Received: 3 July 2023 | Accepted: 17 December 2023

DOI: 10.1111/bjet.13425

ORIGINAL ARTICLE

Investigation of the moderation effect of

gender and study level on the acceptance
and use of generative AI by higher education
students: Comparative evidence from Poland
and Egypt

Artur Strzelecki1 | Sara ElArabawy2

1
Department of Informatics, University of
Economics in Katowice, Katowice, Poland Abstract
2
Business Department, The British This study delves into the implications of incorporat-
University in Egypt, Cairo, Egypt ing AI tools, specifically ChatGPT, in higher educa-
Correspondence tion contexts. With a primary focus on understanding
Artur Strzelecki, Department of Informatics, the acceptance and utilization of ChatGPT among
University of Economics in Katowice, 1 university students, the research utilizes the Unified
Maja 50, 40-­287 Katowice, Poland.
Email: [email protected] Theory of Acceptance and Use of Technology
(UTAUT) as the guiding framework. The investiga-
tion probes into four crucial constructs of UTAUT—
performance expectancy, effort expectancy, social
influence and facilitating conditions—to understand
their impact on the intent and actual use behaviour of
students. The study relies on data collected from six
universities in two countries and assessed through
descriptive statistics and structural equation model-
ling techniques, and also takes into account partici-
pants' gender and study level. The key findings show
that performance expectancy, effort expectancy, and
social influence significantly influence behavioural
intention. Furthermore, behavioural intention, when
considered alongside facilitating conditions, influ-
ences actual use behaviour. This research also ex-
plores the moderating impact of gender and study
level on the relationships among these variables.
The results not only augment our comprehension of
technology acceptance in the context of AI tools but
also provide valuable input for formulating strategies
that promote effective incorporation of ChatGPT in
higher education. The study underscores the need

© 2024 British Educational Research Association.

Br J Educ Technol. 2024;55:1209–1230.  wileyonlinelibrary.com/journal/bjet | 1209

1210 |    STRZELECKI and ElARABAWY

for effective awareness initiatives, bespoke training

programmes, and intuitive tool designs to bolster stu-
dents' perceptions and foster the wider adoption of AI
tools in education.

KEYWORDS
ChatGPT, higher education, moderating effect, technology
acceptance

Practitioner notes
What is already known about this topic
• ChatGPT is a tool that is quickly gaining worldwide recognition.
• ChatGPT helps with writing essays and solving assignments.
• ChatGPT raises ethical concerns about authorship, plagiarism and ethics.
What this paper adds
• This study explores students' acceptance of ChatGPT as an aid in their education,
which has not been studied previously.
• We used the extended Unified Technology Acceptance and Use of Technology
theory to test what factors mostly influence the use of ChatGPT by students.
• We conducted a multiple study in Poland and Egypt based on sampling strategy
from six universities.
Implications for practice and/or policy
• ChatGPT is a global game changer and should be incorporated into study
programmes.
• The limitations of ChatGPT should be well explained and known since it is prone
to making mistakes.
• Higher education teachers should be aware of ChatGPT's capabilities.

I NTRO DUCTI O N

Generative AI is a subfield of artificial intelligence (AI) and machine learning (ML) that is
capable of creating new and original content, in varied formats such as text, audio, video,
pictures and code (Lv, 2023), based on deep learning algorithms that learn to recognize pat-
terns and relationships from vast amounts of input data, which then generate new outputs
that are similar in style and structure to the data they were trained on. This subfield has been
developing over several decades and is rapidly evolving, due to advances and availability
in computational power, large data sets and significant improvements in machine learning
algorithms that can be used to create new content (Feuerriegel et al., 2023). The ability of
these models to self-­formulate new and varied outputs represents a paradigm shift in the
field of AI because they are not being explicitly programmed to follow pre-­determined rules,
or generate specific outputs, like other AI (Feuerriegel et al., 2023).
ChatGPT, created by OpenAI, is an advanced language model based on a generative
pre-­trained transformer. It utilizes natural language processing techniques to learn from
enormous data volumes and produce human-­like responses to questions. With a data set of
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1211

570gb representing 300 billion words and 175 billion parameters, ChatGPT can be thought
of as a computerized conversational partner that can answer questions, provide analysis
and even offer opinions (OpenAI, 2023). This version of ChatGPT, GPT-­3.5, was utilized at
the time the study was conducted (OpenAI, 2023).
Teaching and learning could be revolutionized by ChatGPT in higher education by serv-
ing as an AI-­powered tool for various tasks (Lim et al., 2023). ChatGPT has the flexibility to
function as an independent tool or seamlessly integrate into various systems and platforms
utilized within higher education institutions (HEIs). The use of ChatGPT can facilitate stu-
dents' learning experiences, generate alternative ways of expressing ideas, and based on
data supplied by students or teachers, immediately give each student personalized feed-
back (Hwang & Chen, 2023). Moreover, ChatGPT can be used as a collaboration coach
to assist groups in researching and solving problems together, as a guide on the side to
navigate physical and conceptual spaces, and as a codesigner to help in the designing or
updating of curricula. Additionally, ChatGPT can be employed as an exploratorium to pro-
vide tools for exploring and interpreting data, as a study buddy to help students reflect on
learning material and as a motivator that offers games and challenges to extend learning.
It can act also as a dynamic assessor for students' assignments and other evaluation tasks
(Ivanov & Soliman, 2023).
The utilization of ChatGPT in the realm of higher education presents a multitude of chal-
lenges and ethical considerations (Lund et al., 2023). A primary concern expressed by nu-
merous universities and educators revolves around the potential escalation of plagiarism
and cheating among students. Furthermore, there exist apprehensions regarding the ef-
ficacy of current plagiarism detection tools when faced with written content generated by
ChatGPT (Cotton et al., 2023; Perkins, 2023). Moreover, the absence of regulation sur-
rounding ChatGPT is also a concern, as it facilitates rapid development without adequate
exploration of potential risks and shared protocols. Additionally, the tool's inability to dis-
cern between veracity and falsehood, right and wrong, raises concerns pertaining to cog-
nitive bias (Kasneci et al., 2023; Lund & Wang, 2023). Other concerns encompass privacy,
accessibility and commercialization, necessitating meticulous deliberation and regulation
to ensure fairness and equity in the application of AI tools in higher education (Rudolph
et al., 2023). The unique nature of AI calls for dedicated research. Recent study found that
providing teachers with AI knowledge and practical experience can reduce their concerns
and improve their trust in AI tools. Sessions on AI-­powered assessment and the use of an
AI tool positively impacted teachers' knowledge, perceptions, attitudes, trust and willingness
to adopt AI tools (Nazaretsky, Ariely, et al., 2022).
At present, there is a dearth of peer-­reviewed studies specifically concentrating on the
usage of ChatGPT by students in third-­level education. This research gap is acknowledged,
given that ChatGPT was only made available on 30 November 2022, and is still in the re-
search preview stage, allowing users to offer feedback on its functionality (OpenAI, 2023).
Nevertheless, it has aroused substantial interest across diverse stakeholder groups, includ-
ing higher education students who employ this AI-­powered tool to assist them in completing
their educational assignments (Crawford et al., 2023; Strzelecki, 2023).
The primary aim of this investigation was to analyse the acceptance and utilization of
ChatGPT across students in higher education. To attain this goal, the research evaluated
four concepts adopted from prior literature that can potentially impact the acceptance and
utilization of ChatGPT by students in the context of university education. The study inves-
tigated the ‘Unified Theory of Acceptance and Use of Technology’ (UTAUT) (Venkatesh
et al., 2003) towards ChatGPT. The four concepts affecting the acceptance and usage of
technology, according to this theory, are: ‘performance expectancy’, ‘effort expectancy’, ‘so-
cial influence’ and ‘facilitating conditions’. Although some might argue that current literature
contains a plethora of studies utilizing UTAUT, suggesting that everything has already been
1212 |    STRZELECKI and ElARABAWY

covered in this area, we contend that the sudden and swift increase in ChatGPT users and
the tool's nearly global accessibility necessitates a re-­examination of this background. This
would enable us to contribute to the theory by examining how tertiary education students
perceive ChatGPT.
The research rationale was to study the higher education students' expectations towards
using the AI-­powered tool ChatGPT. Data for this study were collected from higher educa-
tion students from six universities in Poland and Egypt. The theoretical foundation of the
study is the use of the UTAUT framework and the partial least squares method of structural
equation modelling. The theoretical model was tested in both samples (one for Polish data
and one for Egypt data) and revealed satisfying results for tested hypotheses and moderat-
ing effects.
The study is organized as follows. The introduction section presents background informa-
tion on ChatGPT and briefly describes its potential applications in higher education. This is
followed by current literature review section, which covers the limited field of ChatGPT use
in academia. The section also introduces the UTAUT theory and its relevance to the use
of ChatGPT by students. The method section outlines the process for creating the theoret-
ical model, followed by the results section which presents the estimation of the theoretical
model. Finally, the results are discussed, and the implications, limitations and potential ave-
nues for next research are presented.

TH EO R ETI CA L BACKG ROUN D

There has been an increase in interest in the field of technology acceptance and usage,
as individuals increasingly rely on technology for their daily activities. An influential theory
utilized to elucidate and predict technology acceptance is UTAUT, developed by Venkatesh
et al. (2003). UTAUT was constructed by integrating and synthesizing eight pre-­existing mod-
els of information technology acceptance. These models include the ‘Innovation Diffusion
Theory (IDT)’ (Rogers, 1962), the ‘Theory of Reasoned Action (TRA)’ (Fishbein & Ajzen, 1975),
the ‘Social Cognitive Theory (SCT)’ (Bandura, 1986), the ‘Technology Acceptance Model
(TAM)’ (Davis, 1986), the ‘Theory of Planned Behavior (TPB)’ (Ajzen, 1991), the ‘Model
of PC Utilization (MPCU)’ (Thompson et al., 1991), the ‘Motivational Model (MM)’ (Davis
et al., 1992) and a ‘model Combining the TAM and the TPB’ (Taylor & Todd, 1995). According
to Yu et al. (2021), UTAUT is a theory that combines various concepts and user experiences
to form a basis for understanding the acceptance process of an information system. This
comprehensive model encompasses constructs such as ‘Performance expectancy’, ‘Effort
expectancy’, ‘Social influence’ and ‘Facilitating conditions’, which significantly impact indi-
viduals' ‘Behavioural intention’ to adopt and use a technology. Furthermore, UTAUT consid-
ers individual differences, such as gender, age and experience, as moderating factors that
influence the aforementioned constructs within the model.
‘Performance expectancy’ pertains to the extent to which a person perceives that using a
technology will enhance their work performance. ‘Effort expectancy’ characterizes user per-
ception of the ease of use associated with utilizing a technology. ‘Social influence’ pertains
to the degree to which an individual believes that others expect them to use a technology.
‘Facilitating conditions’ refer to the extent to which a person believes that the resources
and support required to use a technology efficiently are readily available. ‘Use behaviour’
describes the actual utilization of technology by the user and ‘Behavioural intention’ is the
plan to behave in a certain way.
The UTAUT theory has been found to be a reliable measure of higher education students'
adoption of technology. Various scholars have tested this theory to assess the intention to
use smartphones in the study process (Hoi, 2020; Nikolopoulou et al., 2020), the Google
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1213

Classroom platform (Jakkaew & Hemrungrote, 2017; Kumar & Bervell, 2019), blended
learning (Azizi et al., 2020), e-­learning system (El-­Masri & Tarhini, 2017; Samsudeen &
Mohamed, 2019), learning management systems (Raman & Don, 2013; Zwain, 2019) and
collaborative Web 2.0 applications (Huang et al., 2013) for learning in higher education.
The original UTAUT theory incorporates four moderating elements. However, UTAUT2
version removed one of them (voluntariness of use) and only kept three moderating el-
ements: gender, age and experience. Dwivedi et al. (2019) noted that previous research
using UTAUT/UTAUT2 models typically utilized only a portion of the model and frequently
neglected moderators. Following this recommendation, we did not include the moderating
variable of experience since the use of ChatGPT does not have a long history of availabil-
ity. Instead, we replaced the moderating variable of age with ‘Study level’, and we used the
original moderating variable of ‘Gender’.
Based on the analysis of prior studies that utilized the UTAUT theory, we suggest em-
ploying the same hypotheses as those used in the original UTAUT model. Performance
expectancy ‘is the degree to which a person believes that using the system will enable them
to improve their level of job performance’ (Venkatesh et al., 2003). In the context of using
ChatGPT by students in their study process, it refers to the students' perception that utilizing
the system will result in advantages for their learning practices. Specifically, in this particular
context, it indicates the students' belief in the usefulness of ChatGPT for their academic
tasks. We propose following hypothesis:

H1. The influence of performance expectancy on behavioural intention will be

direct and positive and moderated by gender and a study level.

Effort expectancy ‘is characterized as the level of comfort associated with using the sys-
tem’. (Venkatesh et al., 2003). In the study of ChatGPT, ‘Effort expectancy’ pertains to the
ease of use of the system and the extent to which it minimizes distractions and requires
minimal effort from students. We propose following hypothesis:

H2. The influence of effort expectancy on behavioural intention will be direct

and positive and moderated by gender and a study level.

Social influence ‘is characterized as the extent to which a person believes that significant
others should implement the new system’ (Venkatesh et al., 2003). In the present investiga-
tion, ‘Social influence’ is used to describe the extent to which individuals perceive that their
immediate social circle believes that they should integrate ChatGPT into their higher educa-
tion practices. We propose following hypothesis:

H3. The impact of social influence on behavioural intention will be direct and
positive and moderated by gender and a study level.

Facilitating conditions ‘are the extent to which a person believes that an administrative and
technological framework is in place to support use of the system’ (Venkatesh et al., 2003).
Within the study aimed at examining the acceptance and adoption of ChatGPT among col-
lege students, ‘Facilitating conditions’ could be especially significant as it has the potential
to impact the level of ease with which students are able to utilize the AI chat, including its ac-
cessibility. In the original UTAUT model, facilitating conditions is not moderated by gender.
We keep this setting and propose following hypothesis:

H4. The influence of facilitating conditions on use behaviour will be direct and
positive and moderated by a study level.
1214 |    STRZELECKI and ElARABAWY

Similar to the intention models discussed by Venkatesh et al. (2003), we anticipate that
the technology use behaviour will be positively impacted by behavioural intention, as per the
established theory.

H5. Behavioural intention will have a significant positive influence on use

behaviour.

The model's configuration, as per the proposed hypotheses, is illustrated in Figure 1,

which comprises all external and dependent variables from the UTAUT model, along with
two moderating variables.

M ETHO DO LOGY

The study employed an online survey to collect data from students at three universities lo-
cated in Katowice, Poland, namely University of Economics in Katowice, University of Silesia
in Katowice and Silesian University of Technology. We have also collected data making
multiple study from three universities located in Cairo, Egypt, namely The British University
in Egypt, The American University in Cairo and Arab Academy for Science, Technology
& Maritime Transport. Participants were recruited from the entire universities' database of
active students, and there were no prerequisites for participation. The measurement scale
for each variable was developed based on a study by Venkatesh et al. (2012). To measure
‘Performance expectancy’, four items were adapted, while four items were adapted for ‘Effort
expectancy’, three items for ‘Social influence’, four items for ‘Facilitating conditions’ and three
items for ‘Behavioural intention’. Venkatesh et al. (2012) examined the acceptance and use
of ‘mobile internet’, while this study's scale measures the acceptance and use of ‘ChatGPT’.
The items for measuring ‘Use behaviour’ are not known in Venkatesh et al. (2012), but only
the two bordering values as ‘never’ and ‘many times per day’ are available.
In terms of the measurement scale, a seven-­point Likert scale was utilized in Poland,
while a five-­point Likert scale was presented for respondents in Egypt. The difference in

Model proposition

Performance H1
expectancy

H2 H5
Effort expectancy Behavioral intention Use behavior

H3

Social influence

H4

Facilitating conditions

Gender Study level

FIGURE 1 Model proposition.

ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1215

the higher bond of the scale comes from cultural differences, while in Poland, seven-­point
scale is commonly used, in Egypt the scale is five-­point. Each item on the scale ranged
from ‘strongly agree’, which was assigned a value of 7 in Poland or of 5 in Egypt, to ‘strongly
disagree’, which was assigned a value of 1. However, the item ‘Use behaviour’ was mea-
sured with a seven-­option scale that included the following response options: ‘several times
a day’, ‘once a day’, ‘several times a week’, ‘once a week’, ‘several times a month’, ‘once a
month’ and ‘never’. Moreover, the variable ‘Study level’ was assessed using three options.
For Poland, it includes a three-­year duration for the first study cycle, a two-­year duration for
the second study cycle leading to a master's degree and the option of being a PhD student.
This classification aligns with the principles of the Bologna Process, which divides academic
studies into three cycles: the first, second and third cycles. In Egypt, the higher education
system is influenced by the French model, but it has its own unique structure. In Egypt,
the distribution of study level was adapted to undergraduates (university student) and pre-­
master's student (modules) as first level; Master's student (thesis level) as second level; and
pre-­PHD Student (modules), pre-­PHD Student (thesis level) and PHD holders as third level.
The ‘Gender’ question provided respondents with three response options: female, male and
prefer not to disclose. The third option was only available for Polish respondents. Islam, the
predominant religion in Egypt, traditionally recognizes two genders. Therefore, our survey
reflects this binary understanding.

Data collection

In March 2023, an online questionnaire was prepared, and a pilot study was carried out to
evaluate the validity and reliability of the scales employed. In the pilot study, a total of 36
students were surveyed and requested to provide feedback on the comprehensibility of the
scales. The findings indicated that all variables met the pre-­determined criteria for both va-
lidity and reliability, affirming the consistent reliability of the developed scale in measuring
students' levels of acceptance towards ChatGPT in higher education. This conclusion was
supported by composite reliability, Cronbach's alpha and reliability coefficients exceeding
0.7, as well as an average variance extracted (AVE) surpassing 0.5. Subsequent to the pilot
study, the survey was administered in the end of May to students at the universities located
in Katowice, Poland; and Cairo, Egypt. Each invitation clearly outlined the purpose of the
study, emphasized voluntary participation and assured complete anonymity of the collected
data. The option to leave the study at any time was disclosed to the participants. The survey
remained accessible for a duration of 1 month, concluding at the end of June 2023.
The measurement scale utilized in this study is shown in Table 1, providing the corre-
sponding items and the scale source. The scale includes a total of 19 items that were utilized
to measure the acceptance and usage of ChatGPT by students. Furthermore, the survey
included two supplementary questions to inquire about gender and study level. As for study
level, the options were consistent with the previously described cycle division.

Sample characteristics

In this study, the questionnaire was completed by a total of 543 participants in Poland, con-
sisting of 288 males, 232 females and 23 students who chose not to disclose their gender.
The sample population can be further described in terms of study cycle, with 406 partici-
pants from the first cycle, among whom 30 were in their first year of study, 171 were in their
second year of study, and 205 were in their third year of study. Among the second cycle
participants, there were 128 students, with 57 being in their first year of study and 71 in their
1216 |    STRZELECKI and ElARABAWY

TA B L E 1 Measurement scale.

Construct Code Question Source

Performance expectancy PE1 ‘I believe that ChatGPT is useful in my studies’ Venkatesh
PE2 ‘Using ChatGPT increases your chances of et al. (2012)
achieving important things in your studies’
PE3 ‘Using ChatGPT helps you get tasks and
projects done faster in your studies’
PE4 ‘Using ChatGPT increases your productivity in
your studies’
Effort expectancy EE1 ‘Learning how to use ChatGPT is easy for me’
EE2 ‘My interaction with ChatGPT is clear and
understandable’
EE3 ‘I find ChatGPT easy to use’
EE4 ‘It is easy for me to become skilful at using
ChatGPT’
Social influence SI1 ‘People who are important to me think I should
ChatGPT’
SI2 ‘People who influence my behavior believe that I
should use ChatGPT’
SI3 ‘People whose opinions I value prefer me to use
ChatGPT’
Facilitating conditions FC1 ‘I have the resources necessary to use
ChatGPT’
FC2 ‘I have the knowledge necessary to use
ChatGPT’
FC3 ‘ChatGPT is compatible with technologies I use’
FC4 ‘I can get help from others when I have
difficulties using ChatGPT’
Behavioral intention BI1 ‘I intend to continue using ChatGPT in the future’
BI2 ‘I will always try to use ChatGPT in my studies’
BI3 ‘I plan to continue to use ChatGPT frequently’
Use behavior UB1 ‘Please choose your usage frequency for Venkatesh
ChatGPT: Never; Once a month; Several et al. (2012)
times a month; Once a week; Several times a
week; Once a day; Several times a day’

second year of study. The third cycle of study was represented by nine participants who
took part in the survey. In Egypt, a total of 385 participants, consisting of 197 males and 188
females, took part in the study. The sample consisted of 289 undergraduate students (first
level), 44 master students (second level), 29 PhD students and 23 PhD holders (third level).

R ESULTS

Model estimation

To estimate the model, we utilized the partial least squares of structural equation modelling
(PLS-­SEM) algorithm with the path weighting scheme in SmartPLS 4 software (Version
4.0.9.2), using default initial weights and a maximum of 3000 iterations (Ringle et al., 2022).
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1217

TA B L E 2 Item loadings.

Poland Egypt

Item Loading Item Loading

BI1 0.909 BI1 0.825
BI2 0.826 BI2 0.862
BI3 0.940 BI3 0.857
EE1 0.872 EE1 0.809
EE2 0.894 EE2 0.813
EE3 0.897 EE3 0.778
EE4 0.913 EE4 0.702
FC1 0.842 FC1 0.919
FC2 0.896 FC2 0.879
FC3 0.817 FC3 0.705
FC4 0.571 FC4 0.598
PE1 0.905 PE1 0.848
PE2 0.869 PE2 0.846
PE3 0.893 PE3 0.762
PE4 0.868 PE4 0.746
SI1 0.939 SI1 0.807
SI2 0.943 SI2 0.861
SI3 0.935 SI3 0.847
UB 1.000 UB 1.000

TA B L E 3 Construct reliability and validity.

Composite Composite Average

Cronbach's reliability reliability variance
alpha (rho_a) (rho_c) extracted (AVE)

Poland Egypt Poland Egypt Poland Egypt Poland Egypt

Behavioural intention 0.872 0.805 0.887 0.806 0.922 0.885 0.797 0.719
Effort expectancy 0.917 0.784 0.929 0.785 0.941 0.859 0.799 0.604
Facilitating conditions 0.824 0.760 0.831 0.896 0.895 0.849 0.741 0.659
Performance 0.907 0.784 0.909 0.804 0.935 0.860 0.781 0.607
expectancy
Social influence 0.933 0.789 0.935 0.790 0.957 0.877 0.882 0.704

Additionally, we employed bootstrapping, a nonparametric procedure, with 5000 samples to

determine the statistical significance of the PLS-­SEM results. We assessed the reflectively
specified constructs by analysing the indicator loadings, where an indicator loading greater
than 0.7 indicates that the construct explains more than 50% of the variance in the indicator,
indicating an acceptable level of item reliability. We present the loadings in Table 2, which
are all above the lower bound, except for FC4. Therefore, we removed FC4 from further
processing in the model and did not consider it.
Composite reliability is a measure used to evaluate the reliability of a model, with ac-
ceptable to good reliability levels demonstrated by results ranging from 0.70 to 0.95 (Hair
 1218
|
  

TA B L E 4 (a) HTMT values (sample in Poland). (b) HTMT values (sample in Egypt).

(a) HTMT values (sample in Poland)

Facilitating Performance Social Use

Behavioural intention Effort expectancy conditions expectancy influence behaviour
Behavioural intention
Effort expectancy 0.636
Facilitating conditions 0.633 0.801
Performance expectancy 0.841 0.596 0.595
Social influence 0.619 0.427 0.494 0.584
Use behaviour 0.751 0.530 0.559 0.577 0.454

(b) HTMT values (sample in Egypt)

Facilitating Performance Social Use

Behavioural intention Effort expectancy conditions expectancy influence behaviour
Behavioural intention
Effort expectancy 0.425
Facilitating conditions 0.329 0.737
Performance expectancy 0.588 0.708 0.581
Social influence 0.549 0.361 0.350 0.531
Use behaviour 0.449 0.399 0.298 0.427 0.166
STRZELECKI and ElARABAWY
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1219

Model estimation results – sample in Poland

Performance H1 β=.504
expectancy

H2 β=.230 β=.002 R2=.646 H5 β=.610 R2=.528

Effort expectancy Behavioral intention Use behavior

β=-.029
β=-.107 β=.044
H3 β=.192

Social influence β=.093

β=-.018
β=.081
H4 β=.171

Facilitating conditions

Gender Study year

Model estimation results – sample in Egypt

Performance H1 β=.285
expectancy

H2 β=.013 β=-.030 R2=.378 H5 β=.338 R2=.205

Effort expectancy Behavioral intention Use behavior

β=.231
β=.120 β=-.025
H3 β=.398

Social influence β=-.031

β=-.285
β=-.223
H4 β=.206

Facilitating conditions

Gender Study level

FIGURE 2 (a) Model estimation results—sample in Poland. (b) Model estimation results—sample in Egypt.

et al., 2022). Cronbach's alpha is one more measure of internal consistency reliability, with
similar thresholds as composite reliability (ρc). Additionally, a reliability coefficient ρA, de-
veloped by Dijkstra (2010, 2014) and later refined by Dijkstra and Henseler (2015), provides
an exact and consistent alternative. Convergent validity of the measurement models is as-
sessed by calculating the AVE for all the items associated with a specific reflective variable,
with an AVE threshold of 0.50 or higher considered acceptable (Sarstedt et al., 2022). The
quality criteria presented in Table 3 were met by the composite reliability, Cronbach's alpha,
reliability coefficient and AVE.
To assess the discriminant validity of PLS-­SEM, the heterotrait-­monotrait (HTMT) ratio
of correlations method by Henseler et al. (2015) was utilized. The HTMT threshold of 0.90
is recommended, indicating a potential issue with discriminant validity when constructs are
conceptually similar. For more distinct constructs, a lower threshold of 0.85 is suggested
 1220
|
  

TA B L E 5 Path coefficient results.

Sample in Poland Sample in Egypt

Path
Path coefficient f 2 effect size p values Confirmed Path coefficient f 2 effect size p values Confirmed
PE → BI 0.504 0.397 0.000 Yes 0.285 0.035 0.000 Yes
EE → BI 0.230 0.099 0.000 Yes 0.013 0.000 0.734 No
SI → BI 0.192 0.072 0.000 Yes 0.398 0.099 0.000 Yes
FC → UB 0.171 0.043 0.000 Yes 0.206 0.049 0.000 Yes
BI → UB 0.610 0.554 0.000 Yes 0.338 0.127 0.000 Yes
STRZELECKI and ElARABAWY
 STUDENTS

TA B L E 6 Moderating effects of study level and gender.

Sample in Poland Sample in Egypt

Path Path
Moderating effect coefficient f 2 effect size p values Confirmed coefficient f 2 effect size p values Confirmed
Study level × SI → BI 0.081 0.012 0.030 Yes −0.223 0.045 0.007 Yes
Study level × PE → BI 0.044 0.003 0.301 No −0.025 0.001 0.662 No
Study level × EE → BI −0.107 0.023 0.000 Yes 0.120 0.018 0.084 No
Gender × EE → BI −0.029 0.002 0.402 No 0.231 0.013 0.033 Yes
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION

Gender × SI → BI −0.018 0.001 0.620 No −0.285 0.025 0.003 Yes

Gender × PE → BI 0.002 0.000 0.962 No −0.030 0.000 0.799 No
Study level × FC → UB 0.093 0.018 0.002 Yes −0.031 0.001 0.451 No
  
|
1221
1222 |    STRZELECKI and ElARABAWY

(Henseler et al., 2015). In Table 4, all values are below the 0.85 threshold, indicating no
significant discriminant validity concerns.
In the subsequent stage of analysis, the explanatory capability of the model is assessed
through the coefficient of determination (R 2), which measures the amount of variance ac-
counted for in each construct. The scale of R 2 ranges from 0 to 1, with higher values indi-
cating greater explanatory power. To provide a general guideline, Hair et al. (2011) suggest
that R 2 values of 0.25, 0.50 and 0.75 can be considered weak, moderate and substantial
respectively. The effect size of a variable is evaluated using f2 values of 0.35, 0.15 and 0.02,
indicating large, medium and small effects respectively. An effect size below 0.02 indicates
no impact (Sarstedt et al., 2022).
PLS-­SEM findings are illustrated in Figure 2 and Tables 5 and 6, with standardized re-
gression coefficients (β) shown on the path relationships and R 2 values presented in the
variables' squares. The primary observation for sample in Poland reveals that ‘Performance
expectancy’ has the most prominent impact (0.502) on ‘Behavioural intention’, followed by
‘Effort expectancy’ (0.232) and ‘Social influence’ (0.190), explaining 64.6% of the ‘Behavioural
intention’ variance (as indicated by the R 2 value). All three paths also have f2 size effect
above lower bond indicating that each path makes an effect. Conversely, ‘Behavioural in-
tention’ has the most significant effect (0.608) on ‘Use behaviour’, followed by ‘Facilitating
conditions’ (0.173). These two variables also have effect size above lower bond and account
for 52.8% of the ‘Use behaviour’ variance.
The primary observation for sample in Egypt reveals that ‘Social influence’ has the most
prominent impact (0.433) on ‘Behavioural intention’, followed by ‘Performance expectancy’
(0.280), explaining 37.8% of the ‘Behavioural intention’ variance (as indicated by the R 2
value). The effect of ‘Effort expectancy’ was not confirmed in this sample. The two signif-
icant paths also have f2 size effect above lower bond indicating that each path makes an
effect. Conversely, ‘Behavioural intention’ has the most significant effect (0.345) on ‘Use
behaviour’, followed by ‘Facilitating conditions’ (0.206). These two variables also have ef-
fect size above lower bond and account for 20.5% of the ‘Use behaviour’ variance. Table 5
presents results of the significance tests for the structural model's path coefficients and
hypotheses confirmation.
In case of moderating variables of ‘Study level’ and ‘Gender’ results show different results
for both samples. In sample consisting Polish students only, three moderating effects are
statistically significant in the model. Study level significantly moderates the path between
‘Social influence’ and ‘Behavioural intention’, the path between ‘Effort expectancy’ and
‘Behavioural intention’ and the path between ‘Facilitating conditions’ and ‘Use behaviour’.
Other moderating effects are not significant.
In sample consisting Egyptian students, also only three moderating effects are statisti-
cally significant in the model. However, we can notice that the effects are higher, because
the path coefficients have higher value. Study level significantly moderates the path be-
tween ‘Social influence’ and ‘Behavioural intention’, but gender significantly moderates the
path between ‘Effort expectancy’ and ‘Behavioural intention’ and the path between ‘Social
influence’ and ‘Behavioural intention’. Moderating effects of ‘Study level’ and ‘Gender’ are
presented in Table 6.

D I SCUSS I O N

This research utilizes an adapted version of the UTAUT theory to investigate factors affect-
ing the adoption and the use of AI-­powered tool, ChatGPT. The findings shed light on the
impact such tools have on the learning process.
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1223

Our findings show that ‘Performance Expectancy’ is one of the key determinants of
‘Behavioural Intention’, aligning with earlier studies regarding learning management sys-
tems (Al-­Adwan et al., 2022), social networking tools for learning (Al-­Adwan et al., 2022) and
mobile learning systems (Almaiah et al., 2019). Hypothesis H1 has been confirmed positively
for both samples. The study underscores the benefits of AI chat, such as reducing task
completion time and providing immediate responses to queries, which can bolster academic
performance and in turn, foster an intention to utilize such tools.
Moreover, ‘Effort Expectancy’ emerged as a crucial influencer of ‘Behavioural Intention’
confirming hypothesis H2 in the Polish sample. This aligns with studies regarding learning
management systems (Al-­Mamary, 2022) and humanoid robot assistance in academic writ-
ing (Guggemos et al., 2020), but contradicts findings related to massive open online courses
(Altalhi, 2021), where ‘Effort expectancy’ was found to be not significant. Similar results were
achieved in the Egyptian sample, not confirming hypothesis H2 in this sample. The study
indicates that students who find ChatGPT easy to use and less effort-­intensive, providing
multi-­lingual conversational communication and allowing response refinement, are more
likely to use it.
The research revealed that ‘Social Influence’ significantly influences ‘Behavioural
Intention’ in agreement with studies on mobile learning adoption (Alowayr, 2022) and e-­
learning system adoption, but diverging from research on interactive whiteboards (Wong
et al., 2015) and learning management systems (Zwain, 2019). The results suggest that
acceptance and usage of ChatGPT are influenced by external figures, such as instructors,
peers and administrators, underlining their vital role in promoting and encouraging students'
adoption and use of AI chat systems. Hypothesis H3 has been confirmed positively for both
samples, but the hypothesis is defined in a positive way. The current discussion in higher
education in this regard tends to be more negative. There are voices that students may be-
lieve that using the ChatGPT is not a good learning practice, something that one should not
be proud of (Cotton et al., 2023).
Our study unveils that ‘Facilitating conditions’, as outlined in the original UTAUT model,
predict the ‘Use behaviour’ of ChatGPT, thus hypothesis H4 has been confirmed for both
samples. This finding resonates with earlier research on mobile learning systems (Almaiah
et al., 2019) and learning management systems (Al-­Adwan et al., 2022). However, it con-
trasts with studies on the adoption of social networking tools (Alvi, 2021) and massive open
online courses (Altalhi, 2021). This suggests that students deem the open access and online
presence of AI chat as pivotal in enhancing their actual usage. Factors facilitating this usage
include availability across web browsers on all platforms, worldwide accessibility, and user-­
friendly interaction in prevalent languages. Hypothesis H5 was also confirmed positively in
both samples, showing that ‘Behavioural intention’ significantly influences ‘Use behaviour’.
Our study also scrutinizes the potential moderating effects of ‘Gender’ and ‘Study level’;
however, our findings differ across both samples. Findings in the Polish sample reveal that
gender does not significantly influence the model relationships. This is consistent with prior
research, such as Dečman (2015), showing that both male and female students demon-
strate similar motivation levels towards e-­learning. As such, our study does not support
the hypothesis that ‘Gender’ moderates the model relationships. Furthermore, prior studies
have either divided sample groups based on gender to examine its moderating effect or dis-
missed this effect through multigroup analysis, as seen in Alghamdi et al. (2022). However,
in the Egyptian sample, two out of three moderating effects are significant. Hence, we con-
clude that ChatGPT usage by gender needs to be further examine.
Since there was a clear significant and evident difference between the magnitude of gen-
der in Egypt which proved to be much higher and influential than that in Poland, this is a clear
point for further exploration. That exploration would aim to investigate further the potential
causes for such difference of significance that is so considerable between both population
1224 |    STRZELECKI and ElARABAWY

various explanation could enrich our academic understanding of the impact of gender as a
moderator in such important relationships in terms either of religious factors, culture factors
and social policy and regulatory rules implications.
Critical analytical frameworks addressing gender dynamics and relations in higher ed-
ucation within Polish and Egyptian populations warrant emphasis. The application context
of gender norms, contributions and prevailing societal customs plays pivotal roles in deter-
mining how gender functions as a moderating factor in higher education and comparable
settings (Cislaghi & Heise, 2020). In this regard, Poland might be perceived as occupying an
intermediary position within the spectrum of Western and European gender perspectives. It
is often seen as less liberal compared to Western Europe and North America but more lib-
eral when juxtaposed against regions like Russia, Eastern European countries, and certain
South American and Caribbean nations.
Concurrently, within the more traditional paradigms of the southern Mediterranean, Egypt
seems to hold a similar intermediate stance. The Egyptian perspective on gender appears
to be more liberal than that of the Persian Gulf and the majority of African nations. However,
it leans more conservative when compared to North-­western African nations, specifically
Tunisia and Morocco (Ayadi & Forouheshfar, 2023). This study identifies a parallel between
Egypt and Poland regarding gender dynamics. Both nations, whether within their liberal or
conservative contexts, represent a median perspective. This central position could poten-
tially resonate with the more liberal or conservative interpretations of gender within their
respective regional contexts.
Regarding ‘Study level’, the moderating effect was significant in three out of four in-
stances in Polish sample and in one out four in Egyptian sample. It was positively significant
in first sample for the relationship between ‘Social influence’ and ‘Behavioural intention’, and
between ‘Facilitating conditions’ and ‘Use behaviour’. However, the relationship between
‘Effort expectancy’ and ‘Behavioural intention’ showed a negative moderating effect. This
suggests a variance in the acceptance and usage of ChatGPT among students from differ-
ent academic levels. In second sample, positive significant relationship was noticed for the
relationship between ‘Social influence’ and ‘Behavioural intention’. Further research, delving
into the influence of study level on ChatGPT usage by employing multigroup analysis, would
be beneficial but is outside the purview of our current study.

Implications

This study carries significant implications for both theory and practical application. It en-
hances our understanding of key factors influencing the acceptance and integration of AI
chat-­tools, like ChatGPT, within higher education contexts. The insights gained can be used
by policymakers, educators and researchers in higher education to work in tandem with stu-
dents in incorporating AI tools into their study processes, thereby facilitating the successful
adoption of such technologies and fostering productive collaboration among stakeholders.
One crucial finding is the significant role ‘Performance Expectancy’ plays in shaping
students' willingness to adopt and use AI chat-­tools. This necessitates efforts to improve
students' understanding of the benefits and potentialities of such tools. A solution could lie
in deploying effective awareness campaigns, supplemented by training programmes and
workshops focusing on the AI chat-­tool advantages. Awareness could be amplified via var-
ious channels including social media, university websites and open seminars, stimulating
students to engage in insightful discussions.
Moreover, the study highlights the importance of ‘Effort expectancy’, reinforcing the need
for user-­friendly and easily navigable AI chat-­systems. This can incentivize students to per-
sist in using these tools. In response, AI chat-­system developers should focus on designing
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1225

intuitive user interfaces and incorporating accessible features. Such initiatives are aimed at
enticing a broader student demographic to interact with these platforms. It is crucial to re-
member that if students find AI chat-­systems too complex for their learning needs, they may
discontinue use, even if other conditions are favourable.
‘Social influence’ plays a pivotal role in shaping students' intentions to utilize AI chat.
Therefore, it is evident that university authorities and curriculum managers should leverage
‘Social influence’ to promote the adoption of ChatGPT. This can be achieved by encouraging
instructors and fellow students who are already using AI chat to advocate for its responsi-
ble use among students. Peers, especially those who have had positive interactions with
ChatGPT, hold significant influence in persuading others to embrace AI chat. Additionally,
the university should establish policies and regulations to govern and promote responsible
usage of generative AI. For instance, the consideration of optional utilization of ChatGPT for
assignments, exams and feedback can be actively explored to facilitate its integration into
student workflows.
‘Facilitating conditions’ play a crucial role in enhancing the utilization of ChatGPT.
Therefore, policies concerning the integration of this tool into the study process should pri-
oritize ensuring adequate access, establishing appropriate regulations, defining guidelines
for when and how ChatGPT can be used and providing effective instructions on how to for-
mulate prompts and address any potential challenges arising from the system's responses.
These measures aim to minimize barriers that may hinder students and instructors from
utilizing AI chat. The findings of this study can inform university policymakers in directing
their efforts towards increasing student awareness and knowledge regarding the benefits
of integrating ChatGPT into their learning process. This can be achieved through the imple-
mentation of training programmes designed to give students the necessary skills to effec-
tively incorporate AI-­powered chat into the educational system.
AI is more than just a tool; it encompasses profound ethical implications that can transform
societal functions and how individuals engage with and view technology. The ethical design
and decision-­making of an AI system can shape our expectations and perceptions of it, far be-
yond its basic functionality. If students believe AI tools operate ethically, they are likely to trust
and use them more. Understanding this trust is crucial. While the UTAUT theory focuses on the
ease of use, from an ethical standpoint, the effort users make to comprehend the AI's decision-­
making process—its transparency—plays a vital role in acceptance. For instance, if there is so-
cietal discourse suggesting that ChatGPT is biased or unfair, it could deter individual adoption.
HEIs can play a proactive role here. They can establish conditions promoting ethical
AI use, like clear AI policies, audit trails and mechanisms to address grievances arising
from AI-­driven decisions. Though the UTAUT framework traditionally associates facilitating
conditions with infrastructural or technological support, in the AI context, this should be ex-
panded. It should include training programmes that inform students about ethical AI usage,
its societal effects and strategies for managing potential ethical dilemmas.
In recent studies, the adoption and the use of AI-­powered tools are more frequently dis-
cussed. There are proposed programmes aimed at enhancing teachers' skills in using var-
ious AI-­powered platforms to improve education. For instance, teachers can participate in
dedicated professional development programmes to learn how to use platforms that eval-
uate students' written assignments (Nazaretsky, Ariely, et al., 2022). On the other hand,
there are platforms designed for teachers that can be implemented in schools to enhance
learning (Cukurova et al., 2023). Additionally, instruments have been developed to measure
how teachers adopt and use AI-­powered tools and how they address various concerns,
such as trust, the absence of human characteristics and transparency in AI decision-­making
processes (Nazaretsky, Cukurova, et al., 2022).
Our study focuses on the other aspect of the education process. We examined how
students adopt and use generative AI-­powered tools. Therefore, the contribution of this
1226 |    STRZELECKI and ElARABAWY

research is to provide insights into the factors that influence the adoption of generative
AI-­powered tools. It is worth noting that ChatGPT is not adaptive and is not specifically de-
signed for educational purposes, which suggests that further studies are needed to explore
the use of this platform in educational processes.
The integration of ChatGPT in HEIs holds the potential to deliver personalized and rele-
vant learning experiences to students, streamline administrative procedures and advance
research and community engagement. However, it is essential to employ ChatGPT in an
ethical manner, taking into account the need to develop individual and institutional capabili-
ties. While certain states and HEIs have implemented restrictions on ChatGPT, most are ac-
tively seeking ways to adapt to the growing prevalence and accessibility of AI. HEIs should
create platforms for stakeholders to engage in discussions about the impact of ChatGPT
and collaboratively devise strategies to adapt and embrace AI technologies. Clear guide-
lines should be established through negotiations with students and instructors, aligning the
use of ChatGPT with course learning objectives. HEIs should also review and update their
policies concerning academic honesty and integrity in relation to ChatGPT and similar AI
tools. Staff training, peer support, mentoring and the introduction of new programmes and
courses focusing on ChatGPT and AI will enhance research and development capabilities
while equipping students with advanced expertise. It is crucial to build capacity for under-
standing and managing ChatGPT, as it cannot replace the indispensability of human creativ-
ity and critical thinking.
There is a critical requirement for extensive deliberations regarding the potential utiliza-
tion, potential risks and inherent limitations of AI tools, underscoring the significance of up-
holding academic and ethical standards, along with critical thinking and human intelligence
playing a paramount role throughout the research endeavour. Proposed solutions should
contain developing guidelines and standards, promoting responsible use and engaging with
the broader community.

Limitations and future work

This research is subject to certain limitations, although the data collection is not restricted to
a single university and the sample is diversified having respondents from six different univer-
sities in two countries, still the sample maybe biased by self-­reported bias. To enhance the
generalizability of the findings, future research should aim to replicate the study model using
a multi-­site approach that involves students from more universities and other countries.
The survey was distributed during a stormy period concerning AI, wherein each month
had a significant impact on the development of the tool. Respondents who used ChatGPT
at the beginning of the tool deployment are considered early adopters. Moreover, the tool's
usage might be influenced by the time of the year, such as during the submission of course-
work, seminal assignments or based on the discipline of study.
In terms of examining the variables influencing ‘Behavioural intention’ and the use of
ChatGPT, this study focused solely on the core elements of the UTAUT prototype model.
Future studies should investigate other comparable models that look into additional factors to
get a more thorough understanding which contributes to students' acceptance of ChatGPT.
Additionally, this study exclusively employed quantitative methods for data analysis.
To provide a more holistic perspective, a future study could incorporate a combination of
qualitative and quantitative methods, allowing for a deeper exploration of the phenome-
non. Furthermore, the study only considered two moderating variables, ‘Gender’ and ‘Study
year’. To fully capture the impact of ChatGPT on intent and behaviour, it is worthwhile to
investigate the moderating effects of other factors such as field of study and more, in order
to better quantify their influence.
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1227

AUTHOR CONTRIBUTIONS
Conceptualization; investigation; funding acquisition; writing—original draft; methodology;
validation; visualization; writing—review and editing; software; project administration; re-
sources; supervision; data curation; formal analysis: Artur Strzelecki. Data curation; re-
sources; writing—review and editing; investigation: Sara ElArabawy.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declared no potential conflicts of interest with respect to the research, author-
ship and/or publication of this paper.

D ATA AVA I L A B I L I T Y S TAT E M E N T

The data that support the findings of this study are available from the corresponding author
upon reasonable request.

E T H I C S S TAT E M E N T
All procedures performed in this study involving adult human participants were in accord-
ance with the ethical standards of the 1964 Helsinki declaration and its later amendments.
Informed consent was obtained from all individual participants included in the study.

ORCID
Artur Strzelecki https://orcid.org/0000-0003-3487-0971
Sara ElArabawy https://orcid.org/0000-0001-9181-7873

REFERENCES
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2),
179–211. https://​doi.​org/​10.​1016/​0749- ­​5978(91)​9 0020 ​-­​T
Al-­Adwan, A. S., Yaseen, H., Alsoud, A., Abousweilem, F., & Al-­Rahmi, W. M. (2022). Novel extension of the
UTAUT model to understand continued usage intention of learning management systems: the role of learning
tradition. Education and Information Technologies, 27(3), 3567–3593. https://​doi.​org/​10.​1007/​s1063​9 -­​021-­​
10758​-­​y
Alghamdi, A. M., Alsuhaymi, D. S., Alghamdi, F. A., Farhan, A. M., Shehata, S. M., & Sakoury, M. M. (2022).
University students' behavioral intention and gender differences toward the acceptance of shifting regular
field training courses to e-­training courses. Education and Information Technologies, 27(1), 451–468. https://​
doi.​org/​10.​1007/​s1063​9 -­​021-­​10701​-­​1
Almaiah, M. A., Alamri, M. M., & Al-­Rahmi, W. (2019). Applying the UTAUT model to explain the students' ac-
ceptance of mobile learning system in higher education. IEEE Access, 7, 174673–174686. https://​doi.​org/​10.​
1109/​ACCESS.​2019.​2957206
Al-­Mamary, Y. H. S. (2022). Understanding the use of learning management systems by undergraduate university
students using the UTAUT model: Credible evidence from Saudi Arabia. International Journal of Information
Management Data Insights, 2(2), 100092. https://​doi.​org/​10.​1016/j.​jjimei.​2022.​100092
Alowayr, A. (2022). Determinants of mobile learning adoption: Extending the unified theory of acceptance and use
of technology (UTAUT). International Journal of Information and Learning Technology, 39(1), 1–12. https://​
doi.​org/​10.​1108/​IJILT​- ­​05-­​2021- ­​0 070
Altalhi, M. (2021). Toward a model for acceptance of MOOCs in higher education: the modified UTAUT model
for Saudi Arabia. Education and Information Technologies, 26(2), 1589–1605. https://​doi.​org/​10.​1007/​s1063​
9- ­​020-­​10317​-­​x
Alvi, I. (2021). College students' reception of social networking tools for learning in India: An extended UTAUT
model. Smart Learning Environments, 8(1), 19. https://​doi.​org/​10.​1186/​s 4056​1- ­​021- ­​0 0164​-­​9
Ayadi, R., & Forouheshfar, Y. (2023). Cultural and creative industries in Egypt, Lebanon and Tunisia: The role
of the EU in building an enabling environment for the emergence of a regional creative hub. EMEA - Euro-
Mediterranean Economists Association. https://​south.​eunei​ghbou​rs.​eu/​wp-­​conte​nt/​uploa​ds/​2023/​07/​
Policy_ ​paper_​CCI_​compr​essed.​pdf
Azizi, S. M., Roozbahani, N., & Khatony, A. (2020). Factors affecting the acceptance of blended learning in med-
ical education: Application of UTAUT2 model. BMC Medical Education, 20(1), 367. https://​doi.​org/​10.​1186/​
s1290​9 - ­​020- ­​02302​-­​2
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Prentice Hall.
1228 |    STRZELECKI and ElARABAWY

Cislaghi, B., & Heise, L. (2020). Gender norms and social norms: Differences, similarities and why they matter
in prevention science. Sociology of Health & Illness, 42(2), 407–422. https://​doi.​org/​10.​1111/​1467-­​9 566.​
13008​
Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring academic integrity in the
era of ChatGPT. Innovations in Education and Teaching International, 1–12. https://​doi.​org/​10.​1080/​14703​
297.​2023.​2190148
Crawford, J., Cowling, M., & Allen, K.-­A . (2023). Leadership is needed for ethical ChatGPT: Character, assess-
ment, and learning using artificial intelligence (AI). Journal of University Teaching and Learning Practice,
20(3), 2. https://​doi.​org/​10.​53761/​1.​20.3.​02
Cukurova, M., Miao, X., & Brooker, R. (2023). Adoption of artificial intelligence in schools: Unveiling factors influ-
encing teachers' engagement. In N. Wang, G. Rebolledo-­Mendez, N. Matsuda, O. C. Santos, & V. Dimitrova
(Eds.), Artificial intelligence in education (pp. 151–163). Cham: Springer. https://​doi.​org/​10.​1007/​978-­​3 -­​031-­​
36272​- ­​9 _​13
Davis, F. D. (1986). A technology acceptance model for empirically testing new end-­user information systems:
Theory and results. Massachusetts Institute of Technology. https://​dspace.​mit.​edu/​handle/​1721.1/​15192​
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1992). Extrinsic and intrinsic motivation to use computers in
the workplace. Journal of Applied Social Psychology, 22(14), 1111–1132. https://​doi.​org/​10.​1111/j.​1559-­​1816.​
1992.​tb009​45.​x
Dečman, M. (2015). Modeling the acceptance of e-­learning in mandatory environments of higher education: The
influence of previous education and gender. Computers in Human Behavior, 49, 272–281. https://​doi.​org/​10.​
1016/j.​chb.​2015.​03.​022
Dijkstra, T. K. (2010). Latent variables and indices: Herman Wold's basic design and partial least squares. In
Handbook of partial least squares (pp. 23–46). Springer Berlin Heidelberg. https://​doi.​org/​10.​1007/​978-­​3 -­​
540-­​32827​- ­​8 _​2
Dijkstra, T. K. (2014). PLS' Janus face – response to professor Rigdon's ‘rethinking partial least squares
modeling: In praise of simple methods. Long Range Planning, 47(3), 146–153. https://​doi.​org/​10.​1016/j.​
lrp.​2014.​02.​0 04
Dijkstra, T. K., & Henseler, J. (2015). Consistent and asymptotically normal PLS estimators for linear structural
equations. Computational Statistics & Data Analysis, 81, 10–23. https://​doi.​org/​10.​1016/j.​c sda.​2014.​07.​
008
Dwivedi, Y. K., Rana, N. P., Jeyaraj, A., Clement, M., & Williams, M. D. (2019). Re-­examining the unified theory
of acceptance and use of technology (UTAUT): Towards a revised theoretical model. Information Systems
Frontiers, 21(3), 719–734. https://​doi.​org/​10.​1007/​s1079​6 -­​017-­​9774-­​y
El-­Masri, M., & Tarhini, A. (2017). Factors affecting the adoption of e-­learning systems in Qatar and USA:
Extending the unified theory of acceptance and use of technology 2 (UTAUT2). Educational Technology
Research and Development, 65(3), 743–763. https://​doi.​org/​10.​1007/​s1142​3 -­​016-­​9508-­​8
Feuerriegel, S., Hartmann, J., Janiesch, C., & Zschech, P. (2023). Generative AI. Business & Information Systems
Engineering. https://​doi.​org/​10.​1007/​s1259​9 - ­​023- ­​0 0834​-­​7
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research.
Addison Wesley.
Guggemos, J., Seufert, S., & Sonderegger, S. (2020). Humanoid robots in higher education: Evaluating the accep-
tance of Pepper in the context of an academic writing course using the UTAUT. British Journal of Educational
Technology, 51(5), 1864–1883. https://​doi.​org/​10.​1111/​bjet.​13006​
Hair, J. F., Hult, G. T. M., Ringle, C., & Sarstedt, M. (2022). A primer on partial least squares structural equation
modeling (PLS- ­SEM) (3rd ed.). Sage.
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011). PLS-­SEM: Indeed a silver bullet. Journal of Marketing Theory and
Practice, 19(2), 139–152. https://​doi.​org/​10.​2753/​MTP10​69- ­​6 6791​9 0202​
Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for assessing discriminant validity in variance-­
based structural equation modeling. Journal of the Academy of Marketing Science, 43(1), 115–135. https://​
doi.​org/​10.​1007/​s1174​7-­​014-­​0 403-­​8
Hoi, V. N. (2020). Understanding higher education learners' acceptance and use of mobile devices for language
learning: A Rasch-­based path modeling approach. Computers & Education, 146, 103761. https://​doi.​org/​10.​
1016/j.​compe​du.​2019.​103761
Huang, W.-­H. D., Hood, D. W., & Yoo, S. J. (2013). Gender divide and acceptance of collaborative Web 2.0 appli-
cations for learning in higher education. The Internet and Higher Education, 16(1), 57–65. https://​doi.​org/​10.​
1016/j.​iheduc.​2012.​02.​0 01
Hwang, G. J., & Chen, N. S. (2023). Editorial position paper: Exploring the potential of generative artificial intel-
ligence in education: Applications, challenges, and future research directions. Educational Technology and
Society, 26(2), 1–18. https://​doi.​org/​10.​30191/​​ETS.​202304_ ​26(2).​0 014
Ivanov, S., & Soliman, M. (2023). Game of algorithms: ChatGPT implications for the future of tourism education
and research. Journal of Tourism Futures, 9(2), 214–221. https://​doi.​org/​10.​1108/​JTF- ­​02-­​2023- ­​0 038
ACCEPTANCE AND USE OF GENERATIVE AI BY HIGHER EDUCATION
STUDENTS    | 1229

Jakkaew, P., & Hemrungrote, S. (2017). The use of UTAUT2 model for understanding student perceptions using
Google Classroom: A case study of Introduction to Information Technology course. In 2017 International
Conference on Digital Arts, Media and Technology (ICDAMT) (pp. 205–209). IEEE. https://​doi.​org/​10.​1109/​
ICDAMT.​2017.​7904962
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G.,
Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O.,
Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and chal-
lenges of large language models for education. Learning and Individual Differences, 103, 102274. https://​
doi.​org/​10.​1016/j.​lindif.​2023.​102274
Kumar, J. A., & Bervell, B. (2019). Google Classroom for mobile learning in higher education: Modelling the initial
perceptions of students. Education and Information Technologies, 24(2), 1793–1817. https://​doi.​org/​10.​1007/​
s1063​9 -­​018-­​09858​-­​z
Lim, W. M., Gunasekara, A., Pallant, J. L., Pallant, J. I., & Pechenkina, E. (2023). Generative AI and the future of
education: Ragnarök or reformation? A paradoxical perspective from management educators. International
Journal of Management Education, 21(2), 100790. https://​doi.​org/​10.​1016/j.​ijme.​2023.​100790
Lund, B. D., & Wang, T. (2023). Chatting about ChatGPT: How may AI and GPT impact academia and libraries?
Library Hi Tech News, 40(3), 26–29. https://​doi.​org/​10.​1108/​LHTN- ­​01-­​2023- ­​0 009
Lund, B. D., Wang, T., Mannuru, N. R., Nie, B., Shimray, S., & Wang, Z. (2023). ChatGPT and a new academic
reality: Artificial Intelligence-­written research papers and the ethics of the large language models in scholarly
publishing. Journal of the Association for Information Science and Technology, 74(5), 570–581. https://​doi.​
org/​10.​1002/​asi.​24750​
Lv, Z. (2023). Generative artificial intelligence in the metaverse era. Cognitive Robotics, 3, 208–217. https://​doi.​
org/​10.​1016/j.​cogr.​2023.​0 6.​0 01
Nazaretsky, T., Ariely, M., Cukurova, M., & Alexandron, G. (2022). Teachers' trust in AI-­powered educational tech-
nology and a professional development program to improve it. British Journal of Educational Technology,
53(4), 914–931. https://​doi.​org/​10.​1111/​bjet.​13232​
Nazaretsky, T., Cukurova, M., & Alexandron, G. (2022). An instrument for measuring teachers' trust in AI-­
based educational technology. In ACM International Conference Proceeding Series (pp. 56–66). LAK22:
12th International Learning Analytics and Knowledge Conference, ACM. https://​doi.​org/​10.​1145/​35068​60.​
3506866
Nikolopoulou, K., Gialamas, V., & Lavidas, K. (2020). Acceptance of mobile phone by university students for their
studies: an investigation applying UTAUT2 model. Education and Information Technologies, 25(5), 4139–
4155. https://​doi.​org/​10.​1007/​s1063​9 -­​020-­​10157​- ­​9
OpenAI. (2023). ChatGPT: Optimizing language models for dialogue. https://​openai.​com/​blog/​chatg​pt/​
Perkins, M. (2023). Academic integrity considerations of AI large language models in the post-­pandemic era:
ChatGPT and beyond. Journal of University Teaching and Learning Practice, 20(2), 7. https://​doi.​org/​10.​
53761/​1.​20.​02.​07
Raman, A., & Don, Y. (2013). Preservice teachers' acceptance of learning management software: An application
of the UTAUT2 model. International Education Studies, 6(7), 157–164. https://​doi.​org/​10.​5539/​ies.​v6n7p157
Ringle, C. M., Wende, S., & Becker, J.-­M. (2022). SmartPLS 4. SmartPLS GmbH.
Rogers, E. M. (1962). Diffusion of innovations. The Free Press of Glencoe.
Rudolph, J., Tan, S., & Tan, S. (2023). ChatGPT: Bullshit spewer or the end of traditional assessments in higher
education? Journal of Applied Learning & Teaching, 6(1), 1–22. https://​doi.​org/​10.​37074/​​jalt.​2023.6.​1.​9
Samsudeen, S. N., & Mohamed, R. (2019). University students' intention to use e-­learning systems. Interactive
Technology and Smart Education, 16(3), 219–238. https://​doi.​org/​10.​1108/​ITSE-­​11-­​2018- ­​0 092
Sarstedt, M., Ringle, C. M., & Hair, J. F. (2022). Partial least squares structural equation modeling. In C. Homburg,
M. Klarmann, & A. Vomberg (Eds.), Handbook of market research (pp. 587–632). Springer International
Publishing. https://​doi.​org/​10.​1007/​978-­​3 -­​319-­​57413​-­​4 _​15
Strzelecki, A. (2023). To use or not to use ChatGPT in higher education? A study of students' acceptance and use
of technology. Interactive Learning Environments, 1–14. https://​doi.​org/​10.​1080/​10494​820.​2023.​2209881
Taylor, S., & Todd, P. A. (1995). Understanding information technology usage: A test of competing models.
Information Systems Research, 6(2), 144–176. https://​doi.​org/​10.​1287/​isre.6.​2.​144
Thompson, R. L., Higgins, C. A., & Howell, J. M. (1991). Personal computing: Toward a conceptual model of utili-
zation. MIS Quarterly, 15(1), 125. https://​doi.​org/​10.​2307/​249443
Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology:
Toward a unified view. MIS Quarterly, 27(3), 425–478. https://​doi.​org/​10.​2307/​30036540
Venkatesh, V., Thong, J. Y. L., & Xu, X. (2012). Consumer acceptance and use of information technology:
Extending the unified theory of acceptance and use of technology. MIS Quarterly, 36(1), 157–178. https://​
doi.​org/​10.​2307/​41410412
1230 |    STRZELECKI and ElARABAWY

Wong, K. T., Teo, T., & Goh, P. S. C. (2015). Understanding the intention to use interactive whiteboards: model
development and testing. Interactive Learning Environments, 23(6), 731–747. https://​doi.​org/​10.​1080/​10494​
820.​2013.​806932
Yu, C.-­W., Chao, C.-­M., Chang, C.-­F., Chen, R.-­J., Chen, P.-­C., & Liu, Y.-­X. (2021). Exploring behavioral intention
to use a mobile health education website: An extension of the UTAUT 2 model. SAGE Open, 11(4), 1–12.
https://​doi.​org/​10.​1177/​21582​4 4021​1055721
Zwain, A. A. A. (2019). Technological innovativeness and information quality as neoteric predictors of users'
acceptance of learning management system. Interactive Technology and Smart Education, 16(3), 239–254.
https://​doi.​org/​10.​1108/​ITSE- ­​09-­​2018- ­​0 065

How to cite this article: Strzelecki, A., & ElArabawy, S. (2024). Investigation of the
moderation effect of gender and study level on the acceptance and use of generative AI
by higher education students: Comparative evidence from Poland and Egypt. British
Journal of Educational Technology, 55, 1209–1230. https://doi.org/10.1111/bjet.13425