Technology in Society 75 (2023) 102370

Contents lists available at ScienceDirect

Technology in Society
journal homepage: www.elsevier.com/locate/techsoc

Use of ChatGPT in academia: Academic integrity hangs in the balance

Saeed Awadh Bin-Nashwan a, *, Mouad Sadallah b, Mohamed Bouteraa c
a
Department of Accounting, Finance and Economics, Faculty of Business, Curtin University, Miri, Malaysia
b
Tunku Puteri Intan Safinaz School of Accountancy, Universiti Utara Malaysia, Sintok, Malaysia
c
Faculty of Business, Economics and Accountancy, Universiti Malaysia Sabah, Kota Kinabalu, Malaysia

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

Keywords: In today’s academic world, some academicians, researchers and students have begun employing Artificial In­
Artificial intelligence telligence (AI) language models, e.g., ChatGPT, in completing a variety of academic tasks, including generating
ChatGPT ideas, summarising literature, and essay writing. However, the use of ChatGPT in academic settings is a
Academia
controversial issue, leading to a severe concern about academic integrity and AI-assisted cheating, while
Academic integrity
Plagiarism
scholarly communities still lack clear principles on using such innovation in academia. Accordingly, this study
Technology adoption aims to understand the motivations driving academics and researchers to use ChatGPT in their work, and spe­
cifically the role of academic integrity in making up adoption behavior. Based on 702 responses retrieved from
users of ResearchGate and Academia.edu, we found that ChatGPT usage is positively shaped by time-saving
feature, e-word of mouth, academic self-efficacy, academic self-esteem, and perceived stress. In contrast, peer
influence and academic integrity had a negative effect on usage. Intriguingly, academic integrity-moderated
interactions of time-saving, self-esteem and perceived stress on ChatGPT usage are found to be significantly
positive. Therefore, we suggest that stakeholders, including academic institutions, publishers and AI language
models’ programmers, should work together to specify necessary guidelines for the ethical use of AI chatbots in
academic work and research.

1. Introduction academic sphere [8,9]. One of the potential opportunities for such a
revolutionary platform is to assist scientists and researchers in gener­
As a variant of GPT-3 (Generative Pre-trained Transformer 3), ating ideas and overcoming writer’s block [3] and as a system to auto­
ChatGPT is an extensive language model launched by Open Artificial mate time-consuming and repetitive content production tasks [10].
Intelligence (OpenAI) in November 2022, has become one of the most However, there are critical challenges with using ChatGPT in academia:
crucial and unprecedented real-world AI platforms to date [1,2]. It the possibility of plagiarism and academic dishonesty [11]; [8]. AI
specifically uses conversational chatbots to rapidly generate human-like essay-writing platforms assist researchers and students in generating
texts with a unique capability to conduct a wide variety of language essays, articles, research, and assignments on their behalf, leading to a
prompts, including question-answering, translation, summarisation, and severe concern about academic integrity and AI-assisted cheating [3,8,
text generation [3]. Surprisingly, the AI chatbot ChatGPT is likely the 12]. In response, some higher education institutions worldwide have
fastest-growing user application in internet history ever [4], with nearly banned access to ChatGPT or other AI tools on campuses [13] – while
100 million users as of January 2023, and has roughly 1.8 billion some are still reluctant to ban it [14] – over fears handing in unauthentic
website visitors per month currently [5,6]. Since its emergence, or potentially plagiarised content. On the other hand, technology ex­
ChatGPT has been utilized for a wide range of purposes, including perts are urging universities to train faculty, researchers, and students
content generation, as it has been shown to be able to generate articles on how to use ChatGPT and AI platforms appropriately rather than ban
[7,8], stories [3] and other forms of realistic and coherent written them outright [15].
content. Since its release in late 2022, interest in ChatGPT has been incredibly
Despite its unprecedented success, ChatGPT is turning out to be a high, especially among academicians [16]. The available data in Google
double-edged sword that has been making waves throughout the Trends (GT) using “ChatGPT” as a search term for the “Job & Education”

* Corresponding author. Department of Accounting, Finance and Economics, Faculty of Business, Curtin University, Miri, 98009, Malaysia.
E-mail addresses: [email protected], [email protected] (S.A. Bin-Nashwan), [email protected] (M. Sadallah), Bouteraa_med@ums.
edu.my (M. Bouteraa).

https://doi.org/10.1016/j.techsoc.2023.102370
Received 20 June 2023; Received in revised form 5 September 2023; Accepted 9 September 2023
Available online 11 September 2023
0160-791X/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

Fig. 1. GT on global search interest for ChatGPT, precisely “Job and Education” category for the period from late November 2022 (its release) until early June 2023
Note: Topics and queries are marked “breakout” by GT, showing a massive increase in interest and popularity.

category since its release contains interesting outcomes, as displayed in body of knowledge, the outcomes of this work offer profound implica­
Fig. 1. Only a short time after public release, the popularity of ChatGPT tions for stakeholders in the academic community.
soared, with search interest peaking on April 17, 2023. The term has
been used worldwide, as GT aggregated search data from 71 different 2. Literature review
countries, indicating the unprecedented global demand and popularity
for this AI tool. Most revealingly, the data shows that the term 2.1. Theoretical basis
“plagiarism” was ranked as the top related search topic, followed by
“AI”; meanwhile, “plagiarism” was also ranked among the top five SCT theory, recognized as one of the most potent theories of human
related queries. Taken together, these outcomes shed light on the behavior [21], underpins the present research. The central premise of
importance and timeliness of this research. There is a pressing need to SCT is that individuals’ behavioral intentions are influenced not solely
provide a thorough understanding of factors driving researchers to use by their behaviors but also by personal, cognitive, and environmental
ChatGPT and the role of academic integrity in making up their moti­ aspects. According to Cooper and Lu (2016), the fundamental tenet of
vations to adopt such an innovative tool. SCT posits that people’s behavior is shaped by both their cognitive
Although there is a growing literature has begun to explore the po­ processes and external social environment situations [21]. advances the
tential role, implications, opportunities and threats of ChatGPT to concept of triadic reciprocal determinism, which involves personal at­
academia and the higher education sphere [3,8,12,16–20], it is likely tributes like internal cognitive and affective states, as well as physical
that no study to date has attempted to understand the motivations that attributes, including external environmental components, all of which
drive academics and researchers to use ChatGPT in their work, and interact to shape overt behavior. Individuals’ behaviors are driven by
specifically the role of academic integrity in making up adoption their perceptions, expectations, and beliefs. In other words, how people
behavior. Academic integrity has long been a concern in academia and think and feel is related to their behavioral intention [21,22]. Further­
higher education, and with the modern and pervasive technological more, the theory proposes that a person’s abilities, skills, and knowledge
revolution of AI chatbots, it has become more critical than ever [16]. play a role in motivating them to undertake specific actions.
Accordingly, the present study aims to bridge this gap by empirically [21] further emphasizes that human behaviors are predicted by their
examining the factors shaping ChatGPT usage among researchers environment, encompassing external determinants. Such an environ­
worldwide and including academic integrity in the research model [21]. ment comprises both social and physical components. The physical
Social Cognitive Theory (SCT) provides a more holistic understanding of environment encompasses natural and human-made elements within a
various personal, behavioral and environmental aspects that can stim­ person’s surroundings. On the other hand, the social environment in­
ulate the adoption behavior of new technologies [22], including cludes immediate physical surroundings, cultural contexts, and social
ChatGPT usage in academic settings. Specifically, the attempt in this relationships in which specific groups of individuals operate and
study seeks to answer the following questions: (1) How do time-saving interact, as [23] suggested. Indeed, it encompasses factors, such as social
feature, electronic word of mouth (e-WOM), peer influence, academic norms, community access, peer influence, values, and more [21].
self-esteem, self-efficacy, and stress influence researchers’ ChatGPT Notably, the social environment is now recognized to include both real
usage? (2) How does academic integrity moderate the relationships on and virtual worlds [22].
ChatGPT’s adoption model? Besides the research implications to the The third facet of the SCT is behavior, which refers to how an

2
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

Fig. 2. Research model.

individual acts or reacts in specific situations or towards specific objects The chatbot can save time by scheduling appointments, setting re­
(Bandura, 1991). This concept also extends to how a person engages minders, preparing lessons, and consultations, and recommending aca­
with technology and technological innovations (Ratten & Ratten, 2007; demic resources [36]. When academics perceive they can efficiently
[22]. These three factors or components interplay with each other to save time by using ChatGPT, they are more likely to use it to perform
forecast individuals’ actions. Nevertheless, their predictive abilities are their job. Based on the relevant literature and the behavioral time theory
not uniform, and their effects on each other do not necessarily occur [30], the following hypothesis was formed:
simultaneously [21]. The SCT framework has found application across
H1. Time-saving feature increases the use of ChatGPT in academia.
various disciplines, likely due to its adaptable nature, as it is able to
recognize human behaviors as dynamic and ever-changing (Kock,
2004). For instance, this has been extensively utilized in studies related 2.3. e-WOM
to e-government system adoption [24], mobile learning [25], internet
banking [22], telemedicine [26], e-tax systems [27]. However, the Word of Mouth (WOM) is an experimental method for disseminating
application of SCT in examining the adoption of AI platforms, including information. WOM is communications between clients regarding a
ChatGPT, within the academic context has been rarely attempted. As specific service/product or its provider without commercial control [38,
such, in the present study, we predict researchers’ adoption of ChatGPT 39]. The advancement of digital platforms led to a novel form of
in their work and how their behavior can be shaped by a variety of communication, “e-WOM”, the electronic version of traditional WOM.
behavioral, personal and environmental factors, such as eWOM, peer Essentially, e-WOM is defined as optimistic or adverse testimony made
influence, time-saving features, self-esteem, academic self-efficacy, by existing or prospective customers about services, products, or their
perceived stress and academic integrity (Fig. 2). Based on these essen­ providers, made available to other institutions and individuals via cy­
tial aspects, SCT can provide a more holistic understanding of re­ berspace [40]. e-WOM has several unique characteristics, including
searchers’ ChatGPT adoption behavior in academic settings [21,22,28, greater scalability, diffusion speed, persistency, accessibility, measur­
29]. ability and quantifiability [41].
This phenomenon is relevant to marketers as e-WOM is weighed as a
2.2. Time-saving feature crucial factor shaping customer behavior towards technology uptake
(Daugherty & Hoffman, 2014; [42]. A recent cross-cultural systematic
The value of time and its impact on individual behavior has been literature review by Ref. [43] highlighted that e-WOM might influence a
widely praised in behavioral economics [30,31]. In a time-sensitive broad range of behavioral outcomes, including trust, attitude, risk
modern society, time is a paramount intangible resource, the use of perception, e-satisfaction, perceived usefulness and intention to use.
which can be exchanged for another resource, such as wealth or effort Indeed, a prior study confirmed that the information shared among
[32]. The literature consistently concluded that the timeliness of ser­ networking sites through e-WOM reflects the knowledge and experience
vices has always been important in users’ motivation to use technologies of other users [44]. However, the literature has not refined the type of
[33–35]. products; in terms of diagnosticity, the provision of tactile information
The time-saving feature of ChatGPT for data administering and within a mess community significantly impacts user evaluations. Such
assimilation into people’s daily lives has made the timeliness element effects may vary depending on individuals’ perceptions, needs, aca­
increasingly essential to accomplishing tasks, boosting productivity, and demic purposes, and the nature of systems like ChatGPT. However, this
achieving goals [36]. In general, efficiency can be regarded as an particular setting has rarely been examined. Thus, we attempt to fill this
advantage of using ChatGPT for users’ experience. However, its effect on gap by exploring the potential linkage between e-WOM and ChatGPT
consumer behavior has been scarcely studied, particularly in the aca­ usage in academia. Being regarded as the concrete evidence provided by
demic context. ChatGPT can be a fundamental means in academia [37]. web communities, a more reliable source to quantify the usefulness and

3
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

benefits of ChatGPT, e-WOM communication among academics likely more sensitive to positive stimuli and tend to use ChatGPT. Thus, the
has a more significant impact to be integrated into their practices. Under following hypothesis can be formulated:
such conditions, the following hypothesis is postulated:
H4. Self-esteem increases the use of ChatGPT in academia.
H2. e-WOM increases the use of ChatGPT in academia.
2.6. Academic self-efficacy
2.4. Peer influence
Self-efficacy is another concept from Bandur’s (1989) SCT—it de­
The social psychology perspective emphasizes the importance of scribes an individual confidence in his/her ability to do or learn speci­
peer influence (also referred to as social influence) in determining fied tasks. In academia, it often refers to academic self-efficacy, defined
behavior. The theory of conflict elaboration related to social influences as a person’s confidence in achieving academic success and educational
is that an individual decision towards new product innovation is goals [65,66]. It relates to motivation, accomplishment, emotion,
significantly determined by his/her peers and social groups [45–47]. cognition, and self-regulation [65].
[48] social influence theory, the elementary form of influence is In the aftermath of ChatGPT’s launch, educators have been
“compliance”, which occurs when individuals admit influence to make immersed and alarmed; meanwhile, there are proponents and opponents
promising reactions from other people or groups; “internalisation”, also of its use in academia. However, it is beneficial for academics to derive
known as social proof, arises when a person takes others’ views as evi­ insights and make credentialed assessments about ChatGPT’s signifi­
dence of reality; “identification” happens when a person accepts beliefs cance in education (Jürgen-Rudolph et al., 2023) [67]. suggested that
to create or sustain relationships within society or groups of people. AI-powered platforms have tremendous potential for transformative
Therefore, such influences could be described as changes in attitude changes in academic settings. Employing artificial intelligence in higher
produced by extrinsic factors. education could boost academic self-efficacy by giving academics access
According to several well-established technology acceptance models to a complex and powerful tool that enhances their capabilities [68].
like the TBP, TRA, UTAUT and TAM – peers’ influence is a substance- Using ChatGPT in academia may similarly reduce teaching and learning
determining factor of behavioral intention, which refers to the degree workloads, gain insights into students’ learning progress, and facilitate
to which a person’s acceptance of technology is pretentious by others’ classroom innovation by streamlining grading, monitoring plagiarism,
opinions [49]. Many technology-related studies asserted that users’ supervision, and feedback [69]. However, scarce empirical studies have
behavior is significantly affected by individual inducements, preferences investigated the effect of academic self-efficacy on using ChatGPT,
and expectations of others [50–52]. Nevertheless, this later conflicted which demands examination by this study. Based on the earlier argu­
with some studies that failed to report significant evidence [53–55]. The ments asserting that using ChatGPT will assist academics in mastering
inconsistent inferences suggest that peer influence is context-dependent, their skills and overcoming difficulties in work, it can be formally hy­
which motivates the contextual examination included in this study. pothesis that:
However, less consideration has been given to investigating the influ­
ence of peers in the context of ChatGPT usage. It could be inferred that H5. Academic self-efficacy increases the use of ChatGPT in academia.
ChatGPT is modern and trendy, posing weight to the idea that the peers’
influence captures academics’ use of ChatGPT. 2.7. Perceived stress

H3. Peer influence increases the use of ChatGPT in academia. Perceived stress refers to individuals’ feelings or perceptions of
tension or pressure he/she experiences over some time [70,71].
2.5. Academic self-esteem Perceived stress comes when a person must cope uncontrollably with
persistent inconveniences, issues, or obstacles regarding crucial aspects
Core self-evaluation (CSE) by Ref. [56] and SCT (Bandur, 1989) are of life or work. Previous studies have shown that stress has an adverse
well-known personality trait theories reflecting individuals’ funda­ impact on mental health in the e-learning environment [72,73]. One of
mental beliefs about themselves. Self-esteem, also known as “self-­ the significant stresses contributing to the rise in stress levels is the
concept or self-confidence”, is the primary psychological dimension of sudden and forced move to online education during COVID-19 [74].
these theories, creating optimistic self-belief [57]. This creates the Similarly, experiments reported a constructive relationship between
necessary skills or possibilities for completing a task successfully and optimistic perceptions of technology and mental well-being [75]; Zhuo
delivering a positive outcome. Self-esteem is a comprehensive construct et al., 2023), implying that the insight of ChatGPT as an unbiased tool is
focused on personality and feelings of self-worth [58]. The self-concept associated with perceived stress and positive user experience [76].
regarding one’s character has been linked to various phenomena, such conducted a systematic review to evaluate the effectiveness of using
as performance, satisfaction, stress, success, and creativity [59]. Strong chatbots. The findings were lacking and revealed weak evidence sug­
self-esteem is a favorable evaluation of self-worth [60]. Several studies gesting that chatbots can help manage depression, stress, distress, and
suggested that people’s self-esteem can boost their acceptance of sys­ acrophobia. However, using chatbots did not significantly affect sub­
tems [61–64]. For instance, in the context of e-learning acceptance, jective psychological well-being. Thus, the literature was inconclusive
ensuring that an individual has a high self-concept to attain meaningful regarding the relationship between chatbots and anxiety severity, pos­
results [62]. itive and negative effects. Based on the current discussion, ChatGPT has
Self-esteem has considerable implications for ChatGPT use in the potential to improve perceived stress, but further research is needed,
academia due to its association with reducing job anxiety, creating a given the lack of empirical evidence, particularly in academia [76].
positive attitude and optimistic self-belief of having a creative academic
quality in accomplishing tasks related to research, teaching and assess­ H6. Perceived stress increases the use of ChatGPT in academia.
ment by automating tedious components of the task [36,37]. Within
traditional education, the self-esteem construct has been extensively 2.8. Academic integrity (moderator)
explored; however, implementing ChatGPT in academia as an innova­
tive matter remains under research. Little is known about whether Integrity in education and academia requires a dedication to
self-esteem motivates academics to use ChatGPT. Accordingly, the pre­ honesty, fairness, trust, responsibility and respect [77]. Academic
sent work aims to fill this gap by investigating the association between integrity is the practice of researching and completing academic work
self-esteem and using ChatGPT in academia. Against this backdrop, it with fairness and coherence [78,79]. Presently, academic usage of AI is a
can be grounded that academics with higher self-esteem evaluations are trending topic in education with numerous advantages, including

4
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

academic engagement, collaboration, assessment, and accessibility. Yet, Table 1

this technological arms-race has also raised concerns about academic Sample characteristics (n = 702).
honesty and plagiarism [3], which violate educational integrity princi­ Measure Items n %
ples [80].
Gender Male 424 60.4
For instance, academics may find it challenging to effectively assess Female 278 39.6
the understanding of learners on materials when they use chatbot ap­ Age 20–30 years 32 4.6
plications, as it may not accurately reflect their education level [3]. 31–40 years 341 48.6
Similarly, using AI among academic staff brings more technologically 41–50 years 187 26.6
51 years and above 142 20.2
innovative means of perpetrating academic dishonesty and violating the Academic position Postdoctoral Researcher/Fellow/Scholar 15 2.1
norms and standards of academic integrity, including research plagia­ Lecturer/Instructor 205 29.2
rism, teaching practices and service misconduct [80]. The practical Assistant Professor/Senior Lecturer 308 43.9
notion is that “academic integrity hangs in the balance” because using Associate Professor 124 17.7
Professor 50 7.1
ChatGPT in academia is a double-edged sword__ It can either be used to
Experience 1–5 years 177 25.2
save time, boost self-esteem, improve academic self-efficacy, and reduce 6–10 years 190 27.1
stress, commit academic misconduct and plagiarism. Against this 11–15 years 203 28.9
backdrop, it is argued that investigating the moderating role of academic 16 years and above 132 18.8
integrity in using ChatGPT in academia is essential for developing
meaningful insights. Although integrity is well reported in the literature
In terms of data collection technique, we sent over 6000 survey in­
and has been shown to exert a direct impact on academics’ behavior [3,
vitations with URL links to randomly selected users from RG and
77,81]; hardly identify investigations explored its role as moderator to
Academia, while a total of 702 useable responses were retrieved for
strengthen the association between various assumed constructs, espe­
analysis during the period from April 14 to May 15, 2023. The response
cially in the context of using ChatGPT in academia. Accordingly, the
rate of the survey was around 11.7%. This is not surprising as recent
following hypotheses are formulated:
literature with online questionnaires about Academia.edu, RG, or
H7. Academic integrity significantly predicts the use of ChatGPT in Mendeley have obtained a similar response rate, ranging from 8% [86]
academia. to 10% [90]. Prior to carrying out the questionnaire distribution, we
consulted 3 academicians and pilot-tested 20 researchers, and accord­
H7a. Academic integrity positively moderates the association between
ingly, a few revisions were made to attain the reliability and validity of
the time-saving feature and the use of ChatGPT in academia.
the ChatGPT usage model. To mitigate the potential for social desir­
H7b. Academic integrity positively moderates the association between ability bias in the research, a series of meticulous measures were put into
e-WOM and the use of ChatGPT in academia. practice, including ensuring the confidentiality of responses, conducting
a comprehensive pilot test, and employing an online survey format [91];
H7c. Academic integrity positively moderates the association between
[92].
peer influence and the use of ChatGPT in academia.
H7d. Academic integrity positively moderates the association between 3.2. Measurements and analysis tool
self-esteem and using ChatGPT in academia.
In this research, the online survey comprising 47 measurement items
H7e. Academic integrity positively moderates the association between
to evaluate eight constructs, namely time-saving features, e-WOM
academic self-efficacy and the use of ChatGPT in academia.
(electronic word-of-mouth), peer influence, self-esteem, perceived
H7f. Academic integrity positively moderates the association between stress, self-efficacy, academic integrity, and the use of ChatGPT in
perceived stress and the use of ChatGPT in academia. academia. All measurement items were derived from previously vali­
dated scales and carefully rephrased to align with the specific context of
3. Research method ChatGPT usage in academic settings. The measurement of time-saving
features was based on five items adapted from Ref. [31]; while the
3.1. Research design, methodology, and data collection technique assessment of e-WOM involved the adaptation of five items from
Ref. [39]. Peer influence was evaluated using four questions adapted
In this study, quantitative research was adopted due to its suitability from Ref. [46]; and well-established Rosenberg’s self-esteem scale [57]
for examining causal models and relationships. It involved the were adapted to measure self-esteem. We assessed academic
controlled manipulation of a limited set of factors to address theory- self-efficacy by using four items adapted from Ref. [66]; while perceived
driven research questions and hypotheses, ultimately aiming to pro­ stress was evaluated through six items adapted from Ref. [71]. The
vide a comprehensive overview of trends and associations [82]. To measurement of academic integrity drew upon seventeen items adapted
validate the ChatGPT usage model, the study adopted a cross-sectional from Ref. [79]; and the assessment of the use of ChatGPT in academia
research design with a web-based survey distributed to the global aca­ employed five items adapted from Refs. [28,29].
demic community. The internet survey tool is basically undertaken due Participants’ perceptions in the study were assessed using a five-
to its effective features, such as responsiveness, cost-effectiveness and point Likert-type scale, wherein the scale ranged from one, indicating
target population access [83,84], as well as the nature of the present “strongly disagree,” to five, indicating “strongly agree”. The current
research examining a technology-related issue [85]. The target sample study utilized partial least squares structural equation modeling (PLS-
in this research is academic researchers from all around the world, SEM), a robust method for examining intricate interrelationships among
specifically users of ResearchGate (RG) and Academia.edu (Academia) latent constructs [93]. PLS-SEM is a variance-based modeling approach
sites with at least two research items available in their profiles. RG and that has gained popularity in the fields of management and social sci­
Academia have become the two most popular academic social ences due to its capacity to handle small sample sizes, non-normal data
networking sites (ASNSs) in the world today [86,87]. RG and Academia distributions, and complex relationships among latent constructs [94].
have gained immense interest and popularity among researchers Unlike covariance-based SEM (CB-SEM), PLS-SEM is particularly suit­
worldwide with a growing number of users. In 2021, the total number of able for studies that aim to predict outcomes rather than establish causal
users of RG and Academia reached 20 million and 170 million, respec­ relationships among constructs [93].
tively [88]; [89].

5
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

Table 2 31-40-year-old age group, while 26.6% of them were in the 41-50-year-
Measurement model results. old age bracket. With regard to the academic position, the sample
Construct Item Loading Cronbach’s CR AVE comprised 43.9% Assistant Professors/Senior Lecturers, 29.2% Lec­
α turers/Instructors, and 17.7% were Associate Professors. Additionally,
Time-saving feature TSF1 0.723 0.857 0.895 0.631 28.9% of respondents stated that they have 11–15 years of academic
(TSF) TSF2 0.801 experience, while 27.1% had 6–10 years of academic experience.
TSF3 0.739
TSF4 0.826
TSF5 0.875
4.2. Common method bias
Electronic word-of- EWOM1 0.807 0.837 0.891 0.672
mouth (e-WOM) EWOM2 0.876 As an essential analysis, we assess common method bias (CMB) to
EWOM4 0.844 ensure the validity of research results, as it may threaten validity. Hence,
EWOM5 0.748
the current study utilized Harman’s single-factor analysis to determine
Peer influence (PI) PI1 0.945 0.919 0.897 0.688
PI2 0.905 whether a single construct can account for a significant share of variance
PI3 0.711 in the model. The analysis results indicated that the largest single
PI4 0.732 construct explains 25.44% of the variance. This value is lower than the
Self-esteem (SE) SE1 0.894 0.888 0.918 0.693
recommended maximum limit of 50% suggested by Ref. [95]. Conse­
SE2 0.804
SE3 0.887
quently, it can be concluded that there were no validity concerns asso­
SE4 0.839 ciated with CMB in the model. Furthermore, the research performed a
SE5 0.728 comprehensive examination of collinearity by conducting a test of full
Academic self-efficacy ASE1 0.927 0.754 0.848 0.653 collinearity to determine if the variance inflation factor (VIF) values
(ASE) ASE2 0.691
surpassed the maximum threshold of 3.3. The results showed that the
ASE3 0.789
Perceived stress (PS) PS1 0.762 0.817 0.864 0.562 VIF values ranged from 1.09 to 1.80, indicating that the presence of CMB
PS2 0.762 did not threaten the study’s validity.
PS3 0.828
PS4 0.655
PS6 0.730 4.3. Measurement model
Academic integrity AI1 0.913 0.955 0.956 0.735
(AI) AI3 0.902 The evaluation of the reflective model in this research encompassed
AI5 0.920 an assessment of the reliability and validity of the variables. Setting
AI7 0.706
AI9 0.964
standards for internal consistency, discriminant validity, convergent
AI10 0.669 validity, and reliability is crucial. According to Ref. [94]; a factor
AI11 0.765 loading greater than 0.70 is considered acceptable for establishing
AI12 0.964 reliability and validity, while [96] suggests a threshold of 0.4. However
Use of ChatGPT in ChatGPT1 0.896 0.887 0.917 0.691
[94], propose that items with loadings between 0.40 and 0.70 may be
academia (ChatGPT) ChatGPT2 0.853
ChatGPT3 0.899 considered for removal only if their exclusion would improve composite
ChatGPT4 0.852 reliability. In this study, eight items (e-WOM3, SE6, ASE4, PS6, AI2, AI4,
ChatGPT5 0.627 AI6, and AI8) needed to be removed from the initial set of 47 survey
items. Convergent validity was supported by the average variance
extracted (AVE) values, which exceeded the recommended threshold of
4. Data analysis and findings
0.50, as suggested by Ref. [94]. Table 2 presents Cronbach’s α values for
all latent constructs, surpassing the required threshold of 0.70, indi­
4.1. Sample profile
cating strong internal consistency of the measures. Additionally, the
composite reliability (CR) values for each latent variable exceeded 0.70,
The questionnaire revealed a noteworthy gender disparity among the
affirming the reliability of the latent constructs. Ensuring the reliability
respondents, with more than half of (60.4%) of respondents being male,
and validity of the variables is essential to prevent errors in the research
whereas (39.6%) identifying as female. The demographic information
results and draw accurate conclusions.
displayed in Table 1 also shows that 48.6% of respondents fell within the
The evaluation of the measurement model includes an important

Table 3
Discriminant validity.
Fornell-Larcker criterion 1 2 3 4 5 6 7 8

1 TSF 0.795
2 e-WOM 0.477 0.820
3 PI − 0.120 0.282 0.830
4 SE 0.232 0.152 − 0.075 0.833
5 ASE 0.095 0.171 0.105 0.083 0.808
6 PS 0.335 0.121 − 0.300 0.131 − 0.335 0.750
7 AI 0.212 0.155 0.145 0.063 − 0.243 0.083 0.857
8 ChatGPT 0.414 0.281 − 0.332 0.205 0.189 0.322 − 0.254 0.831

HTMT 0.85 criterion 1 2 3 4 5 6 7 8

1 TSF
2 e-WOM 0.613
3 PI 0.228 0.393
4 SE 0.276 0.174 0.092
5 ASE 0.254 0.265 0.305 0.206
6 PS 0.382 0.235 0.286 0.246 0.43
7 AI 0.279 0.289 0.113 0.078 0.253 0.231
8 ChatGPT 0.429 0.307 0.193 0.233 0.198 0.326 0.199

6
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

Table 4 results provide evidence of discriminant validity, demonstrating that

Structural model results. each latent variable measures a distinct construct that is not excessively
Hypotheses Path β Std t- p- Supported? correlated with other constructs. Besides, the heterotrait-monotrait
error value value (HTMT) ratio was used to measure the discriminant validity, quanti­
H1 TSF - > 0.233 0.039 5.918 0.000 Yes fying the correlation between two items of the same construct. The
ChatGPT analysis shows that all construct ratios were below the conservative
H2 e-WOM - > 0.233 0.042 5.539 0.000 Yes threshold of 0.85 [98], indicating that the constructs were sufficiently
ChatGPT different from each other. Statistical tests were conducted to evaluate
H3 PI - > − 0.279 0.152 1.835 0.033 No
ChatGPT
the validity and reliability of the measurement model. These outcomes
H4 SE - > 0.075 0.033 2.263 0.012 Yes confirmed that the measurement scales employed for the factors of in­
ChatGPT terest were both valid and reliable, allowing for the structural model’s
H5 ASE - > 0.146 0.052 2.837 0.002 Yes testing.
ChatGPT
H6 PS - > 0.195 0.028 6.835 0.000 Yes
ChatGPT
H7 AI - > − 0.284 0.036 7.886 0.000 Yes 4.4. Structural model
ChatGPT
H7a TSF*AI - > − 0.173 0.045 3.882 0.000 No A structural model evaluation is done to assess how the exogenous
ChatGPT
constructs influence the endogenous construct. The evaluation involves
H7b e-WOM*AI − 0.039 0.057 0.683 0.247 No
- > ChatGPT examining collinearity amongst constructs, determining the significance
H7c PI*AI - > − 0.333 0.129 2.577 0.005 No of the hypothesized relationships, measuring the explained variance
ChatGPT (R2), evaluating the model’s predictive relevance (Q2), and determining
H7d SE*AI - > 0.206 0.055 3.732 0.000 Yes the effect size (f2) [99]. A bootstrapping approach was used to measure
ChatGPT
the significance of the hypothesized relationships, drawing 5000 sam­
H7e ASE*AI - > 0.050 0.044 1.147 0.126 No
ChatGPT ples. The statistical findings of the analysis indicated that all the path
H7f PS*AI - > 0.150 0.069 2.188 0.014 Yes relationships were significant, with a bootstrap critical t-value above
ChatGPT ±1.65 (one-tailed test).
The analysis outcomes indicate strong positive associations between
several constructs and the use of ChatGPT in academia. Specifically,
component referred to as discriminant validity, which was measured
Time-saving feature, e-WOM, self-esteem, self-efficacy, and stress are
using the approach recommended by Ref. [97]. The findings, as shown
positively linked to the Use of ChatGPT in academia (β = 0.233, p =
in Table 3, present that the square roots of AVE for the latent constructs
0.000), (β = 0.233, p = 0.000), (β = 0.075, p = 0.012), (β = 0.146, p =
were greater than the inter-construct correlation coefficients. These
0.002), and (β = 0.195, p = 0.000), respectively. However, peer

Fig. 3. PLS results for structural model assessment.

Notes: Values in brackets refer to t-statistics; values without brackets refer to standardized beta.

7
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

influence is found to have a negative relationship with the Use of implies that academic self-esteem has considerable implications for
ChatGPT in academia (β = − 0.279, p = 0.033). Further, academic ChatGPT use in academia due to its association with reducing job anx­
integrity significantly and negatively relates to ChatGPT usage (β = iety, creating a positive attitude and optimistic self-belief of having a
− 0.284, p = 0.000). Accordingly, these findings support hypotheses H1, creative academic quality in accomplishing tasks related to research,
H2, H4, H5, H6 and H7, while not H3. teaching and assessment by automating tedious components of the task
Regarding the interaction relationships, Table 4 presents the results. [36,37]. Likewise, academic self-efficacy – an academician’s confidence
The interaction effects of time-saving feature × academic integrity, self- in achieving academic success and performance goals – had a positive
esteem × academic integrity, and perceived stress × academic integrity influence on ChatGPT usage in academic work. Employing AI chatbot
on using ChatGPT in academia are significantly positive. However, the ChatGPT in academia and higher education could boost academic
moderating effect of academic integrity on peer influence and usage is self-efficacy by giving academics access to a complex and powerful tool
significantly negative. Interactions of e-WOM × academic integrity, and that enhances their capabilities and reduces research, teaching and
self-efficacy × academic integrity have no effect. Hence, this supports learning workloads (Jürgen-Rudolph et al., 2023). Perceived stress also
hypotheses H7d, and H7f, while H7a, H7b, H7c, and H7e are not plays a significant role in fostering people’s ChatGPT adoption in
supported. academia. ChatGPT is described as a helpful tool for academics who
Looking at the model’s explanatory power, the coefficient of deter­ struggle with time management, task accomplishment, workload and
mination (R2) for the endogenous variable, ChatGPT use in academia, productivity. By providing prompt and accurate content, the use of
exceeds the recommended threshold value of 0.02. The R2 value in­ ChatGPT can potentially manage depression, stress, distress, anxiety,
dicates that the model has a high degree of predictive accuracy, and the and acrophobia in relation to academic capabilities.
variables included in the study collectively explain 43.3% of the vari­ Intriguingly, some meaningful outcomes were also extracted from
ance in the use of ChatGPT in academia (Fig. 3). To assess the model fit, the moderating effect of academic integrity in the ChatGPT usage model.
the study employed the blindfolding procedure in PLS-SEM, which Initially, the direct effect of academic integrity on academicians’
yielded a Q2 value greater than zero (0.275) for the endogenous vari­ adoption of ChatGPT was significantly negative. This implies that the
able. This indicates the predictive relevance of the model. The structural higher the academic integrity among academicians, the lower their
assessment results are presented in Table 4. usage of ChatGPT in their work. Some academicians may claim that
adopting ChatGPT is as unethical as ordinary plagiarism and academic
5. Discussion and conclusions dishonesty, which will likely cause catastrophic issues for the academic
community. Due to the significant role of academic integrity, specif­
5.1. Discussion ically in the era of AI platforms, we tested the moderating effect of ac­
ademic integrity on the relationships between ChatGPT usage and its
ChatGPT has become a hot topic in the academic community as a determinants. The analysis demonstrates that academic integrity-
language model driven by AI, offering a wide variety of profound ben­ moderated interaction of time-saving feature on ChatGPT usage was
efits, including academic content generation, accessibility, collabora­ significantly negative, in contrast to the previously hypothesized posi­
tion, and assessment. Nevertheless, this technological arms-race is tive relationship. This shows that the higher the perceived academic
raising concerns about academic honesty and plagiarism, facilitating the integrity in relation to ChatGPT in academia, the lower the academics
violation of ethical principles of the academic setting. In this research, believe that using ChatGPT can be beneficial in saving their time, but not
an attempt has been made to pinpoint the motivations that drive aca­ to the detriment of the principles of research ethics. The research also
demicians to use ChatGPT in their academic work, stressing the role of suggests that the association between self-esteem and academicians’
academic integrity in shaping their behavior to uptake such an ChatGPT usage is contingent on the level of their academic integrity.
innovation. This means that academicians perceive that their integrity regarding
This work is a theoretically informed study that developed and ChatGPT usage could enhance self-esteem, thus fostering their adoption
validated a comprehensive adoption model highlighting ChatGPT usage of such an innovative tool in academic settings. Furthermore, academic
behavior among researchers by applying the SLT theory. The outcomes integrity strengthens the positive relationship between perceived stress
of the empirical analysis indicated that the time-saving feature in rela­ and ChatGPT usage behavior. Academicians with high levels of aca­
tion to ChatGPT exerts a significant positive influence on academicians’ demic integrity could strongly use ChatGPT in their academic work as a
behavior toward the use of ChatGPT. This implies that the time-saving potentially effective way to relieve anxiety and stress associated with
feature of ChatGPT for integration and information processing into re­ feeling overwhelmed by the workload. However, contrary to what has
searchers’ academic work has made the timeliness element increasingly been hypothesized, the existence of academic integrity strengthened the
essential to promptly generate content, accomplish tasks, boost pro­ negative association between social influence and ChatGPT usage
ductivity, and achieve goals [36]. We also found that e-WOM has a behavior. This implies that the stronger the influence of academic peers
positive impact on ChatGPT usage in academia. That is, being regarded (e.g., researchers or scholars and colleagues) with high levels of aca­
as the tactile information provided by web communities a more reliable demic integrity, the lower their usage behavior of ChatGPT.
source to quantify the usefulness and benefits of ChatGPT, e-WOM
communication among academics likely has a more significant impact to 5.2. Theoretical implications
be integrated into their practices. This result corresponds with prior
studies but from different contexts (Daugherty & Hoffman, 2014; [42]. This attempt provides invaluable insights into the adoption of AI-
Unexpectedly, the results demonstrate that social influence negatively powered chatbots, such as ChatGPT, within the academic sphere. It
affected academicians’ behavior toward using ChatGPT. This indicates empirically explores a relatively new discipline, enriching the existing
that academics’ peers, such as researchers, scholars and colleagues, literature by considering various significant factors that shape the
negatively influence their decisions to use ChatGPT in the academic adoption behavior of such an innovative tool. Furthermore, it sheds light
setting. This interesting outcome is likely traced to the fact that the use on the crucial aspect of academic integrity in AI adoption among re­
of ChatGPT raises security, ethical and legal concerns related to trans­ searchers. These findings establish a robust groundwork for future
parency, bias, misuse, privacy and copyright. A few research, but in studies in this domain, emerging several profound implications that can
different settings, has affirmed the conclusion of the negative influence advance the AI adoption literature.
of social norms on behavioral decisions [100]. First, given the sensitive timing of this research topic and the existing
The analysis also shows that academicians’ behavioral intention to theoretical gaps in the literature, which highlight the scarcity of
use ChatGPT was positively influenced by academic self-esteem. It research on ChatGPT adoption in academic settings, this study is likely

8
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

Table 5 5.4. Limitations and directions for future studies

Relevant policies for AI platforms adoption in academia.
⁃ Stakeholders, such as higher education institutions, publishing companies, sponsors Despite the profound contributions of this study to theory and policy,
and other entities associated with academic and research ethics, should initiate some limitations are to be highlighted. RG and Academia were chosen in
sensitisation programs to build sufficient guidelines for AI chatbots usage (e.g., this study as the most popular ASNSs; however, they may not represent
ChatGPT) in academic and research settings in particular. Simply banning the use of
all other ASNSs users. Future research may look into other vital ASNSs
AI-driven chatbots in academia is a practical impossibility and would not overcome
issues raised; thus, stakeholders should consider legal ways to embed them into the like LinkedIn, Mendeley and Google Scholar. Moreover, the study solely
academic process. relied on the SLT theory to explain the behavior of academicians in using
⁃ Since the use of AI chatbots ChatGPT has become a reality and inevitable in the ChatGPT in academic settings. By framing this research as a preliminary
realm of academic researchers and students; this research suggests that any investigation, future studies may extend this theoretical framework to
academic content, including research for academic purposes, is to be examined and
include other relevant technology usage theories, such as UTAUT and
validated. Although academic staff and researchers strive for self-esteem and effi­
cacy through achieving ideal performance indicators, it is necessary to familiarize TAM, that can capture the nuances of this significant topic. Researchers
them with ChatGPT limitations, e.g., having limited updated knowledge, generating may also expand the target population to include students and academic
incorrect or falsified information, and relying on biased data. staff to understand better their motivations for shaping AI language
⁃ As a further reassurance to ensure the protection of research ethics, it is of utmost
models’ adoption. Further, other crucial factors, such as academic per­
importance to ethically use all AI-generated content platforms. As such, there should
be cooperation and integration between AI language model programmers, academic formance, competence, and personal best goals, in the ChatGPT era will
institutions, publishers and any other relevant stakeholders to work together to likely warrant investigation by future studies. Since this study mainly
curtail the spread of unethical behaviors, including academic dishonesty, plagiarism employs quantitative cross-sectional research, it is recommended to
and fake citations, to protect researchers’ rights. conduct future quantitative research to explore the finer-grained aspects
⁃ Stakeholders, such as AI language model programmers, academic institutions,
and delve deeper into the perspectives of ChatGPT adoption behavior.
publishers and any other relevant stakeholders, are also advised to concert efforts in
making AI chatbots ChatGPT or upcoming update releases safe with the ability to
detect unethical actions. Author statement
⁃ To ensure academic integrity, some scholars have suggested including the AI
language model as a co-author in AI-generated scientific papers [8,101]. However,
Saeed Awadh Bin-Nashwan: Conceptualization, Methodology,
this is likely to result in similar content but in different papers without any
acknowledgment or citation, which can be known as self-plagiarism. It is crucial
Software, Data curation, Writing- Original draft preparation, Writing -
that publishers, editors, sponsors, as well as academic institutions clearly specify Review & Editing, Project administration. Mouad Sadallah: Writing-
research ethics for authorship in terms of allowing or not allowing adding these Original draft preparation, Conceptualization, Software, Formal anal­
tools as co-authors in case a researcher plans to involve AI language models in ysis. Mohamed Bouteraa: Writing- Original draft preparation, Visual­
writing.
ization, Investigation, Writing - Review & Editing.

the first empirical investigation of OpenAI applications among academic Data availability
researchers. Secondly, drawn on social cognitive aspects, the study en­
riches the literature by examining ChatGPT usage among academicians Data will be made available on request.
from all over the globe using the most popular ASNSs (RG and
Academia). Thirdly, the model is built on various crucial determinants References
that could drive academic community adoption of ChatGPT, which have
[1] S. Sinha, L. Burd, J. Preez, How ChatGPT could revolutionize academia, available
yet to be empirically tested in AI adoption, including time-saving at: https://spectrum.ieee.org/how-chatgpt-could-revolutionize-academia, 2023.
feature, e-WOM, peer influence, academic self-esteem, academic self- [2] E. Koc, S. Hatipoglu, O. Kivrak, C. Celik, K. Koc, Houston, we have a problem!:
efficacy and perceived stress. Accordingly, this developed and vali­ the use of ChatGPT in responding to customer complaints, Technol. Soc. 74
(2023), 102333.
dated model has the potential to serve as a launching point for future
[3] D.R. Cotton, P.A. Cotton, J.R. Shipway, Chatting and cheating: ensuring academic
studies in the discipline of AI applications, providing an opportunity to integrity in the era of ChatGPT, Innovat. Educ. Teach. Int. (2023), https://doi.
gain deeper insights into this innovative technology. In addition, we org/10.1080/14703297.2023.2190148.
[4] M. Timothy, 3 reasons why ChatGPT became the fastest growing app of all time,
provide significant and timely evidence on academic integrity in rela­
available at: https://www.makeuseof.com/how-chatgpt-became-fastest-gro
tion to ChatGPT in academic settings. Academic integrity demonstrated wing-app/, 2023.
significant moderating effects on the ChatGPT usage model in academia. [5] Similarweb, chat.openai.com overview. https://www.similarweb.
com/website/chat.openai.com/#overview, 2023.
[6] F. Duarte, Number of ChatGPT users (2023), available at:, 2023 https://
5.3. Practical implications explodingtopics.com/blog/chatgpt-users.
[7] F. Qasem, ChatGPT in scientific and academic research: future fears and
As scholarly communities still lack clear principles and instructions reassurances, Libr. Hi Tech News 40 (3) (2023) 30–32.
[8] C.K. Lo, What is the impact of ChatGPT on education? A rapid review of the
on using AI applications in academic work, significant and timely literature, Educ. Sci. 13 (4) (2023) 410.
empirical evidence has emerged from this study that can inform policies [9] U.K. Hisan, M.M. Amri, Artificial intelligence for human life: a critical opinion
and practices in academic settings and higher education. First, the re­ from medical bioethics perspective–Part I, Journal of Public Health Sciences 1 (2)
(2023) 100–111.
sults shed light on the drivers that fuel academicians’ usage of ChatGPT. [10] D. Liu, As Uni Goes Back, Here’s How Teachers and Students Can Use ChatGPT to
Academic researchers could be overwhelmed by heavy workloads, and Save Time and Improve Learning, 2023 available at: https://academicmatters.
thus, they perceive that using ChatGPT can help effectively with time, ca/as-uni-goes-back-heres-how-teachers-and-students-can-use-chatgpt-to-save
-time-and-improve-learning/#:~:text=ChatGPT%20can%20help%20teachers%
anxiety, and stress management, as well as to attain their academic self-
20save,choice%20or%20short%2Danswer%20questions.
esteem – academicians’ confidence about their ability (who they are and [11] Shiri, A. ChatGPT and Academic Integrity. SSRN 2023, 4360052. Available at htt
what they can do). On the other hand, there is a severe issue raised ps://ssrn.com/abstract=4360052.
[12] M. Alser, E. Waisberg, Concerns with the usage of ChatGPT in academia and
regarding academic integrity and credibility in using AI-generated
medicine: a viewpoint, American Journal of Medicine Open 9 (2023), 100036.
content platforms. Based on the analysis, we argue that the higher the [13] A. Tlili, B. Shehata, M.A. Adarkwah, A. Bozkurt, D.T. Hickey, R. Huang,
academic integrity among academicians, the lower their usage of B. Agyemang, What if the devil is my guardian angel: ChatGPT as a case study of
ChatGPT in their work. As such, some preliminary and relevant policies using chatbots in education, Smart Learning Environments 10 (1) (2023) 15.
[14] K. Huang, Alarmed by AI Chatbots, Universities Start Revamping How They
could be suggested in Table 5. Teach, 2023 (New York Times).
[15] K. Jimenez, ’This shouldn’t be a surprise’ the education community shares mixed
reactions to ChatGPT. https://www.usatoday.com/story/news/education/20

9
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

23/01/30/chatgpt-going-banned-teachers-sound-alarm-new-ai-tech/110 of Marketing - ESIC 23 (3) (2019) 397–414, https://doi.org/10.1108/SJME-08-

69593002/, 2023. 2019-0067.
[16] Y.K. Dwivedi, N. Kshetri, L. Hughes, E.L. Slade, A. Jeyaraj, A.K. Kar, R. Wright, [45] G. Mugny, F. Butera, M. Sanchez Mazas, J.A. Pérez, Judgements in conflict: the
“So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, conflict elaboration theory of social influence. Perception evaluation
challenges and implications of generative conversational AI for research, practice interpretation, Swiss Monographs in Psychology 3 (1995) 160–168.
and policy, Int. J. Inf. Manag. 71 (2023), 102642. [46] I. Ajzen, Constructing a TPB Questionnaire: Conceptual and Methodological
[17] M. Farrokhnia, S.K. Banihashem, O. Noroozi, A. Wals, A SWOT analysis of Considerations, available at:, 2002 https://people.umass.edu/aizen/pdf/tpb.
ChatGPT: implications for educational practice and research, Innovat. Educ. measurement.pdf.
Teach. Int. (2023), https://doi.org/10.1080/14703297.2023.2195846. [47] P. Mehta, C. Jebarajakirthy, H.I. Maseeh, A. Anubha, R. Saha, K. Dhanda,
[18] B.D. Lund, T. Wang, Chatting about ChatGPT: how may AI and GPT impact Artificial intelligence in marketing: a meta-analytic review, Psychol. Market. 39
academia and libraries? Libr. Hi Tech News 40 (3) (2023) 26–29. (11) (2022) 2013–2038.
[19] M. Rahman, H.J.R. Terano, N. Rahman, A. Salamzadeh, S. Rahaman, ChatGPT [48] H.C. Kelman, Further Thoughts on the Processes of Compliance, Identification,
and academic research: a review and recommendations based on practical and Internalisation. Perspectives on Social Power, 1974.
examples, Journal of Education, Management and Development Studies 3 (1) [49] Venkatesh, Morris, Davis, Davis, User acceptance of information technology:
(2023) 1–12, https://doi.org/10.52631/jemds.v3i1.175. toward a unified view, MIS Q. 27 (3) (2003) 425, https://doi.org/10.2307/
[20] M. Dowling, B. Lucey, ChatGPT for (finance) research: the Bananarama 30036540.
conjecture, Finance Res. Lett. 53 (2023), 103662. [50] K. Al-Saedi, M. Al-Emran, T. Ramayah, E. Abusham, Developing a general
[21] A. Bandura, Human agency in social cognitive theory, Am. Psychol. 44 (9) (1989) extended UTAUT model for M-payment adoption, Technol. Soc. 62 (2020),
1175–1184. 101293, https://doi.org/10.1016/j.techsoc.2020.101293.
[22] H. Boateng, D.R. Adam, A.F. Okoe, T. Anning-Dorson, Assessing the determinants [51] S.A. Bin-Nashwan, Toward diffusion of e-Zakat initiatives amid the COVID-19
of internet banking adoption intentions: a social cognitive theory perspective, crisis and beyond, Foresight 24 (2) (2022) 141–158.
Comput. Hum. Behav. 65 (2016) 468–478. [52] Q. Jia, X. Xu, M. Zhou, H. Liu, F. Chang, Exploring the determinants of continuous
[23] E. Barnett, M.A. Casper, Definition of "social environment, Am. J. Publ. Health 91 intention in TikTok from the perspective of social influence: a mixed approach of
(3) (2001) 465. SEM and fsQCA, Journal of Electronic Business & Digital Economics (2023),
[24] M. Chen, Y. Cao, Y. Liang, Determinants of Open Government Data Usage: https://doi.org/10.1108/JEBDE-07-2022-0016.
Integrating Trust Theory and Social Cognitive Theory, Government Information [53] G.A. Abbasi, L.Y. Tiew, J. Tang, Y.-N. Goh, R. Thurasamy, The adoption of
Quarterly, 2023, 101857. cryptocurrency as a disruptive force: deep learning-based dual stage structural
[25] A.S. Almogren, N.A. Aljammaz, The integrated social cognitive theory with the equation modelling and artificial neural network analysis, PLoS One 16 (3)
TAM model: the impact of M-learning in King Saud University art education, (2021), e0247582.
Front. Psychol. 13 (2022), 1050532. [54] H.M. Al-Hattami, Understanding perceptions of academics toward technology
[26] D. Wu, H. Gu, S. Gu, H. You, Individual motivation and social influence: a study acceptance in accounting education, Heliyon 9 (1) (2023), e13141, https://doi.
of telemedicine adoption in China based on social cognitive theory, Health Policy org/10.1016/j.heliyon.2023.e13141.
and Technology 10 (3) (2021), 100525. [55] M. Bouteraa, R.H. Raja Rizal Iskandar, Z. Zairani, Challenges affecting bank
[27] S.A. Bin-Nashwan, A.E.A. Ismaiel, A. Muneeza, M.Y. Isa, Adoption of ZakaTech in consumers’ intention to adopt green banking technology in the UAE: a UTAUT-
the time of COVID-19: cross-country and gender differences, Journal of Islamic based mixed-methods approach, Journal of Islamic Marketing (2022), https://
Marketing (2023), https://doi.org/10.1108/JIMA-08-2021-0278. doi.org/10.1108/JIMA-02-2022-0039.
[28] E. Katz, H. Haas, M. Gurevitch, On the use of the mass media for important things, [56] T.A. Judge, E.A. Locke, C.C. Durham, A.N. Kluger, Dispositional effects on job and
Am. Socio. Rev. (1973) 164–181, https://doi.org/10.2307/2094393. life satisfaction: the role of core evaluations, J. Appl. Psychol. 83 (1) (1998)
[29] Z. Papacharissi, A.M. Rubin, Predictors of internet use, J. Broadcast. Electron. 17–34, https://doi.org/10.1037/0021-9010.83.1.17.
Media 44 (2) (2000) 175–196. [57] M. Rosenberg, Society and the Adolescent Self-Image, Princeton University Press,
[30] G.S. Becker, A theory of the allocation of time, Econ. J. 75 (299) (1965) 493–517. Princeton, New Jersey, USA, 1965.
[31] K. Martha, Factors affecting academic performance of undergraduate students at [58] J. Neroni, C. Meijs, P.A. Kirschner, K.M. Xu, R.H.M. de Groot, Academic self-
Uganda Christian University, Caribbean Teaching Scholar 1 (2) (2010) 79–92. efficacy, self-esteem, and grit in higher online education: consistency of interests
[32] F. Xu, S. Huang, S. Li, Time, money, or convenience: what determines Chinese predicts academic success, Soc. Psychol. Educ. 25 (4) (2022) 951–975, https://
consumers’ continuance usage intention and behavior of using tourism mobile doi.org/10.1007/s11218-022-09696-5.
apps? Int. J. Cult. Tourism Hospit. Res. 13 (3) (2019) 288–302, https://doi.org/ [59] A. Di Fabio, L. Palazzeschi, Core self-evaluation, in: The Wiley Encyclopedia of
10.1108/IJCTHR-04-2018-0052. Personality and Individual Differences, Wiley, 2020, pp. 83–87, https://doi.org/
[33] K.S.P. Ng, J. Zhang, J.W.C. Wong, K.K. Luo, Internal factors, external factors and 10.1002/9781119547174.ch191.
behavioral intention toward food delivery apps (FDAs), Br. Food J. (2023), [60] M. Rouault, G.-J. Will, S.M. Fleming, R.J. Dolan, Low self-esteem and the
https://doi.org/10.1108/BFJ-07-2022-0586. formation of global self-performance estimates in emerging adulthood, Transl.
[34] C. Hong, H. Choi, E.-K. Choi, H.-W. Joung, Factors affecting customer intention to Psychiatry 12 (1) (2022) 272, https://doi.org/10.1038/s41398-022-02031-8.
use online food delivery services before and during the COVID-19 pandemic, [61] A. Mehrabi, A. Falakdami, A. Mollaei, P. Takasi, P.G. Vajargah, H. Jafari, S.M.
J. Hospit. Tourism Manag. 48 (2021) 509–518, https://doi.org/10.1016/j. H. Mazloum, N. Rahimzadeh, M.J. Ghazanfari, A.E. Zeydi, M. Mobayen,
jhtm.2021.08.012. S. Karkhah, A systematic review of self-esteem and related factors among burns
[35] E.H.T. Yapp, S. Kataraian, Key determinants of continuance usage intention: an patients, Annals of Medicine & Surgery 84 (2022), https://doi.org/10.1016/j.
empirical study of mobile food delivery apps among Malaysians, International amsu.2022.104811.
Academic Symposium of Social Science 2022 (2022) 15, https://doi.org/ [62] H.V. Osei, K.O. Kwateng, K.A. Boateng, Integration of personality trait,
10.3390/proceedings2022082015. motivation and UTAUT 2 to understand e-learning adoption in the era of COVID-
[36] X. Tian, Z. Risha, I. Ahmed, A.B. Lekshmi Narayanan, J. Biehl, Let’s talk it out, 19 pandemic, Educ. Inf. Technol. 27 (8) (2022) 10705–10730, https://doi.org/
Proceedings of the ACM on Human-Computer Interaction 5 (CSCW1) (2021) 10.1007/s10639-022-11047-y.
1–32, https://doi.org/10.1145/3449171. [63] A. Tewari, R. Singh, S. Mathur, S. Pande, A modified UTAUT framework to
[37] C. Kairu, A Review of Chatbots in Higher Education: Current Trends, predict students’ intention to adopt online learning: moderating role of openness
Applications, Challenges and Future Implications, 2022, pp. 9928–9935, https:// to change, The International Journal of Information and Learning Technology 40
doi.org/10.21125/inted.2022.2615. (2) (2023) 130–147, https://doi.org/10.1108/IJILT-04-2022-0093.
[38] S.W. Litvin, R.E. Goldsmith, B. Pan, Electronic word-of-mouth in hospitality and [64] H. Yang, S. Zhang, Social media affordances and fatigue: the role of privacy
tourism management, Tourism Manag. 29 (3) (2008) 458–468, https://doi.org/ concerns, impression management concerns, and self-esteem, Technol. Soc. 71
10.1016/j.tourman.2007.05.011. (2022), 102142, https://doi.org/10.1016/j.techsoc.2022.102142.
[39] A. Reyes-Menendez, J.R. Saura, J.G. Martinez-Navalon, The impact of e-WOM on [65] M. Bong, E.M. Skaalvik, Academic self-concept and self-efficacy: how different
hotels management reputation: exploring tripadvisor review credibility with the are they really? Educ. Psychol. Rev. 15 (1) (2003) 1–40, https://doi.org/
ELM model, IEEE Access 7 (2019) 68868–68877. 10.1023/A:1021302408382.
[40] T. Hennig-Thurau, K.P. Gwinner, G. Walsh, D.D. Gremler, Electronic word-of- [66] C. Midgley, M.L. Maehr, L.Z. Hruda, E. Anderman, L. Anderman, K.E. Freeman,
mouth via consumer-opinion platforms: what motivates consumers to articulate T. Urdan, Manual for the Patterns of Adaptive Learning Scales, University of
themselves on the Internet? J. Interact. Market. 18 (1) (2004) 38–52, https://doi. Michigan, Ann Arbor, 2000, pp. 734–763.
org/10.1002/dir.10073. [67] B. Cope, M. Kalantzis, D. Searsmith, Artificial intelligence for education:
[41] T.T.M.D. Do, L.N. Pereira, Understanding Vietnamese consumers’ perception and knowledge and its assessment in AI-enabled learning ecologies, Educ. Philos.
word-of-mouth intentions towards Airbnb, Journal of Hospitality and Tourism Theor. 53 (12) (2021) 1229–1245, https://doi.org/10.1080/
Technology 14 (2) (2023) 83–101, https://doi.org/10.1108/JHTT-12-2020- 00131857.2020.1728732.
0321. [68] Jürgen Rudolph, Samson Tan, Shannon Tan, ChatGPT: bullshit spewer or the end
[42] Y.K. Liao, W.Y. Wu, T.Q. Le, T.T.T. Phung, The integration of the technology of traditional assessments in higher education? Journal of Applied Learning &
acceptance model and value-based adoption model to study the adoption of e- Teaching 6 (1) (2023) https://doi.org/10.37074/jalt.2023.6.1.9.
learning: the moderating role of e-WOM, Sustainability 14 (2) (2022) 815. [69] A.M. Cox, Exploring the impact of Artificial Intelligence and robots on higher
[43] P. Kusawat, S. Teerakapibal, Cross-cultural electronic word-of-mouth: a education through literature-based design fictions, International Journal of
systematic literature review, Spanish Journal of Marketing - ESIC (2022), https:// Educational Technology in Higher Education 18 (1) (2021) 3, https://doi.org/
doi.org/10.1108/SJME-06-2021-0116. 10.1186/s41239-020-00237-8.
[44] C. Orús, R. Gurrea, S. Ibáñez-Sánchez, The impact of consumers’ positive online [70] S. Fisher, Stress and Strategy, Routledge, 2015, https://doi.org/10.4324/
recommendations on the omnichannel webrooming experience, Spanish Journal 9781315627212.

10
S.A. Bin-Nashwan et al. Technology in Society 75 (2023) 102370

[71] S. Cohen, T. Kamarck, R. Mermelstein, A global measure of perceived stress, [84] R.D. Fricker, M. Schonlau, Advantages and disadvantages of Internet research
J. Health Soc. Behav. 24 (4) (1983) 385–396. surveys: evidence from the literature, Field Methods 14 (4) (2002) 347–367.
[72] J. Heo, S. Han, Effects of motivation, academic stress and age in predicting self- [85] M.J. Kim, C.M. Hall, Do value-attitude-behavior and personality affect
directed learning readiness (SDLR): focused on online college students, Educ. Inf. sustainability crowdfunding initiatives? J. Environ. Manag. 280 (1) (2021),
Technol. 23 (1) (2018) 61–71, https://doi.org/10.1007/s10639-017-9585-2. 111827.
[73] M.H. Almomani, M. Rababa, F. Alzoubi, K. Alnuaimi, A. Alnatour, R.A. Ali, Effects [86] J. Lee, S. Oh, H. Dong, F. Wang, G. Burnett, Motivations for self-archiving on an
of a health education intervention on knowledge and attitudes towards chronic academic social networking site: a study on researchgate, Journal of the
non-communicable diseases among undergraduate students in Jordan, Nursing Association for Information Science and Technology 70 (6) (2019) 563–574.
Open 8 (1) (2021) 333–342, https://doi.org/10.1002/nop2.634. [87] S. Mason, Adoption and usage of academic social networks: a Japan case study,
[74] N. Mheidly, M.Y. Fares, J. Fares, Coping with stress and burnout associated with Scientometrics 122 (3) (2020) 1751–1767.
telecommunication and online learning, Front. Public Health 8 (2020), https:// [88] ResearchGate, Progress report, available at: https://www.researchgate.net/p
doi.org/10.3389/fpubh.2020.574969. rogress-report-2021, 2021.
[75] V.T.T. Nguyen, H.-L. Chen, Examining impacts of information system success and [89] Academia.edu, About us. https://www.academia.edu/about, 2021.
perceived stress on students’ self-regulated learning mediated by intrinsic [90] S. Manca, M. Ranieri, Networked scholarship and motivations for social media
motivation in online learning environments: second-order structural equation use in scholarly communication, Int. Rev. Res. Open Dist. Learn. 18 (2) (2017)
modelling analyses, in: Education and Information Technologies, 2023, https:// 123–138.
doi.org/10.1007/s10639-023-11685-w. [91] R.B. Larson, Controlling social desirability bias, Int. J. Mark. Res. 61 (5) (2019)
[76] A.A. Abd-Alrazaq, A. Rababeh, M. Alajlani, B.M. Bewick, M. Househ, 534–547.
Effectiveness and safety of using chatbots to improve mental health: systematic [92] D. Dodou, J.C. de Winter, Social desirability is the same in offline, online, and
review and meta-analysis, J. Med. Internet Res. 22 (7) (2022), e16021, https:// paper surveys: a meta-analysis, Comput. Hum. Behav. 36 (2014) 487–495.
doi.org/10.2196/16021. [93] M. Sadallah, H. Abdul-Jabbar, S.A. Bin-Nashwan, S.A. Abdul Aziz, Alms tax
[77] O.L. Holden, M.E. Norris, V.A. Kuhlmeier, Academic integrity in online (ZAKAT) compliance intention among entrepreneurs from a social cognitive
assessment: a research review, Frontiers in Education 6 (2021), https://doi.org/ perspective: the moderating role of knowledge, Journal of Islamic Accounting and
10.3389/feduc.2021.639814. Business Research (2023), https://doi.org/10.1108/JIABR-04-2022-0104.
[78] J.G. Guerrero-Dib, L. Portales, Y. Heredia-Escorza, Impact of academic integrity [94] J.F. Hair, G.T. Hult, C.M. Ringle, M. Sarstedt, A Primer on Partial Least Squares
on workplace ethical behaviour, International Journal for Educational Integrity Structural Equation Modeling (PLS-SEM), second ed., Sage Publications, London,
16 (1) (2020) 2, https://doi.org/10.1007/s40979-020-0051-3. 2017.
[79] Z. Ramdani, Construction of academic integrity scale, International Journal of [95] P.M. Podsakoff, S.B. Mackenzie, J. Lee, N.P. Podsakoff, Common method biases in
Research Studies in Psychology 7 (1) (2018) 87–97. behavioral research, a critical review of the literature and recommended
[80] J. Roe, M. Perkins, What are Automated Paraphrasing Tools and how do we remedies, J. Appl. Psychol. 88 (5) (2003) 879–903.
address them? A review of a growing threat to academic integrity, International [96] J. Hulland, Use of partial least squares (PLS) in strategic management research: a
Journal for Educational Integrity 18 (1) (2022) 15, https://doi.org/10.1007/ review of four recent studies, Strat. Manag. J. 20 (2) (1999) 195–204.
s40979-022-00109-w. [97] C. Fornell, D. Larcker, Evaluating structural equation models with unobservable
[81] A. Cerdà-Navarro, C. Touza, M. Morey-López, E. Curiel, Academic integrity variables and measurement error, J. Market. Res. 18 (1) (1981) 39–50.
policies against assessment fraud in postgraduate studies: an analysis of the [98] R.B. Kline, Principles and Practice of Structural Equation Modeling, fourth ed.,
situation in Spanish universities, Heliyon 8 (3) (2022), e09170, https://doi.org/ The Guilford Press, New York, NY, 2015.
10.1016/j.heliyon.2022.e09170. [99] J.F. Hair, G.T.M. Hult, C. Ringle, M. Sarstedt, A Primer on Partial Least Squares
[82] R. Bougie, U. Sekaran, Research Methods for Business: A Skill Building Approach, Structural Equation Modeling (PLS-SEM), Sage Publications, London, 2016.
John Wiley & Sons, 2019. [100] S.A. Bin-Nashwan, H. Abdul-Jabbar, S.A. Aziz, Does trust in zakat institution
[83] K.B. Wright, Researching internet-based populations: advantages and enhance entrepreneurs’ zakat compliance? Journal of Islamic Accounting and
disadvantages of online survey research, online questionnaire authoring software Business Research 12 (5) (2021) 768–790.
packages, and web survey services, J. Comput. Commun. 10 (2005). [101] M.R. King, ChatGPT, A conversation on artificial intelligence, chatbots, and
plagiarism in higher education, Cell. Mol. Bioeng. 16 (1) (2023) 1–2.

11