Technological Forecasting & Social Change 198 (2024) 122967

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

journal homepage: www.elsevier.com/locate/techfore

Technological paradoxes and artificial intelligence implementation in

healthcare. An application of paradox theory
Nidhi Singh a, *, Monika Jain b, Muhammad Mustafa Kamal c, Rahul Bodhi d, Bhumika Gupta e
a
Jaipuria institute of Management, Noida, India
b
Birla Institute of Management Technology, Greater Noida, Uttar Pradesh 201306, India
c
School of Strategy and Leadership, Coventry University, Coventry CV1 5FB, United Kingdom
d
School of Business, University of Petroleum and Energy Studies (UPES), Dehradun, Uttarakhand-248007, India
e
LITEM, Institut Mines-Telecom Business School (IMT-BS), University Evry, Universite Paris-Saclay, France

A R T I C L E I N F O A B S T R A C T

Keywords: AI is transforming healthcare system with many innovations in diagnosis, drug research and advancement in
AI medical treatments. But there are several concerns and dilemmas related to data misuse, AI efficiency for critical
Drivers diagnostic services, users' resistance, investment costs, funding issues, and so on that have been raised by many
Barriers
previous studies on the effective integration of AI in clinical settings. Using paradox theory in the organisational
Dilemmas
settings, the present study discusses several technological paradoxes associated with the adoption of AI in
AI strategy
Health system healthcare. In this regard, the study examines the views of diverse medical practitioners about using AI services
Adoption for several medical needs. The study analyses the efficacy and limitations of AI services which develop several
ethical dilemmas in the mind of medical practitioners and also suggest a few strategies for the adoption. Using
grounded theory approach, the study collected views of 62 medical practitioners on these dimensions. The
primary drivers to the adoption identified in the present study are: ease of use, automation efficacy, diagnostic
accuracy, and cost efficiency. A lack of training and education, cultural and religious considerations, privacy
issues and work insecurity are some of the concerns highlighted by the medical staffs. The study inferred a few
paradoxes or ethical dilemmas of practitioners which need attentions. The study contributes to the existing
literature on paradox theory and AI, and identifies a few under-discussed areas, drivers, and barriers of AI
services are highlighted in the paper, which may lead to ethical concerns and steer AI adoption in healthcare.

1. Introduction full capacity (Tseng et al., 2021). Previous studies have been able to
examine the beneficial and detrimental aspects of AI services in relation
Artificial intelligence (AI) is quickly expanding in the healthcare to advanced medical needs, user convenience, potential impact of
sector, with a predicted market size of 45.2 billion dollars by 2026 automation, and data management to some extent (Wong et al., 2019;
(researchAndMarkets.com, 2020). Literature indicates that AI's effi­ Cubric, 2020; Norori et al., 2021). There remains a dearth of research
ciency among medical practitioners is enhanced when managing vast that comprehensively identifies and elucidates various technological
amounts of data as well as providing preventive care, such as rapid advantages and barriers, which in turn collectively give rise to a
diagnosis and treatment, and other medical services (Kumar et al., 2021; multitude of paradoxes or dilemmas from the standpoint of healthcare
Khanijahani et al., 2022). However, it is observed that the adoption of AI providers. These include inquiries into whether the use of AI and human
services by hospitals is largely only for recording medical records, expertise will lead to decisions made by consensus, whether service
assisting in patients' diagnosis, and for operational issues (Lee and Lee, automation or augmenting is needed in a health setting, and whether the
2021; Gruetzemacher et al., 2021). Adoption is progressing in few other system will affect patients' privacy or give them personalised care. When
areas to leverage the benefits of AI services such as radiology, blood contemplating the adoption of AI, these issues can present medical
testing, genomics, etc. but failed to utilise and explore the services at the practitioners with ethical dilemmas. The present study offers a thorough

* Corresponding author.
E-mail addresses: [email protected] (N. Singh), [email protected] (M. Jain), [email protected] (M.M. Kamal), [email protected].
ac.in (R. Bodhi), [email protected] (B. Gupta).

https://doi.org/10.1016/j.techfore.2023.122967
Received 18 April 2023; Received in revised form 12 October 2023; Accepted 31 October 2023
Available online 10 November 2023
0040-1625/© 2023 Elsevier Inc. All rights reserved.
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

examination of these dimensions and their implications. of AI-enabled healthcare services in several advanced health domains?
Much research presently focuses on the effective utilisation of AI- RQ2: What are the key hurdles medical practitioners encounter while
enabled technologies in various sectors (Di Vaio et al., 2020; Kumar adopting the service for critical diagnosis and advanced care services?
et al., 2021). Among these, a few studies have used a grounded theory RQ3: What are the key ethical dilemmas medical practitioners encounter
approach to assess the potential predictors and barriers to the effective due to their varied and unpredictable experience with the use of AI
deployment of AI in healthcare (He et al., 2021). This research hinted at services? RQ4: What adoption strategy should medical practitioners
the use of AI at various levels for operational convenience, diagnosis adopt to handle the challenges and dilemmas associated with AI-enabled
management, and for cost and time-saving procedures (Lee and Lee, services in hospitals?
2021). However, these studies largely observed the AI role in managing To answer these research questions, we used qualitative study design
day-to-day operations and patient management rather than discussing consisted of open-ended essays to explore various drivers and barriers of
deeper engagement of the service. Previous literature, on the other AI-enabled technologies in healthcare. A total of 62 medical practi­
hand, demonstrated that the adoption of AI in healthcare is difficult due tioners employed in UK healthcare services participated in the study in
to paradoxes regarding its effective integration with traditional business the month of January 2023. The findings revealed four main drivers to
processes, resistance and mindset changes, data privacy issues, biases, use AI-enabled technologies: ease of use, effectiveness of automation,
system-related complexities, and limitations and abilities of the service accurate diagnosis, and cost efficiency. At the same time, five main
(Jussupow et al., 2021; Gruetzemacher et al., 2021; Siala and Wang, barriers: privacy and trust issues; unsuitable for critical illness; lack of
2022). According to Shi et al. (2020), the use of AI for sensitive data training and education; financial constraints; and behavioural resistance
requirements is under research. Several contradictions have been were also highlighted. The study indicated a few ethical dilemmas
observed while employing AI for critical diagnostic services and associated with privacy issues, human-non-human interventions, per­
depending on AI recommendations (Cubric, 2020; Raisch and Kra­ sonalised care requirements, etc. A few strategies for AI adoption were
kowski, 2021). There is still inconsistency in the automation and also identified. The present study contributes to the extant literature on
augmentation functions of AI services, which poses decisional issues technological paradoxes by accumulating and analysing the conflicting
(Tseng et al., 2021; Wamba et al., 2021). Data privacy and biases have views of various medical staff and practitioners about what AI can do
been called into question in a few works, increasing technological par­ and what it cannot. We are extending the existing literature by focusing
adoxes (Wiljer and Hakim, 2019; Jussupow et al., 2021). There is a on a few key drivers, barriers, ethical dilemmas, and strategies for the
limited discussions about how to use AI for advanced treatments, per­ adoption of AI for advanced care services. The current study is organised
sonalised services, and critical disease diagnosis (Wong et al., 2019). into the following sections: First, we provide a brief literature review on
This could lead to dilemmas among healthcare professionals about how the development and use of AI adoption. In the second section, we
to use AI beyond the operational level and calls for more discussion explain the research methodology used to assess several factors associ­
using technological paradox perspective, which is still underway in the ated with AI adoption. The third section reports the findings regarding
literature (Raisch and Krakowski, 2021). the drivers, barriers, and dilemmas associated with adoption. In the
After reviewing the existing literature, we identified a few research fourth section, we detail the theoretical contributions of the study. The
gaps. Although past studies offer rich insights into how AI can transform last section is about directions for future research and limitations.
healthcare, limited discussion is available on the application of AI ser­
vices for various advanced and critical care services. This gap in the 2. Review of literature
literature will restrict the ability to gain trust and extract a return on
investment (Kumar et al., 2023). Second, medical practitioner-based 2.1. AI in organisations
studies are few in the literature and those available are mainly
focused on compiling views to a specific kind of technology, including AI Artificial intelligence (AI) is a branch of computer science that aims
chat boxes (Nadarzynski et al., 2019), PCAS adoption (Paré and Trudel, to simulate human behaviour and intellect in complex systems artifi­
2007), and many more. There is still a lack of understanding of how cially (Wiljer and Hakim, 2019). Over the last few decades, the use of AI
these views on various AI-based services, when brought together, can has been increasing in different areas or businesses, such as online
develop ethical dilemmas for decision makers. Also, how these di­ commerce, manufacturing, financial services, hospitality, and medical
lemmas may enhance or restrict their use of AI services needs further fields (Char et al., 2020). Many organisations have used AI to implement
discussion (Jussupow et al., 2021). We believe that paradox theory is a advanced computer systems or robotics systems to enhance their capa­
good way to evaluate different or contradictory points of view from bilities and competitive edge (Martinez-Millana et al., 2022). However,
medical practitioners. the major challenge was to explain how AI performed tasks, which
Third, many barriers and behavioural aspects of AI applications were enabled machines to simulate human behaviour to achieve the best
explored in a broader social context, such as privacy issues, regulatory expected results (Char et al., 2020). This is due to the complexity of
and ethical concerns, integration aspects, etc. (Guan, 2019; Char et al., algorithms and the large amount of complex data used by AI, which is
2020; Lee and Lim, 2021). Our focus is that the implementation of AI for unknown or unfamiliar in the vast majority of cases (Reddy et al., 2020;
critical healthcare services may not fit with privacy perceptions, critical Sharma et al., 2022). Recent work on these observations explains that AI
illnesses, etc. that may affect staff participative behaviour and develop systems should not be focused only on simulating human behaviour but
ethical dilemmas. Using paradox theory, the present study aims to also aim to explain human learning in the decision-making processes
collate the views of medical practitioners on existing AI-based services (Wiljer and Hakim, 2019). This develops a need for a more simplistic
and whether these services can be used for advanced medical diagnosis and representational learning process by relating each piece of infor­
and automation or not. The study further aims to discuss specific oper­ mation with a simpler concept (Martinez-Millana et al., 2022).
ational, functional, and financial challenges in using AI for advanced
treatments and critical illnesses, along with behavioural resistance from 2.2. AI in healthcare
medical staff and patients. Furthermore, departing from previous IS
research, our goal is to examine ethical dilemmas about the relevance Several studies have found that healthcare is the top priority for
and use of services. Lastly, what are the methods of adoption and governments and policymakers when it comes to AI applications (Jus­
advancement of AI-based health services for intensive and critical care supow et al., 2021). Applications of AI are seen as enablers of medical
services? service effectiveness, with a special focus on personalised care for highly
Based on this, the following four key research questions are framed: vulnerable groups. The use of technology has become crucial in the
RQ1: What attributes do medical practitioners encounter during the use healthcare sector as the demand for health services has increased for

2
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

tracing and identification of disease and various medical interventions applications according to them manifests in a paradoxical tension that
(He et al., 2021). These services are found to be beneficial for medical transcends temporal and spatial boundaries. Mill et al. (2022) provides
practitioners with limited medical expertise, physicians with high insights into the sustainability dilemmas inherent in AI and sheds light
mental load and time pressure, and for other unfamiliar specialised on the Accuracy-Energy Paradox and proposes how the concepts of
services (Lee and Lim, 2021). In the healthcare system, AI is used to paradox theory might provide guidance to the AI community in devel­
analyse patients' medical data and assist medical practitioners in making oping a more sustainable solution. Kumar et al. (2023) in their study
an accurate diagnosis using machine learning techniques and algo­ threw light on Augmentation and Automation paradox and the findings
rithms. AI-enabled fitness trackers, medical robotics assistance, per­ of the study revealed that the presence of role ambiguity, job instability,
sonalised health technology are in high demand in the healthcare and the technological environment were significant factors contributing
business. These are used in a variety of health settings, such as identi­ to the experience of technostress resulting from the implementation of
fying critical diseases, chronic health conditions, and forecasting dis­ machine learning (ML) and AI technologies. The key sources of stress in
orders (Cubric, 2020; Khanijahani et al., 2022). AI systems have reached the technological environment were complexity, ambiguity, depend­
a high demand level, but they are not without biases and errors (Jus­ ability, and usefulness according to them (Sharma et al., 2022).
supow et al., 2021). For example, AI-enabled technologies completely Table 2 shows the drivers, barriers, and dilemmas of AI in healthcare
rely on statistical data to provide the analysis of results, while the scope that have been found in previous studies. After reviewing the literature,
of reasoning is minimal in the process (Wiljer and Hakim, 2019; Cubric, it has been identified that AI technology is increasingly interfering in
2020). Similarly, a few systems, including genomics, blood tests, and almost all healthcare domains, particularly healthcare (Bughin et al.,
radiology, are using deep learning algorithms for diagnosis. These 2019; Wong et al., 2019). AI is utilised for a variety of purposes. For
systems-based results may be found ambiguous in comparison to con­ example, in medical imaging, doctors use AI to assess disease, organ
ventional systems involving human knowledge. This may lead to an function, radiation dosage monitoring, and other medical facilities
under or over estimation of the situation (Wong et al., 2019). Studies (Chen et al., 2021). Liu and Tao (2022) in their paper explored the
show that medical staff may become heavily involved with AI to di­ significance of trust and three specific aspects of AI - namely, custom­
agnose critical medical cases, but they may be incapable of comparing isation, loss of privacy, and anthropomorphism - in relation to the public
the results to their own judgements and making informed decisions adoption of smart healthcare services. In a similar vein, Raghavan et al.
(Jussupow et al., 2021). There is a need to understand the capabilities (2021) elucidated that heightened awareness and knowledge regarding
and potential opportunities of adopting AI in healthcare information cyber rules, along with the promotion of digitalization, could potentially
systems (see Table 1). foster a favourable influence on users' trust and adoption of the services.
In addition, past evidence has revealed that AI can also help pro­
2.3. Theoretical lens fessionals interpret complex data sets, identify patterns, and perform
regular medical activities automatically and more effectively (Kalisz
Organisational paradoxes are persistent over time (Chen et al., et al., 2021). However, there are a few conflicting views too. Studies
2021). There are various studies focusing on paradoxes related to argue that we are still far from a “strong form” AI system that can replace
leadership, human resource development, hybrid processes, sustain­ human intelligence in medical diagnosis. Still, the ‘weak form’ of AI can
ability goals, and structural and technological innovations in organisa­ help doctors communicate the system's key usage and benefits to pa­
tions (Reddy et al., 2020). Technology paradoxes appear to emerge in tients (He et al., 2021). The effectiveness of these services is reinforced
organisations as technology adoption creates a number of complex sit­ by the presence of doctors' expertise who possess the ability to not only
uations or tensions (Chen et al., 2021). According to Leoni and Cristo­ interpret but also provide justifications for their recommendations to
faro (2021), new technology adoption is paradoxical as it can either lead patients, thereby minimising errors within the system (Kalisz et al.,
to new opportunities or develop conflicts in decision making. Further, 2021). Jussupow et al. (2021) found these systems less transparent and
tension resulting from these paradoxes is the main hindrance that in­ less predictive. For example, they argued about the AI results' validity
fluences the adoption and views of the practitioners. and accuracy for diagnosis in comparison to a rule-based self-monitoring
Paradox theory refers to the dilemmas that arise due to the multiple, process. This may result in users' resisting views. Tseng et al. (2021) and
often conflicting purposes of services that develop conflicting views Merhi (2023) have identified privacy concerns as substantial barriers to
(Raghavan et al., 2021). Literature uses paradox theory to discuss a few the proper implementation of AI services.
paradoxical tensions observed in technology adoption (Du and Xie, At the same time, several studies explained that this may also lead to
2021) and renewal (Reddy et al., 2020) in organisational settings. For several technological contradictions in health organisations. Jussupow
example, the theory emphasizes the conflicting views on technological et al. (2021) discovered medical practitioners are sometimes perplexed
changes that are suggested in the organisations; these either lead to between the alternatives (AI advice and their own judgements) when
exploration or exploitation of the processes (Khanijahani et al., 2022). evaluating AI discoveries related to life-threatening scenarios such as
There is a clear clash between the automation and augmentation roles of serious heart problems, lung functionality, and other genomics cases.
AI services and the preference of users for better performance. On the Due to several underlying causes, many dilemmas were discussed (Kooli
other hand, compromising with both may lead to tensions and compli­ and Al Muftah, 2022). For example, diverse patient needs and prefer­
cations with the adoption. These kinds of tensions are most common ences (Balagurunathan et al., 2021), lack of medical staff's education
when two different perspectives or different situations are used in and training (Alhashmi et al., 2019), fear of using technology for
organisational technology adoption or for any dual-perspective pro­ treatments (Chen et al., 2021), data privacy and consent issues (Di Vaio
cesses. Angelucci et al. (2022) discussed a fundamental paradox of the et al., 2020; Tandon et al., 2020), and achieving a delicate balance be­
application of AI According to them the level of involvement of a tween the domains of health, and patient freedom (Laï et al., 2020) are
stakeholder in the life cycle of an AI system is inversely proportional to some of the concern areas that create dilemmas. Despite AI's ability to
the degree of influence they exert on the system's functioning. This comprehend and provide information beyond human intelligence, these
implies that individuals who are less affected by the system have the challenges have raised and caused tensions between medical pro­
responsibility for determining its influence on fairness. Angelucci et al. fessionals and patients (Martinez-Millana et al., 2022). We therefore feel
(2022) in their paper discussed the paradoxes of AI in consumer market that paradox theory is appropriate to assess conflicting or diverse views
and also how companies can influence the trajectory of ethical AI by of medical practitioners. The present study extends the existing litera­
undertaking socially responsible measures to tackle these concerns. ture by both confirming existing conundrums and investigating possible
Raisch and Krakowski (2021) applied paradox theory, to understand the dilemmas that may emerge as a result of paradoxes or competing per­
interdependence of augmentation and automation. These dual AI spectives of healthcare providers about the utilisation of AI-enabled

3
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Table 1
Review of existing literature on AI adoption.
Research outline Purpose and focus Level of Key findings Theoretical framework
analysis

Authors: Kumar et al. To identify critical success factors for AI adoption Macro Technological factors are influencing the most that Organisation-
(2023) in healthcare supply chain in the emerging impacts AI adoption. Environment (TOE)
economy context. Other factors including institutional or framework
Method: mixed-method environmental, human. and organisational also
(qualitative and plays a critical role in AI adoption. Human-Organisation-
quantitative) Technology fit (HOT-
fit)
Sample: 20 expert
opinions and extensive
literature review
Authors: Dicuonzo et al. How AI supports the effective and efficient Macro AI supports the effective and efficient management Theoretical framework
(2023) management of the healthcare system of the healthcare system. outlined by Kazanjian
AI can enable optimisation of healthcare resources and Green (2002)
Method: Qualitative and improve health outcomes.

Sample: 19 semi-
structured interviews
and Case study
Authors: Ma et al. (2023) To provide a comprehensive overview of AI Macro The impact of AI technologies on elderly healthcare No
technologies in elderly healthcare is promising and that AI technologies are capable of
Method: Extensive satisfying the unmet care needs of older adults
literature review
(scoping review)

Sample: 105 studies

Authors: Siala and Wang This article focus on how can AI-related Five core themes (SHIFT: Sustainability, Human Virtue ethics theory
(2022) technologies and initiatives be implemented in a centeredness, Inclusiveness, Fairness, and
responsible manner in healthcare? Transparency) emerged for AI solution developers,
Method: Extensive healthcare professionals, and policy makers
literature review
(PRISMA approach)

Sample: 253 articles

Authors: Khanijahani et al. This article aimed to identify and document Micro The result show that professional characteristics
(2022) organisational, professional, and patient associated with AI adoption in healthcare settings
characteristics influencing the adoption of into three main categories such as psychosocial,
Method: Extensive Artificial Intelligence (AI) in healthcare. experiential, and background elements.
literature review
(PRISMA approach)

Sample: 27 articles
Authors: Oprescu et al. To design and deploy Responsible AI based Macro Responsible AI can help pregnant women trust an TOGAF, DARPA
(2022) solutions for personalising pregnancy healthcare. intelligent solution to prevent pregnancy risks.

Method: Qualitative
study

Sample: Interview of 150

pregnant patients and
case study
Authors: Martinez-Millana To perform a systematic overview of systematic Micro The evidence of AI use is often limited to laboratory No
et al. (2022) reviews on the application of artificial and the level of adoption is largely unbalanced
intelligence in the people's health domains
Method: Extensive
literature review

Sample: 129 studies

Authors: Jussupow et al. To understand how AI influences physicians' Micro Physicians use metacognitions to monitor and No
(2021) decisions control their reasoning while evaluating AI advice.
Method: Interview Right or wrong decision depends on utilising
Sample: 68 novice and 12 metacognitions
experienced physicians
Author: Kumar et al. Determine the components of responsible AI in Individuals There is significant influence of patient's cognitive Means-end theory and
(2021) healthcare and its impact on value creation and engagement AI-solutions and perceived value, service-dominant logic
market performance. leading to market performance
Method: Interview,
online survey and dyadic
data analysis.

Sample: 12 in-depth
interview, 290 users and
174 dyadic data
(continued on next page)

4
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Table 1 (continued )
Research outline Purpose and focus Level of Key findings Theoretical framework
analysis

Author: Mirbabaie et al. Identifying the current state of interdisciplinary Macro and An understanding of the current ethical discourse of No
(2021) academic research on ethical issues and AI in Individual AI in clinical environments such as privacy,
hospitals. accountability, data biases, education and others,
Method: Systematic and related ethical concerns
literature analysis and
interview method

Sample: 15 published
manuscript and 6 semi
structured expert
interviews.
Authors: Cubric (2020) To provide an overview of existing systematic Macro The main drivers are economic. The main barriers No
reviews in this rapidly growing field of research, are technical and social.
Method: Review of and to summarise findings related to AI adoption
Literature in business and management.

Sample: 30 reviews
studies published
between 2005 and 2019
Authors: Reddy et al. The article proposes a governance model to Macro Discussion on ethical and regulatory issues like AI No
(2020) address the ethical and regulatory issues raised model biases and transparency. Developing a
by the use of AI in health care. governance model to address these issues
Method: Conceptual

Sample: NA
Author: Char et al. (2020) An overview of value-based issues, along with Macro Proposed framework to identify ethical No
ethical issues related to the use of ML-based considerations through conceptualisation,
Method: Conceptual technology in healthcare development and calibration, implementation and
evaluation, oversight of events.
Sample: NA

services. and the process follows.

3. Methodology 3.1. Data collection approach

A gap exists in our knowledge to understand the perception of We were interested in a diverse group of medical practitioners who
medical practitioners about the key drivers and barriers associated with had a comprehensive understanding of the use of AI-based technologies
AI-based technologies. Further, what are the ethical dilemmas faced by in healthcare and some experience of the related dimensions to enhance
these practitioners, and how do they suggest overcoming these concerns the credibility and validity of the present research. In this study, a
with the adoption of such technologies in the healthcare industry, as was nonprobability purposive sampling method was employed to recruit a
limitedly explored in the literature? The current study is an exploratory sample of over 200 medical practitioners, encompassing doctors, nurses,
research (Swedberg, 2020) designed to understand the perception of medical engineers, medical technicians, healthcare consultants, medical
medical practitioners about the key drivers, barriers, and ethical di­ staff, physicians, specialist nurses, and other relevant professionals.
lemmas associated with AI-based technologies. Exploratory research These participants were invited to engage in open-ended essay re­
serves the purpose of collecting initial data, discerning patterns and sponses, which were conducted on an online platform called prolific.co,
themes or generate hypotheses that can be subsequently examined in during the month of January 2023. We sampled these respondents based
greater depth. To comprehend this, the study used a qualitative research on three qualifying criteria: a) being aware of AI-based medical services
design using the grounded theory method (Chaudhary et al., 2022). This and technologies in healthcare. b) have carried out a few AI-based ser­
approach helps us theorise inductively from the data to draw insights vices for hospital operations, medical procedures, and diagnostics in
about the views of medical practitioners about AI-based technologies. medical centres. c) have been using these services for at least three years.
Open-ended questions are popular in management research as they can We conducted a screening study and recruited 62 participants who
help collect qualitative data or the views of respondents (Dhir et al., matched our desired criteria and agreed to participate in the study
2017). Open-ended essays allowed us to collect data from different re­ (Malodia et al., 2023). The implementation of these qualifying criteria
spondents, improving the validity of our findings by collecting diverse was to guarantee that participants have an adequate level of experience
perspectives and experiences in different circumstances to examine the in the pertinent techniques, enabling them to effectively articulate their
topic in detail (Chaudhary et al., 2022). responses based on their relevant experiences and formulate their in­
Open-ended essays helped us enhance theoretical perspectives on terpretations. In order to locate additional relevant respondents, snow­
new technology adoption and find novel explanations for observed be­ ball sampling was applied, whereby research participants provided
haviours (Ma et al., 2023). Second, healthcare AI-enabled technology referrals to similar participants from diverse medical backgrounds
adoption is complicated by technical challenges, privacy concerns, and (Christofi et al., 2023). Researchers have employed a small yet diverse
behavioural shifts (Wong et al., 2019). This method will reveal adoption group of participants in their studies (Christofi et al., 2023; Malodia
differences and other issues. The UK has the most healthcare technology et al., 2023). This has also been accomplished through the use of
AI start-ups in Europe (ResearchAndMarkets.com, 2020). However, comprehensive essay questions (Chaudhary et al., 2022). Notably, no
adoption is continuing, and its potential, coverage, and healthcare ad­ negative impact on the data quality or the depth of the findings has been
vantages are unclear (Mckinsey Global Institute, 2019). The study aims documented.
to capture these agendas, fostering a conversation regarding the adop­ All participants are affiliated with diverse and pertinent groups
tion of AI and its resulting impacts. Fig. 1 depicts the research design, operating within various medical domains, including but not limited to

5
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Table 2
Description of various drives, barriers and dilemmas identified in the past literature.
Research profile Purpose of the study Drivers Dilemmas Barriers

Authors: Kooli and Al To examining the ethical – Patient privacy, biased results, –
Muftah (2022) dilemmas associated with AI in patient safety and system errors
Method: Narrative the healthcare sector are some major ethical
review method dilemmas that are likely to be
focusing on faced once AI will be introduced
content analysis in healthcare
Sample: 22 Studies
Author: Merhi To examine the most significant – – Data quality, Transparency, Privacy
(2023) barriers to responsible AI. and security, Skill full workers, Justice
Method: Analytical and fairness, Organisational culture
hierarchy
Sample: 7 experts
Author: Raghavan This article analyses the Contextual factors such as – Contextual factors relate to lack of
et al. (2021) healthcare AI based cloud demographics, economic and infrastructure and technology
Method: adoption drivers and constraints innovation mindset of users. Policy readiness. Organisational barriers like
Conceptual in Japan, South Korea, and factors like promoting digitalisation, cost and technical expertise. Policy
Sample: NA Singapore. data related cloud policies and cyber barriers related to lack of data
security laws. Human factors including standardisation and privacy issues.
trust and acceptance
Author: Kalisz et al. The goal is to examine how Functional benefits. Interaction and – New product ambiguity related to
(2021) healthcare professionals evaluate co-creation benefits. Automation design and features. Technical
Method: Delphi and handle categorical ambiguity benefits. High accessibility, value and challenges and legal issues. Lack of
technique features of social robots in considered cost effective. trust and physical or mental risk.
Sample: 11 experts healthcare sector.
of social robots
Author: Al Badi et al. This study's aim is to evaluate the – – Using AHP method, main challenges,
(2021) challenges of integrating AI in identified were Accuracy, privacy and
Method: semi UAE healthcare. security followed by trust,
structured interpretability and lack of control.
interviews and
AHP technique
Sample: 27
managerial level
executives
Author: Tseng et al. This paper discusses the benefits Improving care management, – Physicians' concerns: Lack of AI
(2021) and disadvantages of screening capacities, outreach in guidelines for serious eye diseases.
Method: implementing AI in Ophthalmology. High Cost. Lack of quality evidence
Sample: ophthalmology treatment from Reduce workload and stress of medical and trails results. Complexities with
both the provider and patient staff system integration. Trust and
perspectives. generalisability issues.
Patients concerns: Data and privacy
risk, lack of knowledge and literacy.
Lack of human interaction and techno
stress.
Author: Laï et al. An overview of how French health Competitive advantages A dilemma involving balance Legal and regulatory issues
(2020) professionals interpret the AI integration into practice among health, social justice, AI Interference between doctors-
Method: semi emergence of AI and how other In Depth results due to big data and ML and freedom. patients relationship
structured stakeholders perceive the issue. involvement Liability concerns
interviews
Sample: 40 French
stakeholders
Author: Ajmera and This article aims to investigate the – – Risk of job disruption, privacy, heavy
Jain (2019) barriers to Health 4.0 adoption in capital investment, skilled labour
Method: Review of India's healthcare sector. requirements, legal issues, lack of
Literature and R&D, lack of stakeholder support,
Delphi technique behavioural apprehensions and others.
Sample: 15 experts

paediatrics, ENT, biolabs, biochemistry, emergency departments, psy­ medical services, and other relevant factors, we were able to gain in­
chiatry, mental health, oncology, endocrinology, genetics, neurosci­ sights into their varying opinions and feedback regarding AI services in
ences, and other relevant fields. As a result of their diverse medical different medical fields. The study's relevance and trustworthiness are
backgrounds, they have acquired extensive firsthand experience in uti­ enhanced through the generalisation of its findings. Demographic de­
lising AI-based technologies. This also facilitated the investigation of tails of the respondents are mentioned in Table 3.
several procedural, diagnostic and medical advantages and obstacles to After reviewing the extant literature and theoretical frameworks
AI-based services across diverse departments. Moreover, the incorpo­ using a grounded theory-based approach, the study developed
ration of input from several medical organisations will augment our comprehensive list of open-ended essay questions on the adoption of AI-
comprehension of the anticipated outcomes and technological obstacles enabled technologies in healthcare (Martinez-Millana et al., 2022;
faced by various medical divisions in the integration of AI-driven ser­ Reddy et al., 2020; Wiljer and Hakim, 2019; Chen et al., 2021). To
vices. The individuals involved in this research exhibit considerable ensure the reliability and validity of the questions, we consulted five
expertise in both the public and private healthcare domains in the medical practitioners specialising in a variety of medical fields and,
United Kingdom. By gathering perspectives from a wide range of re­ based on their feedback, modified, combined, and eliminated some
spondents based on their occupations, genders, departments, type of questions. Next, a preliminary investigation involving three medical

6
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Fig. 1. Research design process.

staff members or doctors was carried out, and their feedback was questions. Authors analysed participant discussions, clarified varying
incorporated into the final list of open-ended essay questions for the observations, and concluded on various themes and concepts. The pro­
study. We conducted the final data collection on Prolific Academics cess of building the data structure encourages researchers to think about
platform. All the participants received a set of pre-defined questions that the data conceptually, which is important for conducting qualitative
encouraged them to share their perspectives on the future of artificial research and maintaining the rigour. We have utilised step by step
intelligence (AI)-enabled health technologies and applications. Also, abstraction procedure to ensure objectivity and improve reliability of
they were asked to share their experiences and how they were involved the findings. A panel of two independent researchers manually reviewed
with the service. They were also asked to talk about some key challenges and examined the participants responses and tried to labelled those re­
with using AI-enabled services as well as the drivers of using the service sponses to identify the key themes. First the researchers aimed to
in different medical fields. We also requested that participants provide identify key drivers and barriers with the use of AI-based technologies in
detailed responses to the predefined questions focusing on the key di­ healthcare. Next, the researchers analysed the data sentences by paying
lemmas faced by medical practitioners when using AI-based technolo­ attention to language that indicated paradoxes or ethical dilemmas and
gies. We also asked for their ideas on how to make AI-enabled services in contradicting views associated with the use of AI based services. For
advanced medical care better and more useful. All these questions were example: the study identified sentences like, “however there is a risk*”,
asked using paradox theory as a lens to investigate these issues. Open- “at the same time may recommend an incorrect scan *”, “agree or
ended questions comprised of nine questions involving various aspects disagree with AI*”, “Will it actually improve the care of my patients?*”
of adoption and use of AI-enabled technology/applications in healthcare and so on. Finally, solutions or strategies suggested in the participants
(see Annexure I). responses were identified and analysed to identify key AI adoption
strategies in the healthcare sector.
3.2. Data analysis In order to accomplish this, open codes were derived from the re­
sponses. The method defines terms, phrases, and descriptions provided
The study used an iterative analysis procedure, using the method­ by respondents to a specific research question as zero order codes. To
ology proposed by Gioia et al. (2013) to analyse the qualitative re­ identify these codes, the study reviewed all the replies line by line for
sponses and apply thematic coding process (Dhir et al., 2017). The data each question. In total 103 zero order codes (e.g., ‘clear and smooth
in the study was gathered, organised, categorised, and subsequently processes’, ‘data security concerns’, ‘fear of utilisation’, ‘cost’, ‘lack of ac­
arranged thematically, providing a structured foundation for a mean­ curacy’, etc.) were identified. Next, with the help of another researcher,
ingful understanding. The Gioia technique not only helps in data these codes were then segregated into broader codes (used for catego­
structure design but also gives research dimensions to answer research rising data) before being synthesised based on similarities and common

7
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Table 3 Table 4
Demographic details. Data structure details.
Respondent details Frequency Percentage First order concepts Second order themes Aggregate
dimensions
Gender
Male 24 38.7 % • High response time. Ease of use Drivers
Female 38 61.3 % • Less errors.
Age • Easy to understand and smooth
Less than 25 years 13 21.9 % processes.
26–35 years 15 24.2 % • Effective utilisation of information.
Over 35 years 34 54.9 % • Makes medical procedures much
Sector easier.
Public health sector 42 67.7 % • Improved efficiency. Effectiveness of
Private health sector 20 32.3 % • Save time in detection, prevention, automation
Occupation and treatment.
Doctor 12 19.35 % • Double the productivity.
Nurse 11 17.74 % • Improve multi-tasking capabilities.
Health care assistant 9 14.52 % • Automated monitoring of treatment
Medical engineer 7 11.29 % and medical services.
Emergency medical technician 3 4.84 % • Accurate diagnosis due to less Accurate diagnosis
Biomedical scientist 2 3.23 % invasive and time saving procedures.
Healthcare consultant 2 3.23 % • Advancement in deep learning
Occupational therapist 2 3.23 % processes and treatment algorithms.
Others 2 3.23 % • Improved functionality.
Acupuncturist 1 1.61 % • Improved medical practitioners'
Audiologist 1 1.61 % efficiency.
Business analyst 1 1.61 % • Reduced working hours due to Cost efficiency
Case manager 1 1.61 % procedural effectiveness.
Dentist 1 1.61 % • Reduced cost of some treatments.
Genetic counsellor 1 1.61 % • Need of a fewer medical staff.
Lab specialist 1 1.61 % • Reduced operational cost.
Mobility specialist 1 1.61 % • Patients' data misuse by hospitals. Privacy and trust Barriers
Ph.D. 1 1.61 % • Data security concerns. issues
Psychologist 1 1.61 % • Lack of patients trust issues due to
Social worker 1 1.61 % misinformation.
Vaccinator 1 1.61 % • Fear of unknown glitches in system
Experience (years) integration.
3 to 5 34 54.83 % • Lack of privacy invasion guidelines.
5 to 10 14 22.58 % • Lack of stakeholders' support.
10 to 20 8 12.90 % • Lack of pertinent information and Unsuitable for
More than 20 6 9.67 % knowledge of the technology. critical illnesses
• Lack of consistency and accuracy in
results.
themes, thereby converting them into 59 first-order concepts (e.g., • Not suitable for personalised and
‘cultural and religious factors’, ‘lack of privacy invasion guidelines’, ‘high unpredictable needs.
• Human expertise is missing.
response time’, ‘improved efficiency’, etc.). First order concepts were • Not useful for critical treatments and
refrained from drawing specific categories. As the study progressed, the mental health issues.
study derived second-order themes from the first-order concepts, pro­ • lack of training about a new Lack of training and
ceeding from comments to theorisation. We created meaningful clusters disruption. education
• Low literacy and awareness about the
of each term and phrases and categorised them in various categories,
use of AI.
thus, emerged as second order constructs. These themes were analysed for • Lack of education to the patients
the common themes to observe the particular concept such as ease of about the benefits.
use, cost, privacy issues, behavioural outcomes, and so on. These • Lack of information about
second-order themes were combined to generate aggregate dimensions, communicating with AI technology.
• Lack of information about job
which are more specific final concepts. The resulting structure comprise security.
of 103 zero order codes, 59 first order themes, 11 s order themes and 4 • Lack of knowledge about data
final aggregate dimensions that are named as ‘Drivers’, ‘Barriers’, cleaning and outcomes.
‘Ethical dilemmas’ and ‘Strategy for adoption’ (Table 4). • Lack of education on complex
designs.
• High investment and maintenance Financial constraints
4. Findings cost.
• Lack of financial aid in the system.
Paradox theory examines tension in an organisation associated with • Fundings constraints in hospitals.
new technology adoption and aims to manage paradoxes by finding a • Difficulty in convincing investors.
• Cultural and religious factors. Behavioural
balance between conflicting objectives or decisions (Raisch and Kra­ • Unwillingness of patients and older. resistance
kowski, 2021). Based on the literature, we adopted technological par­ Population to trust AI.
adoxes perspectives as the frame, as it helped us understand different • Lack of human care and wellbeing
paradoxes or viewpoints of medical practitioners on AI adoption in provided by AI.
• Healthcare equity issue.
healthcare settings. Furthermore, the current study employs paradoxical
• Workload will increase.
theoretical assumptions to identify several dilemmas for medical prac­ • AI may introduce biases between
titioners related to the contradictory yet interconnected roles of AI, certain groups.
which develop inconsistent perspectives about the service. In the • Lack of interest and attitude of
following, the present study identified four drivers of AI technology medical staff..

adoption: ease of use (Lee and Lee, 2021), effectiveness of automation (continued on next page)

8
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Table 4 (continued ) discussed in the study. Fig. 2 presents the summary of the findings.
First order concepts Second order themes Aggregate
dimensions 4.1. Drivers
• Ethical versus pragmatic decisions. Ethical dilemmas Ethical
• Shut down versus implement them dilemmas The findings include four key drivers to the adoption of AI enabled
universally. technologies in healthcare settings.
• Human touch versus technological
advancement.
4.1.1. Ease of use
• AI results need to be verified and by
whom? The first driver is ease of use, which represents the convenience of
• Autonomy of patients versus using AI-enabled technologies for medical care, maintaining records,
treatment benefits. and faster diagnostics. Ease of use is considered an important predictor
• Assessment of benefits versus harms of AI adoption in many past studies (Laï et al., 2020; Kumar et al., 2021).
based on individual needs.
• Need for patients' protection policy. Strategy for AI Strategy for AI
It leads to higher satisfaction that eventually enhances initial and
• Need for stakeholders' financial adoption adoption continued adoption of a services (Alhashmi et al., 2019). In the context
support. of AI adoption, ease of use is considered important for the effective
• Need for more advanced research on utilisation of the service (He et al., 2021). Our findings revealed that the
the use of AI.
ease of use of AI-enabled technologies allows medical staff to handle
• Need for technical and operation
support. large patient data sets and offers ease of service (Hercheui and Mech,
• Need for creating awareness and 2021). One perspective suggests that AI technologies have a primary
education. focus on patient well-being, offer enhanced convenience when dealing
• Role of administrators and with extensive data sets, and accomplish data sequencing more effi­
management is crucial.
ciently (Sipior, 2020). Conversely, it can be argued that these ad­
vancements in technology result in reduced invasiveness and increased
(Reddy et al., 2020), accurate diagnosis (Wong et al., 2019), and cost transparency of processes, leading to enhanced outcomes for patients
efficiency (Laï et al., 2020). Also, five key barriers were identified in the (Di Vaio et al., 2020). Respondents explained that AI involves less
adoption process: privacy and trust issues (Tandon et al., 2020), being invasive procedures, which make things much easier. Further, it is
unsuitable for critical illness (Khanijahani et al., 2022), behavioural observed that it makes medical services much faster and easier for pa­
resistance (Cubric, 2020), funding constraints (Nadarzynski et al., 2019; tients, particularly for those who are inclined towards new technology
Siala and Wang, 2022), and a lack of training and knowledge (Bughin usage. Thus, it will continue to grow as generational care shifts. Consider
et al., 2019). Further to this, the study explained ethical dilemmas or the following interview passages from the participants:
conflicting views of medical staff on AI services on patient autonomy
and privacy of data; the role of AI and human expertise in advanced ‘The AI technology is useful as it provides ease of access to more pertinent
treatments; ethical or pragmatic AI-based decisions, behavioural resis­ information that will further the prevention of major diseases which is a
tance, and many others. Following that, while adopting AI-enabled major positive improvement in the healthcare field.’
technologies, strategies for AI adoption related to stakeholder support (P6, Female, 34, UK, Lab Specialist, Public Healthcare)
(Kumar et al., 2021), legal protection (Reddy et al., 2020), technical
support (Laï et al., 2020), and AI education (Lee and Lim, 2021) are also

Fig. 2. Summary of present findings.

9
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

‘Ease of it's use. It makes the job a thousand times easier in all the ways different diagnosis and provide unbiased recommendations. It is further
that make it a thousand times more annoying, aka documentation and discussed that the use of AI reduces biases and prejudices in the clinical
avoiding lawsuits for malpractice.’ processes and improves data accuracy for better medical treatments.
(P34, Male, 28, UK, Physician, Public Healthcare) Few respondents stated that:

‘It is done to be more accurate and comprehensive It is meant to minimise

4.1.2. Effectiveness of automation
human error and also streamline the accuracy of the analysis. The ap­
The second driver identified in the present study is the effectiveness
plications are designed to better diagnose accurately and more quickly
of automation of AI-enabled technologies. Automation is explained as a
than previously done by a person.’
technique that allows a system or service to operate or control various
(P33, Female, 63, UK, clinical laboratory technologist, Public
functions automatically. Reddy et al. (2020) have posited that the
Healthcare)
integration of AI into system has resulted in notable enhancements in the
effectiveness and operational capabilities of healthcare delivery. The
‘It can aid doctors in making more accurate and reliable diagnoses and
utilisation of automated features has facilitated in many medical treat­
can also reduce the time taken to make these diagnoses. My personal
ments, diagnostic procedures, and clinical care settings inside hospital
experience with AI is the use of observation frequencies for inpatients
environments. In the current study, respondents talked about procedural
being automatically changed by algorithms.’
and communication effectiveness as a result of AI-enabled automation in
healthcare. The study discussed the workflow automation done by AI to (P43, Male, 24, UK, healthcare assistant-surgery, Public Healthcare)
perform routine tasks and allow physicians to focus on critical medical
cases. AI-enabled systems have improved the efficacy and functionality 4.1.4. Cost efficiency
of healthcare delivery. Due to its automated features, it is suggested that The cost of technology is defined as the cost associated with the
it can be used to streamline processes and save time and energy for development, deployment, and implementation of technology. The
healthcare workers for several medical treatments, diagnoses, and findings suggest that cost is directly associated with the treatment pro­
clinical care settings (Laï et al., 2020). A few of the respondents quoted: cedures and patients' long-term outcomes (Bamel et al., 2023; Dicuonzo
et al., 2023). This could be controlled by using well-tested and accurate
‘AI could also free up healthcare workers from rote/administrative tasks technology in the healthcare system to reduce operational and routine
to allow more time for direct patient care.’ costs. According to the present findings, AI algorithms may help in
(P31, Female, 48, UK, Geriatrics, Public Healthcare) obtaining more precise patient-related medical facts. According to Char
et al. (2020), this methodology serves to authenticate and forecast future
‘This technology could help alleviate staffing shortfalls, as fewer spe­ results, hence increasing the probability of mitigating the long-term
cialists would be required to process and investigate patient scans. Of financial strain on patients (Vučinić, 2020). According to Guan
course, this is dependent on the AI being proven to be accurate enough to (2019), AI techniques can be utilised for conducting clinical trials of
be run autonomously.’ drugs, using deep learning techniques for planning, process and data
(P59, Male, 41, UK, Business Analytics-NHS, Public Healthcare) collection. This may reduce the operational cost and save time. Re­
spondents talked about how AI services improve procedural effective­
At the same time, the study explained automated benefits of AI
ness, speed up the process, and save costs. On the other hand, AI
system in improving personalised care services and providing self-
techniques can be utilised for conducting clinical trials of drugs, using
diagnose settings to patients (Cubric, 2020). AI has a potential to pro­
deep learning techniques for planning, processing, and data collection.
vide personalised health services by analysing personal records of pa­
This may reduce the operational cost and save time. Studies recommend
tients and offering personalised medical options. Consider the following
that AI diagnostics can be deployed at a lesser cost in comparison to the
interview passages from the participants:
cost associated with healthcare workers (Dicuonzo et al., 2023).
Consider the following interview excerpts from the participants:
‘I see it being used for patients to self-refer and self-diagnose issues/
problems without needing to see a professional face to face. Automated ‘Simplification, accuracy and efficiency are the main ones but also cost
ways of communicating with patients to provide appointment reminders, savings and elimination of manual labour activities.’
test results etc.’ (P46, Male, 46, UK, Procurement, Private Healthcare)
(P36, Female, 43, UK, Midwife, Public Healthcare)
‘There are potential cost savings to be made in batch processing of clinical
‘Less things are missed with some AI in my opinion. Because it's computer
or scan data and there may be cases where AI systems are more accurate
based, all scenarios can be automatically evaluated via AI.’
at detecting abnormalities than a human eye.’
(P14, Female, 28, UK, medical technician manager, Private Healthcare)
(P58, Male, 52, UK, Academic psychologist and Research Lead, NHS,
Public Healthcare)
4.1.3. Accurate diagnosis
AI enabled technologies are perceived to be accurate due to its effi­
cacy to assess complex data such as radiology images, patients scan with 4.2. Barriers
less errors and time (Dicuonzo et al., 2023). Accuracy of technology is
explained as its capability to measure and interpret the information The findings further revealed five key drivers to the adoption of AI
correctly (Al Badi et al., 2021). According to recent research findings, enabled technologies in healthcare settings.
advancements in deep learning techniques, algorithms, and procedures
have led to enhanced capabilities of AI-enabled devices in medical 4.2.1. Privacy and trust issues
diagnosis and therapies (Mirbabaie et al., 2021; Al Badi et al., 2021). In AI analyses enormous amounts of complex data gathered from a
the present study, participants discussed about the accuracy of AI results variety of sources about a patient's past and current medical information
in diagnosis, which is delivered in the required time period. AI reduces (Raghavan et al., 2021). After obtaining the patients' approval, these
the chances of human error that increases the reliability and accuracy of data are used for diagnosis and subsequent treatments. Nevertheless,
medical recommendations by medical professionals. A few suggested scholarly investigations have expressed apprehensions regarding the
that AI enabled technology assist medical staffs and doctors with entitlement of physicians to utilise said data, safeguarding patient
confidentiality, and the efficacy of legal frameworks (Wiljer and Hakim,

10
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

2019). Several studies revealed the existence of privacy issues related to 4.2.3. Lack of training and education
AI's use of data. This affects the level of trust individuals have as well as Major AI services developed so far use complex deep learning algo­
their willingness to adopt AI technology (Di Vaio et al., 2020; He et al., rithms and techniques (Alhashmi et al., 2019). The findings suggested
2021). According to current research, one of the primary reasons for that to effectively communicate the benefits of AI to patients, medical
mistrust is patient data privacy. Concern is raised about the accessibility professionals must first understand the technology, which is still
of the data and the lack of certainty that it is only accessed by those with evolving due to resistance, old habits, and other behavioural issues
valid medical reasons. Additionally, when working with large pop­ (Talwar et al., 2023). These are the key challenges highlighted in recent
ulations and sensitive data, it is indicated by the respondents that AI studies (Alhashmi et al., 2019). Hercheui and Mech (2021) have raised
systems can make inaccurate predictions. Respondents discussed these similar worries over the insufficient education and training of medical
trust concerns and expressed their doubts about employing AI services teams in relation to AI systems, which subsequently impacts their
because of the possibility of system errors when processing patients' inclination to embrace such systems for enhancing care management.
data. It is observed that medical staff are concerned and feel that the AI Respondents expressed similar concerns about the lack of medical teams
could interfere with a person's right to privacy and confidentiality. education and training on the AI system, which affects their willingness
Consider the following interview excerpts from the participants: to adopt it for improving care management. The present study found
that healthcare workers struggle to learn the system. Participants also
‘Privacy confidentiality of data. who can look at what? difficult to ensure agreed that there is a need for proper training to detect severe illnesses
data is only accessed by those with a valid reason when AI can flag up and critical cases using AI. A slow medical adaptation rate due to various
issues in old records. understanding of limitations and capacity - what is it behavioural and cognitive inabilities was also discussed. Consider the
good for/no good for.’ participants' responses:
(P44, Female, 37, UK, Clinical Engineer, Public Healthcare)
‘Sometimes AI machines and devices can be complex and require training
‘One massive problem within AI in healthcare is there is very little trust in to learn how to use. This is an issue because some health cares will
AI systems to miss something that a human wouldn't. This problem is due disregard this necessary training and not know how to use AI devices
to the system not being as nuanced as a human can be, particularly with which can lead to detrimental effects.’
difficult to interpret scans etc.’ (P12, Female, 19, UK, pharmacy technician, Private Healthcare)
(P55, Male, 24, UK, Biomedical scientist, Public Healthcare)
‘Lack of training especially as so much of the workforce is only part time
or are agency/bank workers so are not always working in the same
4.2.2. Unsuitable for critical illnesses
settings.’
Deep learning techniques capture time-sensitive patient data (Shi
(P36, Female, 42, UK, midwife-Obstetrics, Public Healthcare)
et al., 2020). These methods are perceived as superior to established
ways for assessing critical illnesses and anticipating patterns. However,
the present findings expressed concerns about AI's efficacy in evaluating 4.2.4. Financial constraints
critical illnesses or patients with risky medical histories (Jussupow et al., The use of AI in healthcare is changing in various domains such as
2021; Tandon et al., 2023). Respondents indicated that cases like mental patient care, scanning, radiotherapy, medical treatments, and hospital
health or psychiatric issues demand more professional opinions since administration (Laï et al., 2020). But implementing and deploying new
complex results occur and human involvement is required in place of AI technological innovations across departments requires funds (Sharma
that follows patterns (Wiljer and Hakim, 2019; Tseng et al., 2021). et al., 2022). Several studies have found that funding constraints in the
Moreover, in cases where the patient presents with a complex medical healthcare sector inhibit the collection and analysis of sufficient data for
background, a greater level of human knowledge becomes required, as meaningful information (Garbuio and Lin, 2019; Vučinić, 2020).
the process of diagnosis may necessitate comprehensive assessments Financial support is highly required for long-term drug research, AI-
(Jussupow et al., 2021). Respondents also indicated that AI can be un­ based drug trials, and other medical innovations. In the present study,
reliable as it may stop working, not evaluate all information, or come to participants noted insufficient funds in the healthcare sector to support
wrong conclusions. It may also be very clinical in its' approach, fore­ and implement AI-enabled services (Bamel et al., 2023). It is observed
going the compassionate human component that is critical when eval­ that there is a need for financial aid to implement AI services in medical
uating critical cases. The majority of respondents stated that AI was settings, which is presently missing in the sector (Laï et al., 2020). Some
inefficient in managing severe medical cases or scenarios, requiring participants were concerned about the large, long-term, and largely
communication with patients and families (Balagurunathan et al., irrevocable investments in AI technology, with extremely high invest­
2021). Simultaneously, participants expressed the need for human ment costs for some AI applications (Garbuio and Lin, 2019). Also, the
involvement to validate AI-based results and identify future actions. primary difficulty in implementing AI in healthcare is its financing,
Following are related interview excerpts from the participants: which must cover the initial outlay. This may influence the decision, and
if AI-based technology is too expensive, the healthcare industry may
‘For patients with complex medical histories it becomes more of a hin­ choose not to invest in it. Consider these quotes from participants:
derance than a benefit. I think it would be more useful if we could more
easily modify the algorithm for the AI in the patient's chart to account for ‘Funding is the man issue. We don't have the funds in order to recruit the
their complex medical histories.’ use of applications or technology. The next issue would be learning how to
(P62, Female, 25, UK, medical assistant-cardiology, Private healthcare) use the tech or apps in healthcare; I imagine there would be resistance
from patients and maybe even staff.’
‘the technologies automatically release patient results in the laboratory, (P51, Female, 25, UK, Doctor-emergency, Public Healthcare)
but often need to be checked for consistency and errors. when there are
outliers or variances, then the technology may become confused. There is ‘The cost of technology and research and experimentation is the biggest
still the need for human checks and balances.’ barrier. The current NHS is already struggling with funding and justifying
(P33, Female, 63, UK, Clinical laboratory technologist, Public spending money on AI technology which may take several decades to
healthcare) become in a state of use rather than spending money on helping patients in
the present is another barrier faced by the healthcare sector.’
(P56, Male, 25, UK, Doctor-cardiology, Public Healthcare)

11
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

4.2.5. Behavioural resistance ‘When dealing with things like genomics and personalised medicine in
New technology resistance is not on the research agenda. Individual general, we must make sure that this data doesn't seep into and influence
resistance to new technology is defined as ‘opposing action or delaying things like health insurance coverage decisions.’
the use of new technology’ (Oprescu et al., 2022). Current studies in the (P4, Male, 35, UK, Healthcare consultant, Private Healthcare)
field of healthcare have investigated the phenomenon of patient resis­
tance towards the use of AI for diagnostic and therapeutic purposes (Laï ‘Ethical dilemmas may include privacy, because technology can often
et al., 2020; Koo et al., 2021). Tandon et al. (2023) also assert that utilise a lot of information about a person's life and health condition.
physicians exhibit a hesitancy towards adopting novel technological Patients may not feel comfortable with having artificial intelligence
advancements, primarily attributed to a dearth of adequate education running their healthcare plans, and may prefer a human to conduct most
and training. The present study examined users and patients' resistance of their care.’
to using AI-based diagnosis and treatment. It is expressed that physicians (P17, Female, 22, UK, Nurse-Labor and Delivery, Public Healthcare)
are reluctant to use new technology due to a lack of awareness and
education. A few respondents indicated high cultural and religious ‘Practitioners may encounter situations when in their professional
resistance among staff and users to using AI services. They hinted that judgement they disagree with the AI and must make ethical decisions.’
cultural differences, religious resistance to using technology and sharing (P1, Female, 25, UK, Occupational Therapist, Private Healthcare)
personal information, etc., all affect AI adoption in health settings. When
implementing AI systems in healthcare, respondents expressed concerns
about a lack of personal or human touch, cultural and religious factors, 4.4. Strategy for AI adoption
resistance to change, and job insecurity (Kumar et al., 2020). Consider
these participant quotes: Participants have suggested a number of strategies as potential ways
to effectively implement and adopt AI-enabled healthcare technologies
‘Healthcare has the expectation of a human touch/experience, some and services. In order to safeguard the confidentiality of patients' data in
groups of people do not and will not trust technology due to abstract fears the context of utilising AI, it is imperative to implement novel policies,
and lack of experience and misinformation. protecting sensitive data will laws (Jussupow et al., 2021; Ma et al., 2023), or modifications to current
not be preferred by these patients.’ legislation (Sharma et al., 2022). The study identified a few areas that
(P13, Female, 37, UK, physician network CX designer, Private require attention and advancement for AI adoption. One such area is
Healthcare) medical staff training and knowledge. This includes training that in­
‘key dilemmas probably include resistance to change since this is not how volves understanding of AI structure, input data requirements, and
we were trained. Getting used to new technology after some people are not interpreting patient outcomes (Shi et al., 2020). Another field of
technically inclined, not everyone may agree with it ethically, there may research mentioned is improving AI's alignment with human intelli­
be legal challenges which may arise from any error that results.’ gence for critical judgements. Some further expressed their concerns
about the lack of technical, human, and financial resources, which are
(P22, Female, 25, UK, medical-RN, Public Healthcare)
highly required to enable the role of AI technology (Wong et al., 2019;
Cubric, 2020). The role and support of stakeholders and administrators
4.3. Ethical dilemmas are found to be crucial in the study. Consider the following participant
quotes:
The current findings on patient data privacy, AI accuracy resulting
from the utilisation of extensive and intricate data sets, and the tech­ ‘AI needs technical support on the back end and through development:
nological advancements associated with AI that impact outcomes have strong network connections, data recall accuracy, and a more neutral
given rise to several ethical concerns (Hercheui and Mech, 2021; Jus­ starting point that reduces engineer or data input prejudices and bias in
supow et al., 2021). Past studies supporting the present study revealed programming caused by existing data entry from human engineers and
that users are perplexed with conflicting outcomes, such as the use of programmers.’
advanced AI algorithms that may bring medical breakthroughs and lead (P13, Female, 37, UK, physician network CX designer, Private
to undiscovered solutions (Raisch and Krakowski, 2021). Similarly, on Healthcare)
the other hand, some respondents expressed that there is a risk associ­
ated with using large patient data sets that may arise when the data is ‘Some healthcare facilities may not be able to afford the implementation
not sufficient or unrepresentative. This may lead to a dilemma as to how of such AI enable technologies and may lack the financial backing from
accurate these results are and how they will influence outcomes. outside sources. In an effort to assist in receiving these technologies
Another ethical challenge indicated in the present study is related to healthcare facilities may have to reach out to their stakeholders, other
data privacy issues with AI, as it contains patients' medical histories and financial institutions, or patrons to assist in purchasing these
sensitive information. There are concerns about data misuse and the technologies.’
sharing of private information without consent (Wiljer and Hakim, (P7, Female, 39, UK, Nurse-urgent care, Public Healthcare)
2019; Di Vaio et al., 2020; He et al., 2021). This may lead to a dilemma
or privacy concerns while using AI services. Regarding the significance
of human interaction versus AI in healthcare contexts and the extent to 5. Concluding discussion
which technology can replace it, a number of respondents also voiced
their ambivalence. The study explores the concept of paradox using paradox theory and
Along with these, a few other dimensions such as patients' autonomy, the use of new technology adoption in organisational settings (Chen
technological discomfort or resistance, unethical labour practices and et al., 2021; Kalisz et al., 2021). The use of paradox theory is relevant in
treatment with medical staff, biases, and many others are discussed in the present study as it helps to understand the paradox of technological
the present study (Tandon et al., 2023). Consider these participant development (the use of AI) in healthcare settings. In this regard, the
quotes: study observed the diverse perspectives of medical practitioners about
the use and efficacy of AI services, as well as some emerging ethical
dilemmas associated with the implementation of AI in different health
domains. For this, the study used a qualitative research design and an
inductive analysis of open-ended essay questions. The study identified a

12
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

list of paradoxes and their linkage with paradox theory: inform the process of ethical decision-making by recognising and
resolving inherent conflicts, hence guaranteeing the ethical and
• Paradox: Patient Autonomy and Privacy of data responsible application of AI. The paradox hypothesis promotes
o Data Paradox: The acquisition of comprehensive patient data is the acceptance of ethical paradoxes among stakeholders, wherein
imperative for the optimal functioning of AI systems, as it enables they recognise the presence of several ethical norms that may
them to make informed decisions and perform accurate predictive contradict with one another, and strive to include them into AI
analytics. This leads to the emergence of a paradoxical situation systems. It fosters the facilitation of open discourse and the
wherein there is a conflict between the necessity for data acces­ development of moral frameworks aimed at resolving these
sibility and the obligation to safeguard patient confidentiality and dilemmas.
adhere to regulatory frameworks. Striking the right balance is • Paradox: behavioural resistance, and many others.
deemed crucial. o Healthcare organisations frequently encounter the paradoxical
o Application: The use of paradox theory can provide guidance in challenge of simultaneously pursuing innovation and integrating
the creation of AI systems that effectively navigate the delicate AI technology to enhance patient care and operational efficacy,
equilibrium between safeguarding data privacy and facilitating while also complying to stringent regulatory mandates. Healthcare
data exchange. This may entail the utilisation of methodologies organisations are perpetually engaged in the pursuit of innovation
such as federated learning or differential privacy to safeguard and the integration of novel technologies such as AI to enhance the
patient data while concurrently capitalising on the advantages of quality of patient care and optimise health outcomes. Neverthe­
shared data. less, it is imperative for them to prioritise both patient safety and
data security. This situation gives rise to a dilemma wherein in­
• Paradox: AI and human expertise in advanced treatments dividuals or organisations must concurrently adopt and foster
o The Efficiency Paradox: This paper explores the Efficiency innovation while also effectively mitigating associated dangers.
Paradox associated with the integration of AI in healthcare, o Application: The use of paradox theory facilitates the identifica­
wherein the utilisation of AI technologies can lead to both positive tion of an optimal equilibrium between innovation and regulation,
and negative impacts on efficiency. The present discourse high­ so assuring that AI solutions possess both state-of-the-art capabil­
lights a paradox that exists within the healthcare industry, spe­ ities and adherence to healthcare standards. The application of
cifically pertaining to the delicate balance between the imperative paradox theory can assist healthcare professionals in effectively
for efficiency and the potential consequences of dehumanising managing this situation by highlighting the significance of
patient care. Nonetheless, a conundrum arises when considering achieving equilibrium between two ostensibly conflicting objec­
the juxtaposition between the imperative for optimal efficiency tives. Organisations are encouraged to take a dynamic approach,
with the potential consequence of depersonalising the provision of wherein they continuously modify their plans in response to
healthcare to patients. Excessive dependence on AI might poten­ evolving conditions.
tially result in the erosion of interpersonal connections and the
deterioration of the rapport between patients and healthcare Next, in response to RQ1, the study observed that there are four main
professionals. This paradox underscores the tension between the attributes influencing the perceptions of medical practitioners and en­
desire to optimise healthcare processes and the potential negative courages the adoption of AI services: ease of use (Alhashmi et al., 2019),
impact on the humanistic aspects of caregiving. The healthcare effectiveness of automation (Lee and Lee, 2021), accurate diagnosis (Al
industry is fundamentally centred around human beings, and the Badi et al., 2021), and cost efficiency (Char et al., 2020). Our findings
integration of AI may give the impression of diminishing the per­ indicated that AI-based services have a high response time and are less
sonal nature of exchanges. prone to errors (Kumar et al., 2021). Research on ease of use indicated
o Application: The use of paradox theory can provide guidance in that the variable explains the efficacy and convenience of a particular
the creation of AI systems that aim to provide standardised pro­ technology. This, in turn, has a good impact on performance, as it en­
cedures for general care, while also allowing for the customisation hances the ability to efficiently utilise information and facilitates
of therapies or suggestions for particular patients. Healthcare seamless execution of operations. (Khanijahani et al., 2022). At the same
professionals place a high level of importance on their autonomy time, automation reflects the reliability and efficiency of the technology
when making decisions. However, the use of AI may provide in saving time in procedural activities that enhance multi-tasking ca­
valuable benefits such as automation and decision assistance. pabilities and productivity (Mirbabaie et al., 2021). As per Guan (2019),
Achieving an optimal equilibrium can be a formidable task. The automated AI-enabled services improve the procedural effectiveness of
use of paradox theory can facilitate the development of AI systems the system and initiate automated monitoring of basic medical treat­
that enhance human skills in the field of healthcare delivery, while ments. This provides more time for practitioners to handle critical cases
still preserving the fundamental human element. and provide personalised care to patients, which improves critical
diagnosis productivity.
• Paradox: Ethical or pragmatic AI-based decisions – In order for AI Accurate diagnosis procedures has been identified as a third attribute
systems to achieve efficacy, it is imperative that patients and of AI enabled medical services. AI decisions are derived from large
healthcare professionals develop a sense of trust in these systems. conclusive analysis of facts, considering more advanced deep learning
However, scepticism and apprehension may arise due to concerns processes and treatment algorithms than traditional methods. This is
about the dependability and ethical implications of AI. The inte­ possible as AI collects all relevant and critical patient medical facts and
gration of AI in the healthcare sector gives rise to a range of ethical takes decisions entirely based on these facts. Such decisions are
quandaries, including algorithmic prejudice and the risk of deper­ considered to be accurate and improve medical practitioners' efficiency
sonalising patient interactions. as they are not influenced by their subjective judgements, emotions,
o Application: The use of paradox theory can provide valuable in­ biases, and other heuristic factors. The final positive attribute identified
sights into the development of strategies aimed at fostering con­ by respondents about AI-enabled technologies is their cost efficiency
fidence in AI systems. These strategies may include transparent (Dicuonzo et al., 2023). AI reduces operational and labour costs by
communication regarding the application of AI, the establishment minimising technical glitches and reducing working hours due to
of explicit standards for AI-assisted decision-making, and the avoiding procedural repetitions (Char et al., 2020). It is observed that
implementation of continuous monitoring and accountability advanced AI services may lower costs associated with a few clinical trials
mechanisms. The utilisation of paradox theory may effectively (Guan, 2019). Further, due to the fast treatment and diagnosis services,

13
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

it may reduce the cost of experts (doctors) as well as the cost of basic shared different views; a few were in favour of implementing it for
medical services (Dicuonzo et al., 2023). advanced treatment benefits (Dicuonzo et al., 2023); however, a few
In response to RQ2, the study observed some barriers to the AI researchers recommended it due to its potential side effects on indi­
adoption: privacy and trust issues (Raghavan et al., 2021), unsuitable to vidual needs and a lack of human touch and expertise (Jussupow et al.,
critical illnesses (Jussupow et al., 2021), lack of training and education 2021; Mirbabaie et al., 2021).
(Kalisz et al., 2021), financial constraints (Garbuio and Lin, 2019), and In response to RQ4, the paper suggests a few AI adoption strategies
behavioural resistance (Di Vaio et al., 2020). In regard to the issue of based on respondents' recommendations. The most important point is
privacy and trust, prior studies have expressed apprehensions regarding patients' legal protection and awareness of copyright and data privacy
the security of data, potential misuse of patients' data, the absence of issues, which are critical to understand (Jussupow et al., 2021). The
privacy guidelines, insufficient information, and other trust-related healthcare system should consider the legal and regulatory issues of AI
concerns associated with AI services (Cubric, 2020). The respondents deployment with more clarity on data usage, transferability, and sharing
raise concerns about the illegal or uninformed use of patient data, which policy rights (Ma et al., 2023). The study also identified the need to
might undermine patient privacy and healthcare system trust (Di Vaio educate medical professionals on utilising AI services. An issue raised by
et al., 2020). According to He et al. (2021), any violation of these rules Kalisz et al. (2021) is the lack of medical team learning in the AI system,
compromises patient confidentiality and weakens their trust in the which inhibits their acceptance of AI services to enhance clinical man­
system. The findings highlighted concerns about misinformation and agement. Participants also discussed the significance and support of
experience regarding patient data privacy guidelines, which leads to a internal stakeholders such as management, administrators, and practi­
lack of trust in the system. Next, it has been discovered that AI's auto­ tioners in providing technical and human expertise to manage data is­
matic feature may provide a great deal of ease to medical personnel. sues and errors (Ajmera and Jain, 2019). Also, managing AI-based
However, due to its reliance on patterns, AI is limited in its ability to studies and medical diagnostics requires technological, human, and
anticipate the highly particular and unpredictable requirements of in­ financial resources (Wong et al., 2019). Support for various stakeholders
dividual patients. According to Wong et al. (2019), the use of this is also highlighted by few of the medical practitioners (Wiljer and
technology may have adverse consequences for both patients and Hakim, 2019). A proactive role of external stakeholders including
medical practitioners. Its applicability is limited to non-urgent medical financial institutions, governments, and law firms for creating funding
examinations and therapies, as it is deemed inadequate for managing programs and providing financial aid for advanced AI research and AI-
severe medical conditions. Insufficient information and a lack of un­ based trials is also found significant by the participants (Martinez-
derstanding of the technology might also have an impact on the con­ Millana et al., 2022).
sistency and quality of the outcomes (Oprescu et al., 2022).
Another concern raised is related to lack of training and education of 5.1. Theoritical implications
staff using AI-based healthcare services (Laï et al., 2020). The study
reveals that this is primarily due to employees' unfamiliarity with AI The current study contributes to the literature on application of AI in
services and their ignorance of the training requirements for medical the health sector by offering insights from personal perspectives of
staff (Cubric, 2020). This increases the difficulty of using the service for health care providers adding significant value and relevancy to the
data cleaning and interpreting the outcomes for patients, who also lack current healthcare context. While many studies have examined the
information about the new system. The study reveals fourth barrier to useful uses and benefits of AI in healthcare, very few have examined the
service is lack of funding for AI-enabled technology which may be used viewpoints and experiences of healthcare professionals in this field. The
for drug research and personalised care services (Garbuio and Lin, present study addresses this gap by investigating the viewpoints of
2019). The use of AI in healthcare is changing in various and deploying healthcare providers on application of AI. The novelty of this research is
these innovations across departments requires funds. Several studies examination of health care provider's perspectives and viewpoints on
indicated that AI-based services have high investment and maintenance technological paradoxes in clinical setting. To identify potential bar­
costs, which are difficult to maintain due to funding constraints and a riers, allay concerns, and encourage the successful integration of this
lack of financial aid in the system (Garbuio and Lin, 2019). This is also technology, it is crucial to get insights into the opinions of healthcare
highlighted in the present study. The final barrier identified is behav­ practitioners about the efficacy and integration of AI. This study suc­
ioural resistance of users using AI services in healthcare (Norori et al., cessfully takes into account a broad range of perspectives with a vast
2021). Lack of interest and attitude, perceived high workload, and un­ array of healthcare experts. The study highlights the drivers, barriers
willingness to change were identified as major behavioural obstacles and strategies for wider adoption of AI in clinical setting. The analysis of
(Mirbabaie et al., 2021). Also, a few other common behavioural barriers healthcare professionals' perspectives on the effectiveness and use of
to AI adoption are a lack of human touch and wellbeing, as well as artificial intelligence on technological paradoxes is a developing area of
cultural and religious barriers (Garbuio and Lin, 2019). research that is novel and innovative.
In response to RQ3, the study revealed a few ethical dilemmas from In addition to concerns about patient privacy and data security, the
the perspective of medical practitioners while discussing the pros and problems also include issues of training and education, automation
cons of the use of AI services in healthcare. The study's primary ethical inferred job displacement, ethical ramifications, and cultural and reli­
concern is patients' autonomy versus advanced treatment benefits gious considerations. These revelations add significantly to the corpus of
(Mirbabaie et al., 2021). Issues of accountability and ambiguity sur­ research as they offer a deeper understanding of the barriers that need to
round AI-based outcomes, as there is a lack of clarity about how and to be addressed for the successful integration of AI in a clinical setting. It
what extent patient data will be utilised for the treatment (Kumar et al., makes it easier for us to comprehend the many concerns and reserva­
2023). The study raises another dilemma regarding ethical versus tions healthcare professionals can have about the application of AI
pragmatic decisions that explains concerns about human and non- technology. Moreover, this research output not only elucidates the ob­
human reliance on medical services as a crucial paradox (Oprescu stacles but also identifies the determinants that facilitate the imple­
et al., 2022). Medical professionals have emphasised the necessity of mentation of artificial intelligence in the healthcare domain. The main
establishing greater levels of confidence regarding the efficacy of AI factors that contribute to the adoption, as identified in the current study,
systems, their compatibility with human expertise, and their reliability include ease of use, effectiveness of automation, accuracy of diagnostics,
in making vital diagnoses. This is due to the fact that AI outcomes are and cost efficiency. With the use of this information, strategies may be
derived from patterns and predictability (Laï et al., 2020). Participants developed to support the effective application and use of artificial in­
further shared their views about its implementation on a global scale or telligence in healthcare settings.
total elimination; several contradictions were also raised. Studies also The study offers a few important insights into technological

14
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

paradoxes developed during the adoption of a new technology using 5.2. Managerial implications
paradox theory. The study explains the relevance of paradox theory for
the adoption of AI-based services and unravels medical practitioners' This research will help the healthcare industry better understand the
dilemmas and paradoxes related to the adoption of the service. drivers and barriers to implementing AI enabled solutions. Few impli­
Following the theory, we conceptualised technology adoption drivers, cations are identified. First, practitioners see that AI services improve
challenges, ethical dilemmas, and a few adoption strategies from the data administration and reduce administrative burden of medical staff.
viewpoints of medical practitioners, resulting in AI adoption in health­ Primarily, hospitals may use AI services for data cleaning and prioriti­
care. Subsequently, the study explained how the use of AI is progressing sation, allowing employees to focus on patient care. Second, while
in various health domains but argued that the use of AI is largely to looking at the current research's identified barriers that may impede the
improve operational efficiency and simplify administrative operations effective adoption of AI services. Hospitals should be aware of issues and
rather than to improve critical diagnosis and advanced care services. work towards improving patient and staff perception of AI services (Ma
These findings require additional research into critical medical treat­ et al., 2023). To gain patients' trust, informed consent and legal pro­
ments and the validation of AI-based services in these specialised tection of patients should be enforced. Moreover, hospitals must
healthcare domains. While some responses focused on concerns about continue to research and enhance AI capabilities for critical diagnosis.
protecting patients' privacy and sensitive information, no definitive Also, effective technology alignment is necessary (Laï et al., 2020). Staff
suggestions or standards were presented. Misuse of sensitive patient should be trained on how to use AI services and be involved in AI trials
data and the use of ineffective AI can undermine patients' trust. It has and research to improve their knowledge and perception.
been recognised that this is due to a lack of patient awareness and Next, the findings related to ethical dilemmas will help the man­
inaccurate information about the service. Future studies may evaluate agement understand existing technological paradoxes and ethical con­
these issues and must comprehend the perspectives of all stakeholders, cerns in the system. There exists a paucity of medical practitioner-
especially patients, who are the primary consumers or beneficiaries of centered research in this domain, thereby necessitating the consolida­
any such new healthcare technologies. tion of perspectives on diverse AI-driven services to gain a comprehen­
Second, past qualitative studies have used a piecemeal approach and sive understanding of the ethical predicaments that decision-makers
focused on particular aspects of healthcare like the use of AI in radiology could encounter. In order to attain a thorough comprehension of the
(Chen et al., 2021); potential impact of AI in primary care (Balagur­ ethical predicaments linked with AI-driven healthcare services, it is
unathan et al., 2021); ethical issues related to the use of artificial in­ imperative to investigate the interplay between various viewpoints and
telligence in medical research (Mirbabaie et al., 2021); use of AI during perspectives. The ethical considerations surrounding the implementa­
pregnancy (Oprescu et al., 2022); challenges to implementing artificial tion of AI services require decision makers to carefully navigate various
intelligence in healthcare (Norori et al., 2021). Additionally, the current factors, including but not limited to patient privacy, data security,
research mostly focuses on technology-specific viewpoints, such as AI accountability, bias, and transparency. Comprehending these ethical
chat boxes (Nadarzynski et al., 2019), PCAS adoption (Paré and Trudel, predicaments is pivotal in advancing the conscientious and efficient
2007), etc. utilisation of artificial intelligence in the healthcare industry.
Third, our finding also relies on paradoxical thinking to assess the The adoption and deployment of AI models in clinical care requires a
key limitations of AI service usage in healthcare. The findings show that clinical governance framework with diverse representation and experi­
a weak form of AI is still largely used and accepted. The study observed ence. This will improve integration and ethical governance by consid­
that concerns and views of practitioners still outweigh human expertise ering privacy, security, quality, and ethical considerations alongside AI
over AI for critical medical cases. Further, the findings related to data model efficacy and effectiveness (Reddy et al., 2020). Before AI models
quality and privacy issues also indicated medical staff's concerns with AI become a regular component of healthcare delivery, the study suggests a
(Reddy et al., 2020). We hope to stimulate future research into these broader discussion about regulation, stakeholder support, and financial
complex processes. The paradoxical perspective helps researchers un­ assistance (Ajmera and Jain, 2019). Through a comprehensive under­
derstand these concerns, how deep they exist in the system, and how standing of these perceptions, researchers, and policy makers can
different dilemmas or barriers can change the way practitioners think discern potential areas that may benefit from changes, establish guide­
and the results of AI-based work. lines, and effectively tackle ethical considerations pertaining to the
Fourth, the study further highlights different behavioural con­ implementation and utilisation of AI technologies within the healthcare
straints, such as cultural and religious beliefs, insecurities, and techno­ domain.
logical resistance to AI. Research shows that medical staff are perplexed
about using AI services as they lack personal interaction, which is 6. Limitations and future research directions
necessary to address mental problems and critical ailments. The findings
could be used to develop a conceptual framework for these dimensions. The study provides valuable insights into the key drivers, barriers,
Unlike previous descriptive studies, following paradox theory, the study and technological paradoxes associated with the use of AI services in
identifies key ethical concerns or dilemmas experienced by medical healthcare from the viewpoint of healthcare providers. Despite this,
practitioners employing AI technology. Concerns about data quality, there are a few limitations to the study. First, the study used qualitative,
human intervention, reliability of AI-enabled health devices, informed open-ended essay techniques to collect the data from 62 health experts
consent, unethical labour practices, etc. are a few technological para­ from diverse health domains. However, the findings may not represent
doxes raised in the study (Norori et al., 2021). Some have been identi­ the broader health community due to the small sample size. Future
fied, but very few have been investigated in previous research on studies may collect data from a larger group to establish the general­
technology diffusion in organisational settings. isability of the findings. Second, the present study is only focused on the
Finally, the study suggests a few strategies to improve AI adoption viewpoints of healthcare providers. Future studies may also collect the
among healthcare professionals to reduce paradoxes and concerns with perspectives of other relevant stakeholders, particularly patients, to
AI. The study provides insights on funding sources, stakeholders, clinical build trust in AI to further refine the findings. AI adoption may have a
governance, and training requirements. These elements can be used to dual effect on the patient-provider relationship that may improve
build a framework for strategy. Fifth, our study uses data from open- engagement and trust on the one hand and may have potential side ef­
ended essays. Unlike previous studies, it evaluates respondents' free fects due to data privacy issues and information asymmetry, which can
remarks and opinions on the issue. Our findings thus have a broad only be explored from patients' perspectives. To address this, future
organisational context. studies should access patients experiences with the existing system and
various technological, behavioural, and psychological factors

15
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

influencing patients trust in AI. Further, by including patients' views, 2021). There is a growing need for interdisciplinary collaboration be­
healthcare innovators may also verify or explore ethical dilemmas, tween professionals in AI, physicians, and researchers in order to
specifically the need for human touch and expertise and patients' au­ consistently enhance AI solutions for critical care applications. Behav­
tonomy identified in the present study, and how crucial these challenges ioural resistance can be addressed by change management programs,
are for actual consumers of AI-based medical services. user friendly interface, champion programs. The objective should be to
Next, the study was conducted solely in the UK. Future studies may develop artificial AI systems that include user-friendly interfaces,
compare the issues in developed markets with those in developing allowing for smooth integration into established processes, and ensuring
markets with respect to diverse cultural values (liberal or conservative) intuitive use for healthcare professionals.
that may affect the use of AI and data sharing behaviour and with Funding constraints, collaborations with governmental entities,
respect to regulatory and legal systems, as some countries have data commercial enterprises, and charitable institutions is recommended in
protection policies from a sector perspective (USA). On the other hand, a order to get financial support for research and AI initiatives in its
few markets have a standard legal system (Europe) (Curtiss, 2016). application in healthcare domain. Lack of training and knowledge can
Further, some constraints, including financial support, a lack of in­ be addressed by continuous education, ongoing training programs and
vestors, and high investment costs, may not be applicable to developed making available resources to keep medical practitioners informed
markets in comparison to developing systems (Vučinić, 2020). Also, about the latest advancements in artificial intelligence, centres of
some systems are highly developed and some are still at a nascent stage excellence and encouraging inter disciplinary training programs that
of adoption; hence, barriers and dilemmas may differ. Thus, under­ bring together AI specialists, data scientists, and healthcare practi­
standing such diverse perspectives may enhance the relevance of find­ tioners. By using these strategies to overcome these challenges, health­
ings. Further, using different ways to collect data, like observations or care organisations may establish a suitable environment for the effective
individual interviews, would have given a deeper understanding of the integration of AI, hence enhancing patient care, operational efficiency,
social problems that come with putting AI into place. The results ob­ and the overall quality of healthcare services.
tained from qualitative methodologies can be further utilised to
construct comprehensive scales that measure the factors influencing the CRediT authorship contribution statement
adoption AI. These scales can then be subjected to empirical testing
across many domains. Moreover, the utilisation of extensive semi- Nidhi Singh: Conceptualization, Original draft preparation and
structured interviews can prove advantageous in facilitating a novel Writing - Reviewing and Editing.
and unexplored line of investigation, hence shedding light on previously Monika Jain: Literature and Theoritical background, Data curation
unidentified domains that warrant consideration. The utilisation of and Writing - Reviewing and Editing.
observation methods, namely case study and longitudinal analysis, Muhammad Mustafa Kamal: Findings and Implications and
holds significance in examining the consequences of adopting new Reviewing and Editing. Revisions.
technology within a hospital or medical center setting over a period of Rahul Bodhi: Tables and Data curation and Writing - Reviewing and
time. This approach allows researchers to immerse themselves within Editing.
the context under investigation (Choudrie and Dwivedi, 2005). These Bhumika Gupta: Findings and Implications and Reviewing and
methodologies facilitate the prediction of AI based outcomes, the ex­ Editing. Revisions.
amination of current impediments, and many facets of technology
dissemination.
Declaration of competing interest
6.1. Recommendations
None.
The findings of the study identified five key barriers in the adoption
of AI in healthcare. Privacy and trust issues were identified as significant Data availability
barriers. The implementation of robust data governance in the health­
care sector, specifically focusing on the secure handling of patient data Data will be made available on request.
and stringent data governance policies and procedures to safeguard
sensitive information is one of the ways for addressing privacy and trust Acknowledgements
issues. By examining existing literature and industry practices, this study
highlights the importance of establishing effective data governance None.
frameworks to mitigate risks and protect patient privacy (Panch et al.,
2019; Rajpurkar et al., 2022). The implementation of encryption, access Annexure I
controls, and audit trails is of paramount importance in ensuring the
security and integrity of sensitive data. Bartoletti (2019) suggested four 1) What is your opinion on the use of Artificial Intelligence (AI) enabled
step mechanism to address the privacy issue, which includes data pri­ technology/applications in healthcare? Please elaborate your answer
vacy impact assessment, algorithmic impact assessments, audit trails by using suitable examples. Do you have any experience? If yes then
and governance framework. The implementation of transparency and please elaborate?
consent mechanisms for data usage, as well as the conduct of third-party 2) According to you, what are the main reasons behind the adoption of
audits, can prove to be beneficial in effectively addressing this issue AI-enabled technology/applications in healthcare? Please discuss
(Reddy et al., 2020). using suitable examples.
Unsuitable for critical illness was another of the limitations of the use 3) What are the different benefits offered by AI-enabled technology/
of AI in health care. Allocation of the resources towards the development applications in healthcare? Please elaborate your answers by using
of specialised AI models that are especially designed to address critical suitable examples?
ailments can help address this issue. It is imperative to subject these 4) According to you, what are the key challenges or problems faced by
models to rigorous testing and validation procedures in order to ascer­ the healthcare sector when it comes to the adoption of AI-enabled
tain their accuracy and dependability in critical care situations. Also, it technology/applications? Please elaborate your answer by using
is equally important to emphasize the support role of AI in critical care. suitable examples.
AI serves as a valuable support tool for healthcare professionals, aiding 5) According to you, what are the different barriers that are faced by the
them in expeditiously making well-informed judgements (Norori et al., healthcare sector when it comes to the adoption of AI-enabled

16
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

technology/applications? Please elaborate your answer by using Jussupow, E., Spohrer, K., Heinzl, A., Gawlitza, J., 2021. Augmenting medical diagnosis
decisions? An investigation into physicians’ decision-making process with artificial
suitable examples.
intelligence. Inf. Syst. Res. 32 (3), 713–735.
6) According to you, what are the key dilemmas faced by healthcare Kazanjian, A., Green, C.J., 2002. Beyond effectiveness: the evaluation of information
practitioners when it comes to the adoption and use of AI-enabled systems using a comprehensive health technology assessment framework. Comput.
technology/applications in healthcare? Please discuss with suitable Biol. Med. 32 (3), 165–177.
Kalisz, D.E., Khelladi, I., Castellano, S., Sorio, R., 2021. The adoption, diffusion &
examples. categorical ambiguity trifecta of social robots in e-health–insights from healthcare
7) According to you, what are the different ethical issues that are professionals. Futures 129, 102743.
associated with the adoption and use of AI-enabled technology/ap­ Khanijahani, A., Iezadi, S., Dudley, S., Goettler, M., Kroetsch, P., Wise, J., 2022.
Organizational, professional, and patient characteristics associated with artificial
plications in healthcare? intelligence adoption in healthcare: a systematic review. Health Policy Technol. 11
8) How do you see the future of AI-enabled technology/applications in (1), 100602.
healthcare? Please elaborate on your choices with suitable examples. Koo, B., Curtis, C., Ryan, B., 2021. Examining the impact of artificial intelligence on hotel
employees through job insecurity perspectives. Int. J. Hosp. Manag. 95, 102763.
9) Are there any other important issues related to AI-enabled technol­ Kooli, C., Al Muftah, H., 2022. Artificial intelligence in healthcare: a comprehensive
ogy/application adoption in healthcare that you would like to review of its ethical concerns. Technol. Sustain. 1 (2), 121–131.
mention? Please elaborate. Kumar, S., Raut, R.D., Narkhede, B.E., 2020. A proposed collaborative framework by
using artificial intelligence-internet of things (AI-IoT) in COVID-19 pandemic
situation for healthcare workers. Int. J. Healthc. Manag. 13 (4), 337–345.
Kumar, P., Dwivedi, Y.K., Anand, A., 2021. Responsible artificial intelligence (AI) for
References value formation and market performance in healthcare: the mediating role of
Patient’s cognitive engagement. Inf. Syst. Front. 1–24.
Kumar, A., Krishnamoorthy, B., Bhattacharyya, S.S., 2023. Machine learning and
Ajmera, P., Jain, V., 2019. Modelling the barriers of Health 4.0–the fourth healthcare
artificial intelligence-induced technostress in organizations: a study on automation
industrial revolution in India by TISM. Oper. Manag. Res. 12 (3), 129–145.
augmentation paradox with socio-technical systems as coping mechanisms. Int. J.
Al Badi, F.K., Alhosani, K.A., Jabeen, F., Stachowicz-Stanusch, A., Shehzad, N.,
Organ. Anal. https://doi.org/10.1108/ijoa-01-2023-3581.
Amann, W., 2021. Challenges of AI Adoption in the UAE Healthcare. Vision,
Laï, M.C., Brian, M., Mamzer, M.F., 2020. Perceptions of artificial intelligence in
0972262920988398.
healthcare: findings from a qualitative survey study among actors in France.
Alhashmi, S.F., Salloum, S.A., Mhamdi, C., 2019. Implementing artificial intelligence in
J. Transl. Med. 18 (1), 1–13.
the United Arab Emirates healthcare sector: an extended technology acceptance
Lee, S.M., Lee, D., 2021. Opportunities and challenges for contactless healthcare services
model. Int. J. Inf. Technol. Language Stud. 3 (3), 27–42.
in the post-COVID-19 Era. Technol. Forecast. Soc. Chang. 167, 120712.
Angelucci, A., Li, Z., Stoimenova, N., Canali, S., 2022. The paradox of the artificial
Lee, C., Lim, C., 2021. From technological development to social advance: a review of
intelligence system development process: the use case of corporate wellness
Industry 4.0 through machine learning. Technol. Forecast. Soc. Chang. 167, 120653.
programs using smart wearables. AI & Soc. https://doi.org/10.1007/s00146-022-
Leoni, L., Cristofaro, M., 2021. To adopt or not to adopt? A co-evolutionary framework
01562-4.
and paradox of technology adoption by small museums. Curr. Issue Tour. 1–22.
Balagurunathan, Y., Mitchell, R., El Naqa, I., 2021. Requirements and reliability of AI in
Liu, K., Tao, D., 2022. The roles of trust, personalization, loss of privacy, and
the medical context. Phys. Med. 83, 72–78.
anthropomorphism in public acceptance of smart healthcare services. Comput. Hum.
Bamel, U., Talwar, S., Pereira, V., Corazza, L., Dhir, A., 2023. Disruptive digital
Behav. 127, 107026 https://doi.org/10.1016/j.chb.2021.107026.
innovations in healthcare: knowing the past and anticipating the future.
Ma, B., Yang, J., Wong, F.K.Y., Wong, A.K.C., Ma, T., Meng, J., Zhao, Y., Wang, Y., Lu, Q.,
Technovation 125, 102785.
2023. Artificial intelligence in elderly healthcare: a scoping review. Ageing Res. Rev.
Bartoletti, I., 2019. AI in healthcare: ethical and privacy challenges. In: Artificial
83, 101808.
Intelligence in Medicine:17th Conference on Artificial Intelligence in Medicine,
Malodia, S., Chauhan, C., Jabeen, F., Dhir, A., 2023. Antecedents and consequences of
AIME 2019, Poznan, Poland, June 26–29, 2019, Proceedings 17. Springer
open innovation: a conceptual framework. Int. J. Entrep. Behav. Res. Vol. ahead-of-
International Publishing, pp. 7–10.
print No. ahead-of-print.
Bughin, J., Herring, L., Mayhew, H., Seong, J., Allas, T., 2019. Artificial Intelligence in
Martinez-Millana, A., Saez-Saez, A., Tornero-Costa, R., Azzopardi-Muscat, N., Traver, V.,
the United Kingdom: Prospects and Challenges. McKinsey Global Institute.
Novillo-Ortiz, D., 2022. Artificial intelligence and its impact on the domains of
Char, D.S., Abràmoff, M.D., Feudtner, C., 2020. Identifying ethical considerations for
universal health coverage, health emergencies and health promotion: an overview of
machine learning healthcare applications. Am. J. Bioeth. 20 (11), 7–17.
systematic reviews. Int. J. Med. Inform. 166, 104855.
Chaudhary, S., Dhir, A., Gligor, D., Khan, S.J., Ferraris, A., 2022. Paradoxes and coping
Mckinsey Global Institute, 2019. Artificial Intelligence in the United Kingdom:
mechanisms in the servitisation journey. Ind. Mark. Manag. 106, 323–337.
PROSPECTS and Challenges. Retrieved from McKinsey Global Institute analysis:
Chen, Y., Stavropoulou, C., Narasinkan, R., Baker, A., Scarbrough, H., 2021.
chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.mckinsey.
Professionals’ responses to the introduction of AI innovations in radiology and their
com/~/media/McKinsey/Featured%20Insights/Artificial%20Intelligence/Artificial
implications for future adoption: a qualitative study. BMC Health Serv. Res. 21 (1).
%20intelligence%20in%20the%20United%20Kingdom%20Prospects%20and%
Choudrie, J., Dwivedi, Y.K., 2005. Investigating the research approaches for examining
20challenges/Artificial-intelligence-in-t (Febuary 12).
technology adoption issues. J. Res. Pract. 1 (1), D1.
Merhi, M.I., 2023. An assessment of the barriers impacting responsible artificial
Christofi, M., Hadjielias, E., Mahto, R.V., Tarba, S., Dhir, A., 2023. Owner-manager
intelligence. Inf. Syst. Front. 25 (3), 1147–1160.
emotions and strategic responses of small family businesses to the COVID-19
Mill, E., Garn, W., Ryman-Tubb, N., 2022. Managing sustainability tensions in artificial
pandemic. J. Small Bus. Manag. 1–42.
intelligence. In: Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and
Cubric, M., 2020. Drivers, barriers and social considerations for AI adoption in business
Society. https://doi.org/10.1145/3514094.3534175.
and management: a tertiary study. Technol. Soc. 62, 101257.
Mirbabaie, M., Hofeditz, L., Frick, N.R., Stieglitz, S., 2021. Artificial intelligence in
Curtiss, T., 2016. Privacy harmonization and the developing world: the impact of the
hospitals: providing a status quo of ethical considerations in academia to guide
EU’s general data protection regulation on developing economies. Wash. JL Tech. &
future research. AI & Soc. 1–22.
Arts 12, 95.
Nadarzynski, T., Miles, O., Cowie, A., Ridge, D., 2019. Acceptability of artificial
Dhir, A., Chen, G.M., Chen, S., 2017. Why do we tag photographs on Facebook?
intelligence (AI)-led chatbot services in healthcare: a mixed-methods study. Digit.
Proposing a new gratifications scale. New Media Soc. 19 (4), 502–521.
Health 5, 2055207619871808.
Di Vaio, A., Palladino, R., Hassan, R., Escobar, O., 2020. Artificial intelligence and
Norori, N., Hu, Q., Aellen, F.M., Faraci, F.D., Tzovara, A., 2021. Addressing bias in big
business models in the sustainable development goals perspective: a systematic
data and AI for health care: a call for open science. Patterns 2 (10), 100347. https://
literature review. J. Bus. Res. 121, 283–314.
doi.org/10.1016/j.patter.2021.100347.
Dicuonzo, G., Donofrio, F., Fusco, A., Shini, M., 2023. Healthcare system: moving
Oprescu, A., Miró-Amarante, G., García-Díaz, L., Rey, V., Chimenea-Toscano, A.,
forward with artificial intelligence. Technovation 120, 102510.
Martínez-Martínez, R., Romero-Ternero, M., 2022. Towards a data collection
Du, S., Xie, C., 2021. Paradoxes of artificial intelligence in consumer markets: ethical
methodology for responsible artificial intelligence in health: a prospective and
challenges and opportunities. J. Bus. Res. 129, 961–974. https://doi.org/10.1016/j.
qualitative study in pregnancy. Inf. Fusion 83–84, 53–78.
jbusres.2020.08.024.
Panch, T., Mattie, H., Celi, L.A., 2019. The “inconvenient truth” about AI in healthcare.
Garbuio, M., Lin, N., 2019. Artificial intelligence as a growth engine for health care
NPJ Digit. Med. 2 (1) https://doi.org/10.1038/s41746-019-0155-4.
startups: emerging business models. Calif. Manag. Rev. 61 (2), 59–83.
Paré, G., Trudel, M.C., 2007. Knowledge barriers to PACS adoption and implementation
Gioia, D.A., Corley, K., Hamilton, A.L., 2013. Seeking qualitative rigor in inductive
in hospitals. Int. J. Med. Inform. 76 (1), 22–33.
research: notes on the Gioia methodology. Organ. Res. Methods 16 (1), 15–31.
Raghavan, A., Demircioglu, M.A., Taeihagh, A., 2021. Public health innovation through
Gruetzemacher, R., Dorner, F.E., Bernaola-Alvarez, N., Giattino, C., Manheim, D., 2021.
cloud adoption: a comparative analysis of drivers and barriers in Japan, South Korea,
Forecasting AI progress: a research agenda. Technol. Forecast. Soc. Chang. 170,
and Singapore. Int. J. Environ. Res. Public Health 18 (1), 334.
120909.
Raisch, S., Krakowski, S., 2021. Artificial intelligence and management: the
Guan, J., 2019. Artificial intelligence in healthcare and medicine: promises, ethical
automation–augmentation paradox. Acad. Manag. Rev. 46 (1), 192–210. https://doi.
challenges and governance. Chin. Med. Sci. J. 34 (2), 76–83.
org/10.5465/amr.2018.0072.
He, W., Zhang, Z.J., Li, W., 2021. Information technology solutions, challenges, and
Rajpurkar, P., Chen, E., Banerjee, O., Topol, E.J., 2022. AI in health and medicine. Nat.
suggestions for tackling the COVID-19 pandemic. Int. J. Inf. Manag. 57, 102287.
Med. 28 (1), 31–38. https://doi.org/10.1038/s41591-021-01614-0.
Hercheui, M., Mech, G., 2021. Factors affecting the adoption of artificial intelligence in
healthcare. Glob. J. Bus. Res. 15 (1), 77–88.

17
N. Singh et al. Technological Forecasting & Social Change 198 (2024) 122967

Reddy, S., Allan, S., Coghlan, S., Cooper, P., 2020. A governance model for the (Emerald), NMIMS management review (Web of Science), IJICBM (Inderscience), IJSSM
application of AI in health care. J. Am. Med. Inform. Assoc. 27 (3), 491–497. https:// (Inderscience), Decision (Springer), FIIM (Sage), MLS (Sage), SERD, GSCSR etc.
doi.org/10.1093/jamia/ocz192.
ResearchAndMarkets.com, 2020. $45.2 billion worldwide artificial intelligence in
Monika Jain is PhD from the University of Rajasthan. She is UGC-NET qualified and holds
healthcare industry to 2026 - featuring Lunit, Magnea & Maxq AI Among Others -
Master's Degree in Economics from the University of Rajasthan, Graduation from Sophia
ResearchAndMarkets.com. Retrieved from. https://www.businesswire.com/:. https:
College, Ajmer. She holds a bright scholastic record, has won National scholarship for her
//www.businesswire.com/news/home/20200622005457/en/45.2-Billion-Worldwi
academic achievement from the Ministry of Education, Government of India. Dr. Monika
de-Artificial-Intelligence-in-Healthcare-Industry-to-2026—Featuring-Lunit-Magnea-
Jain has been engaged in teaching for the last Seventeen years at postgraduate level at
Maxq-AI-Among-Others—ResearchAndMarkets.com (June 20).
various Management Institutes in Bangalore, Delhi, Jaipur, Chennai, Ahmedabad and
Sharma, S., Singh, G., Islam, N., Dhir, A., 2022. Why do smes adopt artificial intelligence-
Greater Noida. Prior to BIMTECH, she has worked at Army Institute of Management &
based chatbots? IEEE Trans. Eng. Manag. https://doi.org/10.1109/
technology, Greater Noida. She has also worked as an Associate Professor with IILM
TEM.2022.3203469 (early access).
College of Management Studies Greater Noida, Asia-Pacific Institute of Management New
Shi, F., Wang, J., Shi, J., Wu, Z., Wang, Q., Tang, Z., Shen, D., 2020. Review of artificial
Delhi, Indus Institute of Engineering and Technology Ahmedabad etc. She has publications
intelligence techniques in imaging data acquisition, segmentation, and diagnosis for
in renowned journals listed in ABDC, Scopus etc.
COVID-19. IEEE Rev. Biomed. Eng. 14, 4–15.
Siala, H., Wang, Y., 2022. SHIFTing artificial intelligence to be responsible in healthcare:
a systematic review. Soc. Sci. Med. 296, 114782. Muhammad Mustafa Kamal is an Associate Professor of Supply Chain Management,
Sipior, J.C., 2020. Considerations for development and use of AI in response to COVID- Curriculum Lead in Decision Making (Business Analytics) and Risk Management, and
19. Int. J. Inf. Manag. 55, 102170. Director of the Transnational Research Degree (TNRD) Programme at the School of
Swedberg, R., 2020. Exploratory research. In: The Production of Knowledge, pp. 17–41. Strategy and Leadership. Dr. Kamal is a Senior Fellow of HEA and a Certified Management
Talwar, S., Dhir, A., Islam, N., Kaur, P., Almusharraf, A., 2023. Resistance of multiple and Business Educator (CMBE). Currently, he is the Deputy Editor for the Journal of En­
stakeholders to e-health innovations: integration of fundamental insights and terprise Information Management (JEIM), Senior Editor for Information Systems Man­
guiding research paths. J. Bus. Res. 166 (2023), 114135. agement and on editorial board for Government Information Quarterly. He has published
Tandon, A., Dhir, A., Islam, A.N., Mäntymäki, M., 2020. Blockchain in healthcare: a over 70 papers in refereed academic journals, conference proceedings, book chapters, and
systematic literature review, synthesizing framework and future research agenda. magazine article. His research work has appeared in several leading ABS ranked journals,
Comput. Ind. 122, 103290. such as International Journal of Production Research, Journal of Business Research, In­
Tandon, A., Dhir, A., Islam, N., 2023. Mobile health interventions for cancer care and formation and Management, and many more.
support: the next level of digitalization in healthcare? IEEE Trans. Eng. Manag.
https://doi.org/10.1109/TEM.2023.3243724 (early access).
Rahul Bodhi is an Assistant Professor of Human Resource Management and Organiza­
Tseng, R.M.W.W., Gunasekeran, D.V., Tan, S.S.H., Rim, T.H., Lum, E., Tan, G.S.,
tional Behaviour at the School of Business, UPES, Dehradun, India. He obtained his
Tham, Y.C., 2021. Considerations for artificial intelligence real-world
Doctorate in Management from the School of Management Studies, Motilal Nehru National
implementation in ophthalmology: providers’ and patients’ perspectives. Asia-Pac. J.
Institute of Technology Allahabad, India. His research interests lie in the area of the Future
Ophthalmol. 10 (3), 299–306.
of work, Knowledge sharing, Exponential technologies, Social media, and Work-related
Vučinić, M., 2020. Fintech and financial stability potential influence of FinTech on
outcomes. He was a Visiting Research Fellow at the Graduate Institute of Digital
financial stability, risks and benefits. J. Cent. Bank. Theory Pract. 9 (2), 43–66.
Learning and Education, National Taiwan University of Science & Technology (NTUST),
Wamba, S.F., Bawack, R.E., Guthrie, C., Queiroz, M.M., Carillo, K.D.A., 2021. Are we
Taipei, Taiwan. His research appears in various leading journals, including the Interna­
preparing for a good AI society? A bibliometric review and research agenda.
tional Journal of Emerging Markets, International Journal of Educational Management
Technol. Forecast. Soc. Chang. 164, 120482.
and IEEE Transactions on Engineering Management among others. He has taught several
Wiljer, D., Hakim, Z., 2019. Developing an artificial intelligence–enabled health care
bachelor's, master's, and doctoral-level courses.
practice: rewiring health care professions for better care. J. Med. Imaging Radiat.
Sci. 50 (4), S8–S14.
Wong, S.H., Al-Hasani, H., Alam, Z., Alam, A., 2019. Artificial intelligence in radiology: Bhumika Gupta is an Associate Professor in the area of Human Resource Management and
how will we be affected? Eur. Radiol. 29 (1), 141–143. Program Director of the IMT-BS Master of Science in International Management Her PhD
in Human Resource Management is from the University of Pau, France and she has been a
visiting researcher at Copenhagen Business School, Denmark and at Stanford University,
Nidhi Singh has an experience of more than 17 years in teaching and corporate. At present
USA. In her research, she compares motivation levels in management practice between
she is working with Jaipuria Institute of Management, Noida as Associate Professor in
traditional and virtual project teams and studies corporate culture and psychological
Finance Area. She has done her Ph.D. from GGSIP University, Delhi. She has qualified UGC
contracts in the work place. Her research on Human Resource Management, Organization
Net also. She has presented many papers in various Seminars & Conferences including
Behaviour, Strategic Human Resource Management, and Corporate Social Responsibility
IIMR, IICA, NLSIU etc. and published papers in journals of National & International Repute
has been widely published in international journals and has been funded by French and
like IJIM (Elesvier), JRCS (Elsevier), IJBM (Emerald), WHMM (Taylor and Francis), IJCHM
international government agencies.

18