Xiaobing is a chatting
acceptance in telecommunication industry. robot developed by Microsoft, has been steering
transformation from traditional GUI to interactive
2.8 Analysis of Chatbot Implementation interfaces with emotional understanding and natural
Chatbot applications can be used to automate firm’s language [71].
internal business process and customer interactions.
These applications are powered by AI and use NLP In June of 2016, Microsoft obtained LinkedIn and
for communicating with humans either through reconstructed the Internet community with use of AI
voice or text. This subsection will discuss the status of technology. Additionally, the Watson application [72]
chatbot implementation globally, regionally and in which was developed by IBM is being used in
Kenya. hospitals in screening hundreds of millions of patient

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

106
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

records to find history of cancer treatment. It bank uses Ada, Stanbic IBTC uses Sami, Access bank
provides recommendations to diagnose leukemia and uses Tamara.
gives therapeutic schedules thus modifying the
paradigm of oncotherapy and clinical diagnosis. These chatbot applications are used in performing
simple daily banking operations like airtime purchase,
Nowadays, people connect and converse virtually via stock trading, bills payment, conversing with a
messengers and chatbot technology is the foundation customer and transfer of money via social network
of this connections. The leading artificial intelligence [76]. The Rwandan government, for example, started
powered chatbot applications include; Bold 360, using Chatbot Babyl, which is capable of relieving
LivePerson, Watson Assistant, Ada, Inbenta, Vergic medical doctors from dealing with simple medical
and Rulai.[73] In early days, chatbot applications inquiries and refer only advanced and serious
were developed to converse with people over the cases [77].
internet. This technology was limited to people who
wanted to connect and have social conversations. 2.8.3 Chatbot Implementation in Kenya
However, current trends show that majority of
There is increased uptake of chatbot applications
companies are now using AI chatbot when
locally, for instance, UBA Kenya’s Leo, Safaricom’s
communicating to their customers, respond to
Zuri, and Jubile Insurance’s Julie. These applications
queries and receive feedback [73]. The AI chatbot
perform several functions including digital sales,
responds like a real human due to combination of ML
transactional queries and customer support [78].
and predefined scripts.
Local adoption is slow due to little understanding of
how AI chatbot application can bring value to
Chatbot application responses are defined by the
business. Nowadays, consumers buy experiences
underlying software and access to knowledge
instead of products and services. AI chatbot can help
database. Apple’s Siri, Amazon’s Alexa, and
businesses offer personalized experiences because
Samsung’s Bixby, can provide information on latest
they make conversations interactive, relevant and
news and weather conditions [73]. E-commerce
engaging. Additionally, some farmers in Kenya
companies focus on using chatbot applications to
utilizes Arifu chatbot application to learn about
delight their CX. The adoption of chatbots globally
agronomy and finances. It offers free educational
has been rising as businesses recognize the usefulness
content from credible sources and organizations
of offering great CX which is cost effective and
about farming to farmers via SMS [79]. Some
personalized [74].
examples of chatbot applications under development
include; Bot Auditor, Insurance Bot, Business
2.8.2 Chatbot Implementation Regionally
Registration Bot, Book Slot Bot, Approval Work Bot
In Nigeria, AIICO Ella is a chatbot application for and Leave Management Bot [80].
insurer running on Facebook Messenger and web. It
III.Methodology
provides customer support with live human customer
care call center agents. The application helps
The study used mixed method to establish the
customers shop and purchase auto within seconds
relationship between AI chatbot adoption and
[75]. Rule-based chatbot applications are used in the
telecommunication industry. This design was applied
banking sector. The UBA bank uses LEO, Diamond
in previous research studies and revealed to be
effective and efficient in describing relationships

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

107
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

between different attributes. The research adopted 3.3 Sampling.

Delphi technique. In Delphi method, the researcher
reaches out to experts in the area of the study. This Saunders [81] defines a sampling design as the
method is useful in gaining an in-depth knowledge of technique that researchers use to select a subgroup
the issue being researched on. This research from the total population which will be involved in
employed descriptive survey in order to achieve the study. It acts as a guide to help the researcher
research study objectives. determine how the study samples will be selected
from the study population. Purposive sampling was
Descriptive survey was appropriate in determining applied in the study. It is non-random sampling
implementation status of AI chatbots in procedure where the members of target population
telecommunication industry. The focus of the study who meet a certain criterion are chosen, in this case,
was to identify key metrics that will eventually guide AI experts in the selected telecom firms were chosen.
the researcher develop a framework that could be The study involved physically visiting the selected
used as a guide for adoption of AI chatbot application telecommunication industries, explaining the purpose
in the telecom firms. The study employed expert of the visit to helpdesk team, who later connected the
opinion in validating the proposed adoption researcher to the IT team lead. After the IT team lead
framework. Expert opinion is an interpretation by understood the purpose of the study, he selected from
specialists or experts with experience in a particular IT team, AI experts who participated in the study.
area about a scientific evidence concerning that
specific topic and in evaluation of products. The IV.Data Analysis and Presentation
research study adopted this technique in order to
validate the developed research model. Table 4.1 Reliability Test

3.1 Location of the Study Variable Cronbach alpha

The research study was carried out in two Organizational Factors 0.871
telecommunication firms in Kenya which have a Technological Factors 0.761
higher percentage of market share. The firms are AI chatbot adoption 0.891
identified as firm’s X and Y for the purpose of this
study and in keeping anonymity of the firms. The findings showed the recommended Cronbach’s
Alpha above 0.70 was achieved for internal
3.2 Target Population consistency of data after testing all variables [82]. The
Target population is a whole group of individuals validity of data is extent by which the test measures
from which research study data is obtained and what it is expected to measure [83]. Mugenda [82]
conclusions made. The target population were describe validity as extent to which the research
experts in the field of artificial intelligence in two study findings attained from data analysis, represents
telecommunication firms in Kenya. For study the phenomenon under study. In agreement with
purposes, AI technologies included, NLP, Machine Table 4.1, Kaiser –Meyer -Olkin, the measure of
learning, chatbots and voice recognition. sampling sufficiently denoted KMO value above 0.5,
thus sample size was adequate to consider the data
sample as normally distributed. The KMO value that

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

108
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

is greater than 0.5, is sufficient to treat data as data from participants for all effective variables, was
distributed normally. identity matrix. The Bartlett’s Test was analyzed via
Chi-Square and presented as in Table 4.2. All
The Bartlett’s Test Sphericity tested null hypothesis variables were significant at 5% significance level,
item to item correlation matrix. Based on received signifying that the null hypothesis is declined.

Table 4.2 Test for Validity

Factors KMO test Bartlett’s test of sphericity

Chi-Square df Sig.
Organizational Factors .906 221.26 4 0.000
Technological Factors .907 340.74 4 0.003
AI Chatbot adoption .891 334.70 4 0.002

Extraction Method; Principal Component Analysis.

4.1 The Relationships

This strived to establish Correlation Analysis relationship between variables of research study, that is:
Organizational Factors, Technological Factors and AI Chatbot Adoption. The findings of Correlation Analysis
are indicated in Table 4.3.
Table 4.3 Correlation Analysis
Organizational Technological AI Chatbot
Factors Factors Adoption
Organizational Pearson Correlation 1
Factors Sig. (1-tailed)
N
Technological Pearson Correlation .634* 1
Factors Sig. (1-tailed) .001
N
AI Chatbot Pearson Correlation .306 .146* 1
Adoption Sig. (1-tailed) .095 .020

N 20 20 20
*. Correlation is significant at the 0.05 level (1-tailed).
The research study results in table 4.3 showed that a strong positive and significant relationship between
organizational and technological factors exist. This was portrayed by a Pearson Correlation Coefficient of, r =
0.634, p-value = 0.001 < 0.05, significant at 0.05 level of significance. This meant that improving on
organizational factors would lead to more comfortable and easy use of new technology in telecommunication
industry. The results in table 4.3 shows that there was a strong positive but non-significant relationship
between the organizational factors and the adoption of chatbot software in customer support services with a

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

109
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

Pearson Correlation Coefficient of, r =0.306, p-value = 0.095 > 0.05, not significant at 0.05 level of significance.
There was a non-strong positive but significant relationship between technological factors and adoption of
chatbot software with a Pearson-Correlation Coefficient of, r = 0.146, p-value = 0.02< 0.05, that is significant at
0.05 significance level. This implies that use of technology in an organization directly relates to the adoption
and use of chatbot software in customer support services.

4.2 The Regression Analysis

ANOVA test was used in determination of whether the developed model is essential in predicting the adoption
and usability of AI Chatbot software in telecommunication companies/organizations in Kenya. From the
analysis of findings in table 4.4, the value of R-Square is 0.613. This implies that, 61.3% of the variation of AI
Chatbot adoption in an organization was explained from the Organizational and Technological factors in the
telecommunication firms under the study.
Table 4.4 The Model Summary
Model Summary
Change Statistics
R Adjusted R Std. Error of the R Square F Sig. F
Model R Square Square Estimate Change Change df1 df2 Change
1 .783a .097
.009 1.32284 .613 .915 2 17 .003
.613
a. Predictors: (Constant), Technological factors and Organizational factors
b. Dependent Variable: AI Chatbot Adoption

At 0.05 significance level, ANOVA test showed that independent variables namely, Organizational factors and
Technological factors, were both predictors of the adoption of AI Chatbot software in the telecommunication
industry in Kenya in this model. In table 4.17, the ANOVA test analysis is shown by a p-value = 0.003, that is
less than 0.05 level of significance (p-value = 0.003 < 0.05).
Table 4.5: The ANOVA Table
Model Sum of Squares Df Mean Square F Sig.
1 Regression 3.202 2 1.601 .915 .003b
Residual 29.748 17 1.750
Total 32.950 19
a. Dependent Variable: AI Chatbot Adoption
b. Predictors: (Constant), Technological factors and Organizational factors
Table 4.6: The Regression Coefficient
Unstandardized Coefficients Standardized Coefficients
Model B Std. Error Beta T Sig.
1 (Constant) 5.022 2.290 2.192 .043
Organizational factors .783 .261 .356 1.195 .248
Technological factors .009 .033 .080 .268 .792

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

110
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

a. Dependent Variable: AI Chatbot Adoption

From the results in Table 4.6 above, at 0.05 significance level, the connection between the dependent and
independent variable is as shown in the equation below
𝑌 = 𝛽0 + 𝛽1 𝑋1 + 𝛽2 𝑋2 + 𝜀 ……………………………….…. (4.1)
Here, the Y value represent the AI Chatbot Adoption, X1 represent the organizational factors, X2 represent the
technological factors, and ε value stands for the error term in the model
Therefore, using the Regression Coefficient in Table 4.18, we get;
𝑌 = 5.022 + 0.783 ∗ 𝑋1 + 0.009 ∗ 𝑋2 + 𝜀 …………………...… (4.2)
The equation shown above, shows that an increase by one unit in organizational factor, the AI Chatbot
adoption in an organization will increase by 0.783. A unit increase in the technological factor will increase the
AI Chatbot adoption by 0.009.
4.3 Analysis of Intervening Effects
It was vital to investigate whether government regulation, need for automation and customer pressure
intervened AI chatbot adoption. To determine the effects of intervening variables, multiple R2 and Beta weights
were used. The Beta value should be greater than 0.1, and if the value is above 1, then it shows multi-
collinearity. The scale used were as follows;
i. A small effect is indicated by a Beta value between 0.1 and 0.2.
ii. A medium effect is indicated by a better value between 0.3 and 0.5.
iii. Values above 0.5 indicates a large effect.
iv. Values less than 0.1 indicates no effect.

OF = Organizational factors
TF = Technological factors
GR = Government regulation.
NA = Need for automation
CP = Customer pressure

Table 4.7 : Intervening Effects

R2 Beta Significance
OF + GR + NA 0.025 0.502 0.285
OF + GR + CP 0.084 0.384 0.012
OF + CP + NA 0.029 0.356 0.328
TF + GR + NA 0.006 0.521 0.478
TF + GR + CP 0.013 0.121 0.903
TF + CP + NA 0.017 0.016 0.129

The results in table 4.7 indicated how the different Beta value of 0.502. Customer pressure and
intervening factors such as government regulations, government regulation had a medium effect on
need for automation and customer pressure affect organizational factors as indicated by a Beta value of
dependent variable in this research study. 0.384. The need for automation and customer
Government regulation and need for automation had pressure had medium effect on organizational factors
a large effect on organizational factors as shown by a as shown by a Beta value of 0.356. The need for

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

111
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

automation and government regulation had large iii. Organizational readiness: It is the organizational
effect on technological factors as indicated by a Beta preparedness to implement a new innovation.
value of 0.521. Customer pressure and government Readiness in terms of willingness to implement
regulations had a small effect on technological factors chatbot technology, and being ready with both IT
as indicated by a Beta value of 0.121. The need for and financial resources.
automation and customer pressure had no effect on iv. Compatibility: It is the capacity for two systems to
technological factors as indicated by a Beta value of work together without having to be altered to do
0.016. so, for example, AI chatbot and human customer
care representatives support infrastructure.
V. Discussion and Framework v. Complexity: It is the simplicity or difficult of a
system while using it. The system that is simple to
The study indicated that the adoption of AI chatbot use tend to be accepted easily while a system that
technology for real-time customer care service is difficult to use is easily rejected i.e. ease of use.
support is influenced by organizational factors, vi. Technology availability/readiness: It is individual’s
technological factors and intervening factors. The tendency to embrace a new technology for
factors identified were in correlation relationship fulfilling in home life or work. The new
with each other. Based on results of data analysis, technology must be available with support
nine distinct AI chatbot adoption factors were infrastructure and must be secure and reliable.
identified that will influence adoption of AI chatbot vii. Customer pressure: It’s the pressure from
technology in customer care support service customers who need prompt and frequent
provision in telecommunication industry. These feedback when they make customer care support
factors are: service requests. The need for automation comes
as a result of competition from other industries
i. Perceived benefits: It’s the perception of positive that are adopting new innovations for their daily
consequence or feedback as a result of adopting business operations.
chatbot for real-time customer care support forviii. Need for automation: The company’s willingness
instance, cut costs, create better CX, and real time to automate some functions in its operation.
support i.e. perceived usefulness. ix. Government regulation: The regulations given by
ii. Top management support: It is the support from the government on how a company should
executives who translate policies into goals. They perform its functions.
make decisions which affect everyone in the firm.
They fund an innovation.

Figure 4.2: AI Chatbot Adoption Framework

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

112
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

VI. CONCLUSION [5]. Amin, A., Anwar, S., Adnan, A., Nawaz, M.,
Alawfi, K., Hussain, A., & Huang, K. (2017).
The researcher discovered that AI chatbot technology
Customer churn prediction in the
is implemented to a small extend. The results
telecommunication sector using a rough set
confirmed that telecom firms were ready to adopt the
AI chatbot technology. The new technology is approach. Neurocomputing, 237, 242-254.
believed to enhance customer insights and cut costs [6]. Coussement, K., Lessmann, S., & Verstraeten, G.
as fewer human customer care representatives will be (2017). A comparative analysis of data
required as a result of automation. The AI chatbot preparation algorithms for customer churn
technology should be less complex and compatible. prediction: A case study in the
The benefits that come with adoption of this AI
telecommunication industry. Decision Support
chatbot technology include cost cutting and saving
Systems, 95, 27-36.
on time as long waits while calling or messaging will
be significantly reduced since the application can [7]. Hou, R., Wu, J., & Du, H. S. (2017). Customer
respond to many customers at once. social network affects marketing strategy: A
simulation analysis based on competitive
Based on the findings of the study, the framework diffusion model. Physica A: Statistical Mechanics
developed can be used as a guiding principle, to guide and its Applications, 469, 644-653.
stakeholders in adoption of AI chatbot technology in
[8]. Gustafsson, A., Högström, C., Radnor, Z.,
telecommunication industry. The adoption will
ensure that customers do not spend significant Friman, M., Heinonen, K., Jaakkola, E., & Mele,
amount of time to get solutions to their requests, C. (2016). Developing service research paving
which will result in better customer experience. the way to transdisciplinary research. Journal of
Service Management, 27 (1), 9-20.
VII. REFERENCES [9]. Bijmolt, T. H., Leeflang, P. S., Block, F.,
Eisenbeiss, M., Hardie, B. G., Lemmens, A., &
[1]. Huang, M. H., & Rust, R. T. (2013). IT-related Saffert, P. (2010). Analytics for customer
service: A multidisciplinary perspective. Journal engagement. Journal of Service Research, 13(3),
of Service Research, 16(3), 251-258. 341-356.
[2]. Rust, R. T., & Huang, M. H. (2014). The service [10]. Demirkan, H., & Parts, D. (2013). Leveraging the
revolution and the transformation of marketing capabilities of service-oriented decision support
science. Marketing Science, 33(2), 206-221. systems: Putting analytics and big data in the
[3]. Hartmann, PM, Hartmann, PM, Zaki, M., Zaki, cloud. Decision Support Systems, 55 (1), 412-
M., Feldmann, N., Feldmann, N., ... & Neely, A. 421.
(2016). Capturing value from big data-a [11]. Klaus, P., 2014. Measuring customer experience:
taxonomy or data-driven business models used How to develop and execute the most profitable
by start-up firms. International Journal of customer experience strategies. Springer.
Operations & Production Management, 36 (10), [12]. IBM (2016). Using Watson to improve customer
1382-1406. service. from:
[4]. Kunz, W., Aksoy, L. Bart, Y., Heinonen, K., https://www.ibm.com/blogs/watson/2016/04/usi
Kabadayı, S., Ordenes, FV, ... & Theodoulidis, B. ng-watson-improve-customer-service/
(2017). Customer engagement in a big data [13]. Chakrabarti, C. and Luger, G.F., 2015. Artificial
world. Journal of Services Marketing, 31 (2). conversations for customer service chatter bots:
Architecture, algorithms, and evaluation metrics.

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

113
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

Expert Systems with Applications, 42(20), [22]. ALLEN, James, REICHHELD, Frederick, and
pp.6878-6897. HAMILTON, Barney. The Three “Ds” of
[14]. Heinonen, K., Strandvik, T., Mickelsson, K-J., Customer Service. Harvard Business School,
Edvardsson, B., Sundström, E., and Andersson, P. 2005. Available at: http://hbswk.hbs.
(2010): A Customer Dominant Logic of Service, edu/archive/5075.html.
Journal of Service Management, 21 (4) 531-548 [23]. WANG, Yonggui, LO, Hing-Po, and YANG,
[15]. Ordenes, F., Theodoulidis, B., Burton, J., Gruber, Yongheng. An integrated framework for service
T and Zaki, M. (2014). Analyzing Customer quality, customer value, satisfaction: Evidence
Experience Feedback Using Text Mining: A from China’s telecommunication industry.
Linguistics-Based Approach. Journal of Service Information systems frontiers, v. 6, n. 4, pp. 325-
Research. 340, 2004.
[16]. Keiningham, T., Ball, J., Benoit, S., Bruce, H. L., [24]. KUO, Ying-Feng, WU, Chi-Ming, and DENG,
Buoye, A., Dzenkovska, J., ... & Zaki, M. (2017). Wei-Jaw. The relationships among service
The interplay of customer experience and quality, perceived value, customer satisfaction,
commitment. Journal of Services Marketing, and post-purchase intention in mobile value-
31(2). added services. Computers in Human Behaviour,
[17]. Walczuch, R., Lemmink, J., & Streukens, S. v. 25, n. 4, pp. 887– 896, 2009.
(2007). The effect of service employees’ [25]. KIM, Moon-Koo, PARK, M Myeong-Cheol, and
technology readiness on technology acceptance. JEONG, Dong-Heon. The effects of customer
Information & Management, 44(2), 206-215. satisfaction and switching barrier on customer
[18]. Carlson, J., & O'Cass, A. (2010). Exploring the loyalty in Korean mobile telecommunication
relationships between e-service quality, services. Telecommunications policy, v. 28, n.2,
satisfaction, attitudes and behaviors in content- pp. 145-159, 2004.
driven e-service web sites. Journal of Services [26]. MICIACK, Alan and DESMARAIS, Mike.
Marketing, 24 (2), 112-127. Benchmarking service quality performance at
[19]. HOLBROOK, Morris, and HIRSCHMAN, business-to-business and business-to-consumer
Elizabeth. The Experiential Aspects of call centers. The Journal of Business & Industrial
Consumption: Consumer Fantasies, Feelings, and Marketing, v. 16, n. 5, pp. 340-53, 2001.
Fun. Journal of Consumer Research, v. 9, n.2, pp. [27]. SANTOURIDIS, Ilias, and TRIVELLAS,
132–140, 1982. Panagiotis. Investigating the impact of service
[20]. MEYER, Christopher, and SCHWAGER, Andre. quality and customer satisfaction on customer
Understanding Customer Experience. Harvard loyalty in mobile telephony in Greece. The TQM
Business Review, 2007. Available at: Journal, v. 22, n. 3, pp. 330-343, 2010.
https://hbr.org/2007/02/understanding- [28]. Michael, J and Mitchell, T. (2015), “Machine
customer-experience. learning: Trends, perspectives, and prospects.”
[21]. ZWILLING, Martin. Customer Experience’ Is Science 349(6245): 255.
Today’s Business Benchmark. Forbes, 2014. [29]. Alan Turing, "Computing Machinery and
Available at: Intelligence".(Turing 1950) In 1956, at the
https://www.forbes.com/sites/martinzwilling/20 original Dartmouth AI summer conference, Ray
14/03/10/customer-experience-is-todays- Solomonoff, "An Inductive Inference
business-benchmark/#4a190d8e5011. Machine".(Solomonoff 1956)

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

114
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

[30]. Tom Mitchell: "A computer program is set to International, 333 Ravenswood Avenue, Menlo
learn from an experience E with respect to some Park, CA 94025.
task T and some performance measure P if its [39]. Computer Science and Technology Board ^
performance on T as measured by P improves National Research Council (ed.): 1997, More
with experience E. Than Screen Deep: Toward Every-Citizen
[31]. Luger, George; Stubblefield, William (2004). Interfaces to the Nation's Information
Artificial Intelligence: Structures and Strategies Infrastructure, National Academy Press,
for Complex Problem Solving (5th ed.). Washington, D.C.
Benjamin/Cummings. ISBN 978-0-8053-4780-7. [40]. Helander, M. G., Landauer, T. K., and Prabhu, P.
[32]. Jordan, M. I.; Mitchell, T. M. (16 July 2015). V. (Eds.): 1997, Handbook of Human-Computer
"Machine learning: Trends, perspectives, and Interaction, (Second, Completely Revised ed.),
prospects". Science. 349 (6245): 255–260. Elsevier Science Ltd., Amsterdam.
Bibcode:2015Sci...349255J. [41]. Fischer, G.: 1993a, beyond human computer
doi:10.1126/science.aaa8415. PMID 26185243. interaction: Designing useful and usable
[33]. Russell, Stuart J.; Norvig, Peter (2003), Artificial computational environments. In: People and
Intelligence: A Modern Approach (2nd ed.), Computers VIII: Proceedings of the HCI'93
Upper Saddle River, New Jersey: Prentice Hall, Conference (Loughborough, England),
ISBN 0-13-790395-2 Cambridge University Press, Cambridge, UK, pp.
[34]. Poole, David; Mackworth, Alan; Goebel, Randy 17^31.
(1998). Computational Intelligence: A Logical [42]. Schermerhorn, R. Cantrell, M. Scheutz, and X.
Approach. New York: Oxford University Press. Wu (2010), Robust Spoken Instruction
ISBN 978-0-19-510270-3. Understanding for HCI. Bloomington: Indiana
[35]. Mittal, "Versatile question answering systems: University.
seeing in synthesis" Archived 1 February 2016 at [43]. R. Cole, L. Hirschman, et al. (1995), The
the Wayback Machine, IJIIDS, 5(2), 119–142, Challenge of Spoken Language Systems:
2011 Research Directions for the Nineties. Oregon:
[36]. Russell & Norvig. Applications of natural Graduate Institute.
language processing, including information [44]. Abu Shawar, B., & Atwell, E. (2007). Chatbots:
retrieval (i.e. text mining) and machine Are they really useful? Journal for Language
translation: 2003, pp. 840–857, Luger & Technology and Computational Linguistics,
Stubblefield 2004, pp. 623–630 22(1), 29-49. Retrieved from
[37]. Cambria, Erik; White, Bebo (May 2014). http://www.jlcl.org/2007_Heft1/Bayan_Abu-
"Jumping NLP Curves: A Review of Natural Shawar_and_Eric_Atwell.pdf
Language Processing Research Review Article]". [45]. Brandtzaeg, P. B., & Følstad, A. (2017). Why
IEEE Computational Intelligence Magazine. 9 people use chatbots. In I. Kompatsiaris, J. Cave,
(2): 48–57. doi:10.1109/MCI.2014.2307227. A. Satsiou, G. Carle, A. Passani, E. Kontopoulos,
[38]. John Dowding, Jean Mark Gawron, Doug S. Diplaris, & D. McMillan (Eds.), Internet
Applet, John Bear, Lynn Cherney, Robert Science: 4th International Conference, INSCI
Moore, Douglas Moran "GEMINI: A NATURAL 2017 (pp. 377-392). Cham: Springer
LANGUAGE SYSTEM FOR SPOKEN- [46]. Accenture. (2016). Chatbots in Customer Service
LANGUAGE UNDERSTANDING", SRI Retrieved from
https://www.accenture.com/t00010101T000000_

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

115
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

_w__/br-pt/_acnmedia/PDF-45/Accenture- [55]. Servion. (2017). AI will power 95% of customer

Chatbots-Customer-Service.pdf interactions by 2025. Retrieved from
[47]. Holtgraves, T. M., Ross, S. J., Weywadt, C. R., & http://servion.com/blog/ai-will-power-95-
Han, T. L. (2007). Perceiving artificial social customer-interactions-2025/
agents. Computers in Human Behavior, 23(5), [56]. Simonite, T. (2017). Facebook’s perfect,
2163-2174. doi: 10.1016/j.chb.2006.02.017 impossible chatbot. from MIT Technology
[48]. Crutzen, R., Peters, G. J., Portugal, S. D., Fisser, Review
E. M., & Grolleman, J. J. (2011). An artificially https://www.technologyreview.com/s/604117/fa
intelligent chat agent that answers adolescents' cebooks-perfect-impossible-chatbot/
questions related to sex, drugs, and alcohol: An [57]. Coniam, D. (2014). The linguistic accuracy of
exploratory study. Journal of Adolescent Health, chatbots: Usability from an ESL Perspective.
48(5), 514-519. doi: Text Talk, 34(5), 545-567. doi:10.1515/text-2014-
10.1016/j.jadohealth.2010.09.002 0018.
[49]. Brynjolfsson, E., & McAfee, A. (2017). Machine, [58]. Atwell, E., & Shawar, B.A. (2007). Chatbots: Are
Platform, Crowd. New York, USA: W.W. they Really Useful? LDV Forum, 22, 29-49.
Norton & Company. [59]. Mott, B., Lester, J., & Branting, K. (2004).
[50]. Fryer, L., & Carpenter, R. (2006). Bots as Conversational Agents. The Practical Handbook
language learning tools. Language Learning & of Internet Computing.
Technology, 10(3), 8-14. doi:10125/44068. doi:10.1201/9780203507223.ch10
[51]. Monsen, P. A. M. (2018). Rekruttering med [60]. Kerly, A., Hall, P., & Bull, S. (2007). Bringing
Chatbot. Retrieved from Chatbots into education: Towards Natural
https://hrnorge.no/aktuelt/rekruttering-med- Language Negotiation of Open Learner Models.
chatbot. Applications and Innovations in Intelligent
[52]. Weizenbaum, J. (1966). ELIZA - A computer Systems XIV, 179-192. doi:10.1007/978-1-
program for the study of natural language 84628-666-7_14
communication between man and machine. [61]. Bickmore, T. W., Schulman, D., & Sidner, C.
Communications of the ACM, 9(1), 36-45. (2013). Automated interventions for multiple
doi:10.1145/365153.365168 health behaviors using conversational agents.
[53]. Brandtzaeg, P. B., & Følstad, A. (2017). Why Patient Education and counseling, 92(2), 142-
people use chatbots. In I. Kompatsiaris, J. Cave, 148. doi:10.1016/j.pec2013. 05.011
A. Satsiou, G. Carle, A. Passani, E. Kontopoulos, [62]. Haubl, G., & Trifts, V. (2000). Consumer
S. Diplaris, & D. McMillan (Eds.), Internet Decision Making in Online Shopping
Science: 4th International Conference, INSCI Environments: The Effects of Interactive
2017 (pp. 377-392). Cham: Springer. Decision Aids. Marketing Science, 19(1), 4-21.
[54]. Zumstein, D., & Hundertmark, S. (2017). doi:10.1287/mksc.19.1.4.15178
Chatbots - An interactive technology for [63]. Rowley, J. (2000). Product searching with
personalized communication, transactions and shopping bots. Internet Research, 10(3), 203-214.
services. IADIS International Journal on doi:10.1108/10662240010331957
WWW/Internet, 15(1), 96-109. Retrieved from [64]. Sadeddin, K. W., Serenko, A., & Hayes, J. (2007).
http://www.iadisportal.org/ijwi/papers/20171511 Online shopping bots for electronic commerce:
07.pdf. the comparison of functionality and

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

116
 Geoffrey Nyongesa et al Int J Sci Res CSE & IT, November-December-2020; 6 (6) : 100-117

performance. IJEB, 5(6), 576. [74]. https://medium.com/@RateMyService/chatbotcx

doi:10.1504/ijeb.2007.016472 -9182f43552f3.
[65]. Mhatre, N., Motani, K., Shah, M., & Mali, S. [75]. https://medium.com/@RateMyService/chatbotcx
(2016). Donna Interactive Chat-bot acting as a -9182f43552f3.
Personal Assistant. International Journal of [76]. Nwosu, John Nwachukwu. "An Investigation
Computer Applications, 140(10), 6-11. into the Extent of the Use of Artificial
doi:10.5120/ijca2016909460. Intelligence in Nigeria Banks."
[66]. Griol, D., Carbó, J., & Molina, J. M. (2013). AN [77]. J. Bizimungu, "Babyl’s chatbot to enhance digital
AUTOMATIC DIALOG SIMULATION healthcare platform", The New Times Rwanda,
TECHNIQUE TO DEVELOPAND EVALUATE Jan 2018, online] Available:
INTERACTIVE CONVERSATIONAL AGENTS. https://www.newtimes.co.rw/section/read/22736
Applied Artificial Intelligence, 27(9), 759- 9.
780.doi:10.1080/ 08839514.2013.835230 [78]. https://medium.com/@RateMyService/chatbotcx
[67]. Kuligowska, K. (2015). Commercial Chatbot: -9182f43552f3.
Performance Evaluation, Usability Metrics and [79]. https://www.cio.co.ke/ai-chatbot-proves-to-be-
Quality Standards of Embodied Conversational a-useful-teacher-in-farming-and-agronomy/.
Agents. PCBR, 2(02), 1-16. doi:10.18483/pcbr.22. [80]. https://technobraingroup.com/digital/artificial-
[68]. CB Insights. The race for AI: Google, Twitter, intelligence-machine-learning/chatbots/.
Intel, Apple in a rush to grab artificial [81]. Saunders M, Lewis P, Thornhill A. Research
intelligence startups Internet]. New York: CB methods for business students. Pearson
Insights. 2016 Dec 6 cited 2016 Dec 10]. education; 2009.
Available from: [82]. Mugenda A. Social science research: Conception,
https://www.cbinsights.com/blog/topacquirers- methodology and analysis. Nairobi: Kenya
ai-startups-ma-timeline/. Applied Research and Training Services. 2008.
[69]. Kepuska, Veton, and Gamal Bohouta. [83]. Porter S, Woodbine G. The effect of ethics
"Comparing speech recognition systems courses on the ethical judgement‐making ability
(Microsoft API, Google API and CMU Sphinx)." of Malaysian accounting students. Journal of
Int. J. Eng. Res. Appl 7.03 (2017): 20-24. Financial Reporting and Accounting. 2010 Oct
[70]. Hebert, Martial H., Charles E. Thorpe, and 26.
Anthony Stentz, eds. Intelligent unmanned Cite this article as :
ground vehicles: autonomous navigation
research at Carnegie Mellon. Vol. 388. Springer Geoffrey Nyongesa, Kelvin Omieno, Daniel Otanga,
Science & Business Media, 2012. "Artificial Intelligence Chatbot Adoption Framework
[71]. Rahim, Siti Rohaya Mat, et al. "Artificial for Real-Time Customer Care Support in
intelligence, smart contract and islamic finance." Kenya", International Journal of Scientific Research
Asian Social Science 14.2 (2018): 145. in Computer Science, Engineering and Information
[72]. Ferrucci D, Levas A, Bagchi S, Gondek D, Technology (IJSRCSEIT), ISSN : 2456-3307, Volume
Mueller ET. Watson: beyond jeopardy! Artif 6, Issue 6, pp.100-117, November-December-2020.
Intell 2013;199–200:93–105. Available at
[73]. https://eastafricadigitalmarketers.com/chatbot- doi : https://doi.org/10.32628/CSEIT20667
messenger-development/. Journal URL : http://ijsrcseit.com/CSEIT20667

Volume 6, Issue 6, November-December-2020 | http://ijsrcseit.com

117