câu hỏi

Computers and Education: Artificial Intelligence 7 (2024) 100274
Contents lists available at ScienceDirect
Computers and Education: Artificial Intelligence

journal homepage: www.sciencedirect.com/journal/computers-and-education-artificial-intelligence
A Structural equation modeling analysis of generative AI chatbots adoption

among students and educators in higher education
Afef Saihi * , Mohamed Ben-Daya , Moncer Hariga , Rami As’ad
Industrial Engineering Department, American University of Sharjah, Sharjah, United Arab Emirates
A R T I C L E I N F O A B S T R A C T
Keywords: In an era where artificial intelligence (AI) is reshaping educational paradigms, this study explores AI-based
Generative AI chatbot adoption in higher education among students and educators. Employing a Structural Equation
Chatbots Modeling (SEM) approach, the research focuses on developing and validating a comprehensive model to un-
Higher education
derstand the multifaceted factors impacting the acceptance and use of these chatbots. The methodology in-
Perception
Adoption
tegrates an extensive literature review, construction of a theoretical model, administration of a detailed
Learning experience questionnaire to a representative sample from the higher education sector, coupled with advanced SEM tech-
niques for data analysis and interpretation. The SEM analysis validates the model’s robustness and highlights the
relationships between several key factors affecting users’ perspectives and chatbots adoption. Results reveal a
predominantly positive perception towards AI-chatbots among both students and educators, underscoring the
potential to substantially enrich their educational journey. However, it also uncovers critical concerns pertaining
to trust, privacy, response bias, and information accuracy. Moreover, the study offers valuable insights into how
moderators such as technological proficiency, user roles, and gender influence the adoption model relationships.
This emphasizes the need for customizing AI-chatbots deployment to meet the diverse needs of users effectively.
Contributing a robust framework for understanding users’ perceptions towards AI-chatbots and their adoption
patterns, this study offers actionable insights for educational leaders, policymakers, and technology developers.
It also lays the groundwork for future research, including longitudinal studies to evaluate the long-term impact of
these technologies, investigations into their effect on learning outcomes, and explorations of the ethical and
privacy considerations involved.
1. Introduction America aid users with financial management and transactions (Mori &
Du, 2023). Within these diverse applications, the domain of higher ed-
The twenty-first century is a transformative era marked by rapid and ucation emerges as a particularly impactful and promising field for the
profound technological advancements. Among the most compelling and integration of AI-based chatbots. These intelligent systems can interact
influential of these innovations are Artificial Intelligence (AI)-based with users in natural language, answer queries, provide feedback, and
chatbots, which are redefining new ways of interaction with technology. even guide learners through complex concepts and procedures, offering
These intelligent conversational agents powered by AI have been transformative potential for teaching and learning processes (Gill et al.,
essential in driving change and improving efficiency across a wide range 2024; Okonkwo & Ade-Ibijola, 2021).
of sectors. To mention a few examples, in healthcare, chatbots like A salient example of this technology is OpenAI’s Generative Pre-
Babylon Health, which was recently acquired by eMed Healthcare trained Transformer (GPT), an AI-driven chatbot that has gained
(Adams, 2023), were among the first to revolutionize patient care by considerable attention for its impressive language generation capabil-
providing preliminary diagnoses and scheduling appointments (Le ities (Ray, 2023). GPT can produce human-like text, answer queries,
Nguyen & Do, 2019). In retail, AI-chatbots like Sephora’s assistant are translate languages, write essays, and even create poetry (Fitria, 2023).
enhancing customer experience by offering personalized product rec- Its ability to understand context and generate meaningful and coherent
ommendations and handling customer inquiries in real time (Lee, 2020). responses has opened up a plethora of applications in customer service
Similarly, in the financial sector, chatbots like Erica from Bank of or virtual assistance, and also in creative fields such as education. The
* Corresponding author.
E-mail addresses: [email protected] (A. Saihi), [email protected] (M. Ben-Daya), [email protected] (M. Hariga), [email protected] (R. As’ad).
https://doi.org/10.1016/j.caeai.2024.100274
Received 20 April 2024; Received in revised form 28 July 2024; Accepted 2 August 2024
Available online 3 August 2024
2666-920X/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
A. Saihi et al. Computers and Education: Artificial Intelligence 7 (2024) 100274
emergence of AI-based chatbots in the higher education environment Therefore, this paper aims to explore the complex dynamics shaping
offers a promising avenue for enhancing both teaching and learning the perceptions of educators and students towards AI-based chatbots and
experiences. These smart conversational agents can facilitate continuous their subsequent impact on the adoption of these technologies in higher
learning support, provide instant responses to student queries, offer education settings. The primary objectives of this study are.
personalized learning pathways, and alleviate administrative burden by
automating routine tasks (Saaida, 2023). These potential benefits align 1. Identify the key variables and factors that shape users’ perceptions
well with the evolving demands of today’s education sector which is and influence the acceptance and use of AI-based chatbots among
characterized by increasing student diversity, growing emphasis on educators and students in higher education.
personalized learning, and the expanding role of online and blended 2. Propose a comprehensive model to measure these variables, explain
learning formats (Bhutoria, 2022; Kamalov et al., 2023). the relationships among them, and understand to what extent do
Several educational institutions worldwide are already leveraging they impact the adoption of AI-chatbots, users satisfaction and
the power of AI-based chatbots. For instance, Georgia Tech’s "Jill Wat- learning outcomes.
son" provides a compelling illustration of how an AI-based chatbot can 3. Clarify the moderating role of the individual differences among users
act as a teaching assistant, seamlessly responding to students’ queries on on the relationships among these variables.
an online platform (Mello et al., 2023). “Beacon”, developed at the 4. Provide actionable insights to enhance the implementation and user
University of Staffordshire, is an AI-powered chatbot designed to acceptance of AI-chatbots in educational contexts.
enhance the university experience for students. It offers personalized
assistance with scheduling, facilitates communication with teachers, Based on a Partial Least Square Structural Equation Modeling (PLS-
and provides information on campus facilities. Similarly, Deakin Uni- SEM) approach, this study intends to develop and validate a model to
versity in Australia employs an AI-based chatbot called "Deakin Genie," elucidate the complex relationships between multiple variables influ-
which helps students manage their studies by answering questions, encing the acceptance and use of AI-chatbots in higher education. This
reminding them of deadlines, and even guiding them to resources and contributes to uncover the multifaceted interdependencies among
services (Monserrat et al., 2022). Such use cases exemplify how AI-based perception factors, trust and privacy factors, interaction and system
chatbots can revolutionize several facets of higher education, ranging factors, social and contextual factors, and the eventual outcome vari-
from student learning to administrative tasks. They have the potential to ables. Exploring these factors involves understanding diverse perspec-
foster learning experiences that are more personalized and engaging, tives from students and educators, with an extension of the analysis to
streamline administrative processes, and provide timely support and encompass diverse fields of study. In addition to presenting technolog-
feedback for students and educators. ical advancements, the integration of AI-based chatbots intersects with
The integration of AI-based chatbots into the higher education sector various educational theories and pedagogical practices. This study
presents several promising opportunities for both students and in- provides insights into how AI-chatbots can enhance and transform
stitutions. Firstly, they offer continuous availability and instant educational experiences, drawing on principles from constructivist
responsiveness, allowing students to access support and information at learning theory, connectivism, self-determination theory (SDT), and
any time regardless of geographical location (Kamalov et al., 2023). This cognitive load theory. To that end, this research design involves a
accessibility ensures that students receive timely assistance and elimi- carefully constructed questionnaire administered to a diverse group of
nates delays often associated with traditional support systems. Secondly, respondents within the higher education sector. The study encompasses
AI-chatbots facilitate individualized learning experiences by custom- both students and educators, with the goal of capturing distinct per-
izing content and recommendations according to each student’s needs. spectives and individual needs associated with the use of generative AI
Through the analysis of students’ data, chatbots can provide adaptive in academic contexts. Furthermore, the research aims to analyze dif-
learning pathways, customized study materials, and targeted feedback, ferences in perception and adoption among various fields of study, such
thereby fostering engagement and addressing specific learning gaps as engineering, medical studies, business studies, among others. The
(Chang et al., 2023). Thirdly, AI-chatbots streamline administrative outcomes of this study contribute to the rapidly growing knowledge base
processes which contributes to freeing up valuable time and resources in this field, and provide actionable insights for stakeholders in higher
for educators and administrators. Routine tasks such as answering education to optimize the utilization of these innovative technologies.
common inquiries, managing course registrations, or providing admin- The remainder of this paper is structured as follows: Section 2 pre-
istrative information can be automated, enabling staff to concentrate on sents a comprehensive review of the related literature. Section 3 dis-
value-added activities like curriculum design, student mentoring, and cusses the different stages of the research methodology employed to
academic support (Ahmad et al., 2022). Lastly, AI-chatbots contribute to achieve the study’s objectives. Section 4 thoroughly outlines the foun-
the enhancement of student engagement and interaction. Through nat- dational framework of factors, variables and indicators that underpin
ural language processing (NLP) and conversational interfaces, chatbots this research. The preliminary model is introduced in Section 5, while
can engage students in interactive conversations, simulate real-world Section 6 focuses on the estimation of the model, establishing its reli-
scenarios, and facilitate discussions, promoting active learning and ability and validity, testing the hypotheses, and discussing the impli-
participation (Belda-Medina & Calvo-Ferrer, 2022). cations of the findings. Finally, Section 7 provides concluding remarks
Despite the transformative potential of AI-based chatbots in higher and outlines avenues for future research exploration.
education, their integration and acceptance are not without challenges
(Hwang & Chang, 2023). Various factors influence educators’ and stu- 2. Related literature review
dents’ perceptions and attitudes towards these technologies. Key con-
cerns include trust in AI technology, the perceived usefulness and ease of 2.1. Emerging research streams in this field
use of AI-chatbots, data privacy issues, and the fear of dehumanization
of education, among others (Chan & Hu, 2023; Robayo-Pinzon et al., The integration of generative AI tools has catalyzed a wave of
2023). Thus, understanding the dynamics that shape perceptions to- innovation within the higher education domain. These advancements
wards AI-based chatbots in higher education and addressing the asso- have introduced fresh methodologies aimed at enhancing student
ciated concerns are of paramount importance. This has made the engagement, providing support, and enhancing overall learning expe-
adoption and integration of such technologies in the higher education riences. An extensive literature review identified numerous research
landscape a dynamic area of research, seeking to understand the factors streams that have emerged in this dynamic and rapidly growing field.
that shape users’ perceptions and the consequent impact on their utili- Firstly, the impact on teaching and learning is a prominent research
zation of generative AI tools. stream that explores how generative AI and chatbots influence teaching
2
methods and student learning experiences. In this context, Gill et al. the computing education community to shape these developments
(2024) discussed the revolutionary impact of ChatGPT on contemporary effectively. Pavlik (2023) explored the transformative potential of
education, emphasizing AI-chatbots’ potential to revolutionize teaching generative AI, focusing on ChatGPT’s capabilities and limitations and its
methods. Similarly, Wang, Lund, et al. (2023) explored the potential implications for journalism and media education.
advantages, such as individualized learning experiences and adaptive Many other research streams emerged in this domain such as
testing, as well as the associated concerns including privacy issues and research on transfer learning (Syed et al., 2021), usability and user
cultural differences. They highlighted the role of AI in enhancing experience within educational chatbots (Lim et al., 2021; Plantak
learning efficiency and supporting customization in education for in- Vukovac et al., 2021), mental health support and wellbeing in higher
ternational students. education (Dhanasekar et al., 2021), building trust and engagement
Secondly, another research stream is devoted to educational chatbots (Muñoz et al., 2023), enhancing administrative processes in higher ed-
and language processing. It focuses on the development of AI-chatbots in ucation (Aloqayli & Abdelhafez, 2023). Interested readers are referred
educational contexts and their NLP capabilities. As a matter of fact, to the works of Bahroun et al. (2023), Montenegro-Rueda et al. (2023),
Iskandar et al. (2023) showcased the implementation of NLP and ma- Pradana et al. (2023), Vargas-Murillo et al. (2023), and İpek et al. (2023)
chine learning in chatbots for new student admission. They utilized for more details pertaining to the various research paths in this field.
Google Dialogflow, a comprehensive platform for developing chatbots
and virtual agents, and achieved a chatbot-based communication me- 2.2. Educational theories and AI-chatbots integration
dium with favorable usability and an 80% accuracy rate in automatic
correction tests for user text. Similarly, Chand and Sharma (2023) dis- The integration of AI-chatbots in higher education intersects with
cussed the potential of using chatbots with NLP capabilities to support several established educational theories, offering new perspectives on
learners when individual teacher attention is limited. their potential impact and application (Jain et al., 2024). Understanding
Furthermore, another line of research highlighted how AI-based these intersections is crucial for developing a comprehensive framework
chatbots enhance the assessment and feedback processes in higher ed- for their adoption and utilization in educational settings. As noted by
ucation. Smolansky et al. (2023) conducted a survey, involving 36 ed- Chen et al. (2020), while AI in Education (AIEd) scholars have explored
ucators and 389 students from two distinct universities, to explore the a variety of perspectives and issues, there has been limited focus on
perspectives of educators and students regarding the impact of genera- educational theories, with constructivism being the primary theory
tive AI tools, like ChatGPT, on assessment methods in higher education. mentioned in existing studies. The authors recommended that AIEd re-
Naidu and Sevnarayan (2023) explored the increasing impact of searchers place greater emphasis on integrating educational theories
ChatGPT on online assessment in remote education, underlining the with AI technologies.
importance of ethical considerations and continued research in this Firstly, constructivist learning theory emphasizes the importance of
domain. Additionally, the convergence of learning analytics is an active, student-centered learning where learners construct knowledge
important research stream, as exemplified by Sharef et al. (2021). In this through meaningful interactions (Tafrova-Grigorova, 2016).
domain, researchers explore the use of AI-driven systems to collect and AI-chatbots can facilitate this process by providing personalized and
analyze data, tailoring learning experiences to meet the unique needs of interactive learning experiences. For instance, chatbots can engage
each student. Such efforts lead to adaptive educational content delivery students in dialogue, offer instant feedback, and adapt to individual
and timely interventions and contribute to revolutionizing the teaching learning needs, thus promoting a more constructivist approach to edu-
and learning processes. Many other studies contributed to this stream cation (Kim & Adlof, 2024). Secondly, connectivism, a theory proposed
such as Bao et al. (2021) who advocated for open-domain chatbots by Siemens (2005), highlights the role of technology and networked
developed through curriculum learning, further enhancing personalized connections in the learning process. AI-chatbots serve as nodes within a
learning pathways. student’s learning network, offering continuous access to information,
Another crucial research stream addressed ethical aspects and chal- resources, and support. This aligns with the connectivist perspective that
lenges posed by generative AI in higher education. To mention a few, knowledge is distributed across a network of connections, and learning
Dalalah and Dalalah (2023) discussed the challenges of false positive consists of the ability to construct and traverse these networks (Jain
and false negative detection in generative AI tools like ChatGPT, high- et al., 2024). Thirdly, SDT, developed by Deci & Ryan (1985), focuses on
lighting that the likelihood of detecting AI-generated text is lower in the role of autonomy, competence and relatedness in enhancing intrinsic
article abstracts compared to literature sections, and emphasizing the motivation and engagement (Xia et al., 2022). AI-chatbots can support
critical need for ethics and responsibility in the use of these models. these components by offering personalized assistance, fostering a sense
Khurma et al. (2023) critically examined ChatGPT’s impact on UAE of competence through timely feedback, and enhancing relatedness by
education, emphasizing equity and governance concerns, providing in- providing continuous support and interaction. Studies have shown that
sights into its benefits and drawbacks, and offering guidance for its technologies that support these psychological needs can significantly
ethical and safe use. enhance learning outcomes (Xia et al., 2023). Finally, cognitive load
Another rich line of research, that is tightly connected to the aim of theory addresses the importance of managing cognitive load to optimize
this study, evolved around users’ perspectives and perceptions. For learning (Van Merrienboer & Sweller, 2005). AI-chatbots can reduce
instance, Chan and Hu (2023) highlighted the crucial role of student extraneous cognitive load by handling routine inquiries and providing
perspectives in shaping the use of generative AI and emphasized the targeted support, allowing students to focus on more complex learning
importance of involving end-users in the development and imple- tasks (Schmidhuber et al., 2021).
mentation of AI tools in education. Guo et al. (2023) provided a
comprehensive analysis by considering both educator and student per- 2.3. Adoption of AI-driven chatbots in education and role of users’
spectives, providing valuable insights into how AI affects teaching and perspectives
learning and offering a holistic understanding of its influence on
educational experiences. The integration of AI-based chatbots in the higher education sector
Moreover, subject specific applications received notable attention by has attracted considerable attention in recent years. Universities
researchers in this field. For example, Currie and Barry (2023) explored worldwide are exploring the potential of these conversational agents to
the impact of ChatGPT on nuclear medicine education. Becker, Denny, transform various aspects of the educational environment (Bahroun
et al. (2023) discussed the educational opportunities and challenges et al., 2023). AI-based chatbots offer the capability to interact with
arising from AI-driven code generation tools in introductory program- students in real-time, provide personalized support, streamline admin-
ming, emphasizing the need for prompt and coordinated action within istrative processes, and enhance the overall learning experience (Antony
3
& Ramnath, 2023). They have been deployed in diverse ways and users, chatbots evolve into adaptable and responsive companions,
commonly integrated into learning management systems, educational thereby fostering overall student success as well as a deeper sense of
websites, and mobile applications to assist students throughout their partnership between students and educational institutions (Chen et al.,
academic journey (Adiguzel et al., 2023). Chatbots can handle routine 2023). In essence, user perspectives are the cornerstone of successful
administrative tasks, answer frequently asked questions, provide course AI-chatbot adoption in higher education. Actively involving students
recommendations, offer study resources, and facilitate student engage- and educators in shaping the development and deployment of these
ment (Ilieva et al., 2023; Kurni et al., 2023). These virtual assistants act virtual assistants allows institutions to unlock the full potential of AI
as continuous companions and ensure prompt access to information and technology to deliver personalized, efficient, and enriching educational
guidance, thus improving the overall efficiency and convenience of experiences (Yusup, 2023).
educational services. Furthermore, AI-chatbots enable customized
learning experiences based on the specific needs of individual students. 2.4. Challenges in the adoption of AI-based chatbots
Through the use of advanced machine learning algorithms, these chat-
bots can analyze student data, track progress, and provide targeted While the adoption of AI-based chatbots in higher education presents
recommendations for academic support or interventions (Bhutoria, significant advantages, it also introduces various challenges that require
2022; Tapalova & Zhiyenbayeva, 2022). They can adapt to different meticulous attention. These challenges have been discussed in recent
learning styles, offer interactive quizzes or simulations, and deliver research studies, shedding light on the complexities associated with
personalized feedback to promote student engagement and compre- integrating AI-chatbots into educational contexts. Firstly, their utiliza-
hension. The dynamic and interactive nature of AI-chatbots provides a tion involves gathering and processing sensitive user data which raises
flexible and learner-centric approach to education and enables students substantial privacy concerns. The potential misuse or mishandling of
to learn at their individual pace and within their preferred learning this data poses a substantial challenge. This brings privacy, data security
environment (Alam & Mohanty, 2023; Shamkuwar & Sharma, 2023). and ethical concerns to the forefront of challenges facing the adoption of
Moreover, AI-chatbots contribute to the digital transformation of AI-chatbots. As emphasized by Chan and Hu (2023), student data, when
higher education institutions. They play a vital role in shaping the shift processed by chatbots, requires stringent protection measures. To
towards online and blended learning formats, particularly in distance address this challenge, institutions must prioritize data security,
education programs (Alshahrani, 2023; Stupina & Paniotova, 2023). compliance with privacy regulations, and transparent communication
These chatbots serve as virtual instructors, guiding students through about data handling practices. The responsible and ethical use of tech-
online courses, facilitating discussions, and providing timely feedback nology, as advocated by Rasul et al. (2023), is crucial in ensuring that
(Hew et al., 2023). Additionally, they alleviate the workload of educa- the potential benefits of chatbots do not compromise user privacy.
tors by automating administrative tasks including grading assignments Secondly, the risk of dehumanization in the educational experience due
and managing course registrations. This allows educators to concentrate to an overreliance on AI-chatbots is a substantial concern. Critics argue
more on the design of instructional content, mentoring, and providing that this overreliance may erode the human element in education,
personalized support to students (Zhai, 2022). The adoption of AI-based impacting empathy, social interaction, and the development of
chatbots in higher education is driven by the promise of improving non-cognitive and soft skills. Thus, the challenge of maintaining a
student experiences, enhancing teaching practices, and optimizing human touch in education, as discussed by Rane (2023), is pivotal.
institutional operations. As universities seek innovative solutions to While AI-chatbots can offer efficiency, they must not replace the
address the evolving needs of today’s learners, AI-chatbots offer op- fundamental interactions that enrich the educational experience.
portunities for these institutions to deliver high-quality, accessible, and Balancing technology with human interaction is essential to address this
personalized learning experiences (Wang, Lund, et al., 2023). challenge. Educational institutions must consider the ethical implica-
Essentially, the integration of users’ perspectives has shed light on tions of replacing human educators with chatbots and work to ensure
the pathway for chatbots to evolve beyond mere automation tools. It has that technology enhances the learning journey rather than diminishing
led to a paradigm shift where these AI-driven entities now act as col- it (Ifelebuegu et al., 2023).
laborators in the educational journey (Chang et al., 2023). Thorough Furthermore, technical issues and limitations such as inaccuracies in
consideration of students’ experiences and preferences significantly NLP, understanding complex queries, and technical glitches can hinder
contributes to improving the design and functionality of these chatbots. the performance of AI-chatbots and affect their reliability and usability.
The active integration of user feedback supports educational institutions As highlighted by Ray (2023), these challenges pose significant obstacles
in refining these tools to become more intuitive, empathetic, and to the seamless adoption of AI-chatbots in higher education. These issues
responsive (Karyotaki et al., 2022). The centrality of users’ perspectives demand continuous development and improvement efforts to ensure
in this evolution cannot be overstated due to their indispensable role in that chatbots deliver reliable and accurate support, and institutions must
ensuring the seamless adoption of these tools in higher education. Stu- invest in refining chatbots’ functionalities and user-friendliness to foster
dents and educators, as the primary end-users, provide invaluable in- trust among both students and educators. Additionally, ethical consid-
sights into the effectiveness and usability of these AI-driven tools erations, including algorithmic bias, fairness, and accountability, are
(Mogavi et al., 2023). Their perceptions and feedback directly influence central to the responsible adoption of AI-chatbots. Ensuring that these
the refinement and optimization of chatbot functionalities. As claimed technologies do not perpetuate discrimination or inequalities remains a
by Chan and Hu (2023), the way users perceive things has a substantial complex challenge. Active engagement and ethical education, as
impact on how they approach learning and the ultimate results they emphasized by Sakib et al. (2023), are essential components in tackling
achieve. Gaining insight into their perspectives allows educators and such ethical implications. Thus, institutions must prioritize responsible
policymakers to customize these technologies to cater to their needs and and ethical use, offering guidance on fairness, bias mitigation, and
address their concerns, ultimately fostering more successful learning accountability to ensure that AI-chatbots enhance users’ learning ex-
outcomes. Furthermore, the active engagement of students in periences (Díaz-Rodríguez et al., 2023). Another challenge that faces
co-creating their educational support systems empowers them and AI-based chatbots adoption in higher education is maintaining academic
promotes a sense of ownership regarding their learning experiences integrity. In this context, Cotton et al. (2023) discussed the risk associ-
(Chang et al., 2023). Thus, the integration of users’ perspectives ensures ated with students unethically using ChatGPT, leading to unintelligent
that AI-chatbots align closely with their specific and diverse needs and and unlearned responses. Educators may face challenges in dis-
preferences. It enables institutions to tailor chatbot interactions to tinguishing between authentic student work and that generated by
match diverse learning styles, cultural backgrounds, and individual chatbots which poses difficulties in assessing real learning outcomes and
learning paces (Zhang et al., 2023). Through ongoing dialogue with achievement.
4
One of the important challenges to this adoption also lies in tech- education sector. Additionally, the review seeks to determine variables
nological readiness. Add to that, compatibility issues with legacy sys- that reflect the degree of adoption and satisfaction experienced by users.
tems and the need for continuous updates further complicate the To ensure comprehensiveness of the extracted components, a search
technological environment. Therefore, integrating AI-chatbots into string is carefully and iteratively designed. Subsequently, articles search
existing systems and platforms requires substantial technical infra- and selection are conducted from Scopus database using the following
structure and expertise. Universities may need to invest in robust IT keywords: ("Chatbot" OR "ChatGPT" OR "GPT*" OR "Google Bard" OR
systems, data security measures, and staff training to ensure the smooth "openai" OR ("Generat*" AND ("Artificial Intelligence" OR "AI"))) AND
functioning of chatbots (Michel-Villarreal et al., 2023). Moreover, ("educat*" OR "teach*" OR "learn*" Or "student*" OR "*graduate" OR
aligning AI-driven chatbots with pedagogical goals and teaching "academi*" OR "instructor*").
methods can be challenging. Developing meaningful and effective in- Given the latent nature of the factors and variables derived from the
teractions that enhance learning rather than disrupting it requires previous research phase, the literature review is extended to identify the
careful planning. Institutions need to invest in instructional design diverse indicators that are capable of collectively capturing the essence
expertise and collaborate closely with educators to create of each variable and reflectively measuring it. Afterwards, building upon
chatbot-driven experiences that complement classroom instruction the insights gained from the literature review, a hierarchical structure of
(Chang et al., 2023). Finally, the study conducted by Chan and Hu relationships is constructed. This structure groups factors and their
(2023) emphasizes a significant long-term challenge in the adoption of corresponding variables and indicators, thereby establishing a compre-
these chatbots in higher education which is the potential impact on in- hensive framework underpinning this research. The relationships
dividual growth, professional opportunities, and societal values. As established within this framework establish the basis for the subsequent
highlighted in their research, concerns have been raised regarding how model development and analysis.
the integration of generative AI technologies may influence students’ In the phase of preliminary model design, the insights gathered from
educational journeys and their broader personal and professional the extensive literature review are transformed into a tangible research
growth. These concerns relate to questions about whether AI-chatbots framework. During this phase, abstract components extracted from the
might unintentionally alter students’ learning experiences, future literature evolved into a structured model that maps out the complex
career opportunities, and even societal values. web of relationships among factors, variables, and indicators. This
model comprises both a measurement model and a path model. The
3. Research methodology measurement model defines the approach to assess and quantify the
various latent constructs based on relevant indicators, whereas the path
The research design adopted in this study follows a well-structured model outlines the hypothetical causal connections between these con-
process that encompasses various consecutive phases including structs, forming the basis of the research hypotheses. This critical step in
comprehensive literature review, model development, questionnaire the research process reflects the complexity of users’ perceptions in the
design, data collection, and advanced data analysis techniques. A visual context of AI-based chatbot adoption and lays the foundation for sub-
representation of the research process is provided in Fig. 1. sequent empirical testing and validation.
A fundamental step in this research involves conducting an exhaus- During the questionnaire design and pilot testing phase, careful
tive literature review. This review aims to identify factors that shape attention is devoted to designing a robust and tailored instrument for
users’ perceptions regarding AI-based chatbots’ adoption in the higher data collection. Building upon the framework established in previous
Fig. 1. Phases of the research process.
5
research phases and the authors’ extensive experience in survey design SmartPLS, and the graphical output and complex statistical results can
and development, a comprehensive questionnaire was meticulously be misinterpreted without a thorough understanding of PLS-SEM
constructed. This questionnaire aligns with the identified factors, vari- methodology. Furthermore, the estimates of path coefficients can be
ables and indicators, ensuring that it effectively captures the various biased if there are collinearity issues or if the indicators do not appro-
dimensions of user perceptions and adoption factors. Prior to its wider priately capture the latent variables. To that end, it is recommended to
application, a pilot test of this instrument was conducted through ensure the reliability and validity of the measurement model and rule
engaging a specific group of educators and students, representing out the collinearity issues before estimating the structural model (Hair Jr
diverse perspectives within the higher education environment. Their et al., 2021).
feedback was indispensable in refining the questionnaire and enhancing
its clarity and relevance. This iterative process of design and refinement 4. Foundational framework of factors and indicators
is vital in ensuring that the questionnaire is in line with the research aim
and extracts meaningful insights. In this section, a comprehensive framework is introduced encom-
In the subsequent phases of the research, the focus shifted towards passing various factors, variables, and their corresponding indicators.
the critical processes of data collection, analysis and model validation. A The identification of these components was achieved through an
comprehensive approach was employed to collect data from a diverse exhaustive literature review which is a fundamental step in the research
group of 192 participants, including 107 students and 85 educators process. Variables were initially identified, then clustered into thematic
across various academic disciplines. The survey was initially sent to factors based on common attributes and relationships. Given their latent
around 400 prospects, resulting in a response rate of 48%. The age range nature, corresponding indicators to measure these variables were also
of participants was between 18 and 65 years or over, with a gender systematically identified. Furthermore, the framework integrates in-
distribution of 54% female and 46% male. Participants came from sights from several educational theories, including constructivist
diverse subject backgrounds including engineering, medical studies and learning theory, connectivism, SDT, and cognitive load theory.
business studies. The data was collected over a period of three months, Many of the variables used in this study are derived from established
and the questionnaire was administered via an online platform to technology acceptance models, such as the Technology Acceptance
facilitate wide distribution and ease of access. Participants were invited Model (TAM) by Davis (1989, pp. 205–219) and the Unified Theory of
to complete the survey through email invitations and professional social Acceptance and Use of Technology (UTAUT) by Venkatesh et al. (2003).
networks such as LinkedIn. The selection of participants was carried out Technology acceptance refers to the process through which individuals
using a blend of purposive, stratified, and snowball sampling methods, or organizations begin to embrace and utilize a new technology to
ensuring a balanced representation of both students and professors from achieve desired outcomes (Yadegari, Mohammadi, & Masoumi, 2022).
various strata. Once the data was assembled, the analytical journey TAM emphasizes the importance of perceived ease of use and perceived
began with the use of PLS-SEM modeling techniques. SEM represents a usefulness as key determinants of technology acceptance (Liesa-Orús
model fitting framework encompassing various multivariate methods et al., 2023). UTAUT extends this framework by incorporating addi-
that seamlessly integrates a range of analytical tools such as measure- tional factors such as social influence and facilitating conditions (Robles
ment theory, factor analysis, regression techniques, path analysis, Urquijo et al., 2019). In addition to these established variables, other
among others (Hair et al., 2011; Kiraz et al., 2020). This method proves relevant factors identified in the literature are also included such as
especially valuable when the research objectives involve the examina- trust, privacy concerns, system quality, and information quality, among
tion of complex and multifaceted constructs, emphasizing the specifi- others. The objective is to create a more comprehensive framework and
cation of a complex network of relationships rather than a simple allow for a holistic examination of the factors influencing AI-chatbot
relationship between a response variable and a set of explanatory vari- adoption and use. The selection of indicators used to measure these
ables. Thus, this approach was chosen due to its suitability for predicting variables was based on the following well-defined criteria.
complex relationships and latent constructs within the model. Addi-
tionally, PLS-SEM, often referred to as PLS path modeling, emerges as • Relevance to the measured concept: Indicators were chosen based on
the preferred choice for investigations oriented towards prediction, their direct relevance to measuring the concept in question.
theory development, and cases where the phenomenon under investi- • Empirical support: Indicators with strong empirical backing from
gation is novel or changing (Gupta et al., 2022). Considering the nature previous studies were prioritized to ensure the robustness of the
and goals of this research, which emphasize the exploration of innova- model.
tive concepts rather than the confirmation of established ones, PLS-SEM • Theoretical grounding: Many indicators were selected based on their
stands out as being the most-suited approach for estimating the pro- alignment with established theories such as TAM and UTAUT.
posed model, assessing its reliability, establishing its validity, and • Clarity and measurability: Indicators that could be clearly defined
rigorously testing the formulated hypotheses based on the relationships and measured were preferred to ensure the accuracy of data collec-
established within the research framework. tion and analysis.
The application of this method is facilitated through the utilization of • Comprehensiveness: The aim was to capture as many manifests and
SmartPLS 4 which is a software tool initially used for PLS-SEM (Ringle attributes as possible to represent the characteristics of each
et al., 2024). It is widely used in academic research for its ability to construct comprehensively. A comprehensive approach was impor-
handle complex models with multiple indicators and latent variables tant because the SEM modeling phase serves as a filtering and
(Sarstedt & Cheah, 2019). The current version of the software also in- confirmation step. That is the estimation and validation of the
cludes Covariance-Based SEM (CB-SEM) capabilities, making it a measurement model ensure that indicators not representing their
comprehensive tool for SEM modeling. SmartPLS 4 offers an intuitive constructs (i.e., those that do not load well) are excluded from the
and user-friendly interface that simplifies the process of model creation model. This process also ensures that indicators reliability, conver-
and analysis. It can manage complex models with numerous indicators gent validity and discriminant validity are established.
and latent variables making it suitable for multifaceted research designs.
The software also provides robust statistical techniques such as boot- Therefore, the selection process was systematic and grounded in the
strapping and blindfolding and generates detailed graphical outputs that existing literature ensuring a comprehensive and representative set of
help in the visualization and interpretation of results, and these graph- indicators. Table 1 presents a detailed overview of the various compo-
ical output capabilities are continually improving (Cheah et al., 2024, nents of the framework (factors, variables and indicators) along with the
pp. 97–107). However, despite its user-friendly interface, there is still a references supporting them.
learning curve for users unfamiliar with SEM and the specific features of
6
Table 1
Foundational framework of factors, variables and indicators.
Factors Variables Indicators/Measures Supporting references
Perception Factor Perceived Ease of Use (Bilquise et al., 2023; Huang et al., 2022; Liu & Ma, 2023;
Nozhovnik et al., 2023; Ratten & Jones, 2023; Tiwari et al., 2023)
- User-friendliness of the chatbot interface (User_Friendliness) (Ali et al., 2022; Khalil & Rambech, 2022; Wadhawan et al., 2023)
- Simplicity of the chatbot’s interface (Simplicity) Stupina and Paniotova (2023)
- Ease of navigation within the chatbot (Ease_Navigation) (Rodriguez-Arrastia et al., 2022; Sarosa et al., 2020)
- Clarity of instructions (Clarity_Instructions) Cheung et al. (2023)
- Ability to handle errors and provide clear guidance (Clear_Guidance) (Currie et al., 2023; Dalalah & Dalalah, 2023)
Perceived Usefulness (Bilquise et al., 2023; Contreras & Valette-Florence, 2023; Liu &
Ma, 2023)
- Utility in academic support (Academic_Utility) (Pesonen, 2021; Sallam, 2023; Wadhawan et al., 2023)
- Relevance to individual learning needs (Individual_Relevance) (Ling & Tan, 2020; Murgia et al., 2023; Ruiz-Rojas et al., 2023)
- Enhancement of the overall learning experience (Chan & Hu, 2023; Su & Yang, 2023; Wang, Lund, et al., 2023)
(Learning_Experience)
- Problem solving ability (Problem_Solving) (Chang et al., 2022; Malik et al., 2022)
- Relevance to academic requirements (Academic_Relevance) (Bilquise & Shaalan, 2022; Opara et al., 2023; Wang, Lund, et al.,
2023)
- Enhancing engagement in Learning activities (Z.Lin, 2023; Alshahrani, 2023; Deng & Yu, 2023)
(Learning_Engagement)
Perceived Efficiency (El Azhari, Hilal, Daoudi, & Ajhoun, 2022a; Houston & Corrado,
2023; Wang, Lund, et al., 2023)
- Time savings when using the chatbot (Time_Saving) (Allagui, 2023; Kesarwani & Juneja, 2023; Sallam, 2023)
- Reduction in effort for completing tasks (Effort_Reduction) (Kesarwani & Juneja, 2023; Thosani et al., 2020)
- Promptness of responses from the chatbot (Response_Promptness) (Chand & Sharma, 2023; Motlagh et al., 2023; Schünke et al.,
2022)
- Ability to automate repetitive and time-consuming tasks (El Azhari, Hilal, Daoudi, & Ajhoun, 2022b; Kooli, 2023; Ahmed &
(Task_Automation) Ganapathy, 2021)
Trust and Privacy Trust in AI-Chatbots (Bilquise et al., 2023; Pesonen, 2021; Sharma et al., 2023)
Factor - Confidence in the chatbot’s accuracy (Accuracy_Confidence) (Bilquise et al., 2023; Kohnke et al., 2023)
- Reliability and consistency of chatbot responses (Breeding et al., 2023; Rasul et al., 2023)
(Response_Reliability)
- Transparency of the chatbot about its functioning and decision (Sallam, 2023; Sharma et al., 2023)
processes (Process_Transparency)
- Trust in the chatbot’s handling of user data (Data_Trust) (Sebastian, 2023; Yang et al., 2023)
Privacy Concerns (Baskara, 2023; Wang, Lund, et al., 2023; Yang et al., 2023)
- Comfort level with data sharing (Comfort_Sharing) (Agarwal et al., 2022; Belen-Saglam et al., 2022)
- Trust in data security measures (Trust_Security) (Ofosu-Ampong et al., 2023; Sebastian, 2023; Yang et al., 2023)
- Concerns about potential misuse of user data (Misuse_Concerns) (Grassini, 2023; Piñeiro-Martín et al., 2023)
- Awareness of chatbot’s data collection and usage policies Kasneci et al. (2023)
(Awareness_Policy)
- Level of control over the data shared with the chatbot (Data_Control) Hettiarachchige and Jahankhani (2021)
Interaction and Perceived Interactivity (Rahim et al., 2022), (Jan et al., 2023), (Wang, Wang, et al., 2023),
System Factor - Responsiveness of the chatbot during interactions (Ram & Pratima Verma, 2023; Wamba et al., 2023)
(Interaction_Responsiveness)
- Ability to engage in meaningful dialogue (Meaningful_Dialogue) (Kalla & Smith, 2023; Shahriar & Hayawi, 2023)
- Personalization and adaptability to user preferences (User_ (Huang et al., 2022; Kaiss et al., 2023; Zhai, 2022)
Personalization)
- Ability to provide real-time feedback during tasks (Firat, 2023; Farrokhnia, Banihashem, Noroozi, & Wals, 2024)
(Realtime_Feedback)
- Ability to adapt interactions based on the user’s level of expertise or Roumeliotis and Tselikas (2023)
familiarity with the subject (Adapt_Interactions)
- Ability to handle a wide spectrum of subjects effectively (Aithal & Aithal, 2023; Božić & Poola, 2023)
(Wide_Handling)
System Quality (Aslam & Nisar, 2023; Chu, 2023; Xiao et al., 2023)
- Ability to maintain uninterrupted service (Uninterrupted_Service) Bilquise et al. (2023)
- Overall performance in handling user interactions (Biswas, 2023; Qureshi, 2023)
(Interaction_Performance)
- Compatibility with different web browsers or platforms (Calle et al., 2021; Chand & Sharma, 2023)
(Platform_Compatibility)
Information Quality (Merkouris et al., 2022), (R.Xu, Feng, & Chen, 2023; Chu, 2023)
- Accuracy and correctness of chatbot responses (Response_Accuracy) (Iskandar et al., 2023; Johnson et al., 2023)
- Relevance of information to user queries (Relevance_Query) Kalla and Smith (2023)
- Perceived credibility and trustworthiness of sources (Huschens et al., 2023; Tiwari et al., 2023; Yilmaz et al., 2023)
(Source_Credibility)
- Depth and comprehensiveness of the information provided (Currie et al., 2023; Totlis et al., 2023)
(Information_Depth)
- Ability to provide up-to-date and current information (Crawford et al., 2023; Jeyaraman et al., 2023; Mzwri &
(Current_Information) Turcsányi-Szabo, 2023)
Social and Social Influence (Bilquise et al., 2023; Foroughi et al., 2023; Rukhiran et al., 2022)
Contextual Factor - Influence of peers and teachers on chatbot use (Peer_Influence) (Antony & Ramnath, 2023; Bin-Nashwan et al., 2023;
Ofosu-Ampong et al., 2023)
- Perceived social acceptance of chatbot technology (Conyette, 2023; De Andrés-Sánchez & Gené-Albesa, 2023)
(Social_Acceptance)
- Prevalence of chatbot use within the educational context (Qadir, 2023; Wardat et al., 2023)
(Prevalence_Use)
(continued on next page)
7
Table 1 (continued )
Factors Variables Indicators/Measures Supporting references
- Perception of how professional network views the use of chatbots in Getchell et al. (2022)
education (Professional_View)
Facilitating Conditions (Foroughi et al., 2023; Romero-Rodríguez et al., 2023; Rukhiran
et al., 2022)
- Perception of organizational support for chatbot use (An et al., 2023; Cheong-Trillo, 2023)
(Organization_Support)
- Assessment of available technical infrastructure (Foroughi et al., 2023; Michel-Villarreal et al., 2023)
(Technical_Infrastructure)
- Perception of training and resources for chatbot use (Ifelebuegu, 2023; Kasneci et al., 2023)
(Training_Resources)
- Availability and effectiveness of technical support for using the (Cheong-Trillo, 2023; Yu & Guo, 2023)
chatbot (Techinacl_Support)
- Availability and usefulness of user guides for using the chatbot Foroughi et al. (2023)
(User_Guide)
Accessibility (Alasadi & Baiz, 2023; Babington-Ashaye et al., 2023; Rawas,
2023)
-Equal access for users with diverse needs (Equal_Access) (Gill et al., 2024; Kasneci et al., 2023; Tlili et al., 2023)
-Accommodations for users with disabilities (Accom_Disability) (Martiniello et al., 2021; Mateos-Sanchez et al., 2022)
-Availability on various devices and platforms (Availability) (Calle et al., 2021; Chand & Sharma, 2023)
-Smooth functioning on different web browsers Rajaobelina et al. (2021)
(Smooth_Functioning)
- Ability to provide support in multiple languages (Aguirre et al., 2022; Cox & Tzoc, 2023; Saadna et al., 2022)
(Multilanguage_Support)
Ethical Considerations (Alasadi & Baiz, 2023; Bahroun et al., 2023; Fuchs & Aguilos,
2023)
- Transparency in disclosing chatbot capabilities (Y. Xu, Bradford, & Garg, 2023)
(Transparency_Capability)
- Fair treatment and absence of bias in chatbot interactions (Bias_Free) (Ferrara, 2023; Mohammad et al., 2023; Zhu et al., 2023)
- Responsible handling of user data and ethical practices (Lang et al., 2023; Lundberg, 2023)
(Responsible_Handling)
- Compliance with ethical and regulatory standards related to data (Kasneci et al., 2023; Sallam et al., 2023; Wu et al., 2023)
privacy and use (Ethical_Compliance)
Outcome Variables Adoption of AI-Chatbots (Foroughi et al., 2023; Rajabi et al., 2023; Tiwari et al., 2023)
- Frequency of use (Frequency_Use) (Paul et al., 2023; Strzelecki, 2023)
- Duration of engagement (Duration_Engagement) Aggarwal et al. (2023)
- Consistency of usage (Consistency_Usage) Strzelecki (2023)
- Integration with learning activities (Integration_Learning) (Baidoo-Anu & Ansah, 2023; Zhai, 2022)
- Intent to continue usage (Intent_Continue) (Park & Lee, 2023),
- Degree of reliance on the chatbot for academic support (Kazemitabaar et al., 2023; Ramachandran et al., 2023)
(Degree_Reliance)
User Satisfaction (Alias et al., 2021; Limo et al., 2023; Suárez et al., 2022)
- Satisfaction ratings (Satisfaction_Rating) Han et al. (2023)
- Quality of interactions (Quality_Interaction) Sonkar et al. (2023)
- Net Promoter Score (NPS) (Net_Prom_Score) (Liu et al., 2020; Muzurura et al., 2023)
- Helpfulness of responses (Helpfulness_Response) Følstad and Bjerkreim-Hanssen (2023)
- Perceived empowerment (Empowerment) Tisland et al. (2022)
Learning Outcomes (Alshahrani, 2023; Chan & Hu, 2023; Deng & Yu, 2023)
- Academic performance (Academic_Perf) (Chaiprasurt et al., 2022; Chetradevee et al., 2022)
- Knowledge acquisition (Acquisition_Knowledge) Aleedy et al. (2022)
- Skill development (Skill_Develop) (Li et al., 2023; Elbanna & Armstrong, 2024)
- Perceived improvement in learning effectiveness (Mogavi et al., 2023; Shoufan, 2023)
(Learning_Effectiveness)
- Application of gained knowledge (Application_Knowledge) Khorshidi et al. (2023)
Personalized Learning Experience (Alshahrani, 2023; Fuchs, 2023), (Y.Lin, 2023)
- Tailored Content (Tailored_Content) (Božić & Poola, 2023; Carr, 2023)
- Customized Recommendations (Customized_Recom) Chukwuere (2023)
- Feedback and Progress Tracking (Feedback_Tracking) (Agustini, 2023; Yeşilçınar, 2023)
- Learning Pace (Learning_Pace) (Bin-Hady et al., 2023; Montenegro-Rueda et al., 2023)
- Adaptive Assessments (Adaptive_Assess) (Alipio et al., 2023; Skavronskaya et al., 2023)
4.1. Perceptions factor constructivist learning theory which emphasizes the importance of
user-friendly tools in facilitating active, student-centered learning.
The perceptions factor includes various variables that collectively • Perceived Usefulness: Within this category, the multifaceted utility
form a critical foundation for understanding user attitudes and adoption of chatbots in academic support is explored. This includes their
behavior towards AI-based chatbots in higher education settings. relevance to individual learning needs, the extent to which they
enhance the overall learning experience, their problem-solving ca-
• Perceived Ease of Use: This variable covers the user-friendliness of pabilities, alignment with academic requirements, and their effec-
chatbot interfaces, encompassing key aspects such as the simplicity tiveness in enhancing engagement in various learning activities.
of the interface, ease of navigation, clarity of instructions, and the Studies emphasize that users are more inclined to adopt and persist
chatbot’s proficiency in effectively handling errors. Research sug- in using chatbots when they perceive them as valuable tools in their
gests that the perceived ease of use significantly influences users’ educational journey (Bilquise et al., 2023; Contreras &
willingness to adopt and interact with chatbot technology (Bilquise Valette-Florence, 2023). This is also supported by SDT, which
et al., 2023; Tiwari et al., 2023). This aligns with the principles of
8
highlights the role of tools that enhance competence and engage- interactions, and its compatibility with different browsers and plat-
ment in learning processes. forms. Users expect a seamless and reliable experience when
• Perceived Efficiency: This variable assesses the extent to which engaging with chatbots, and issues related to system quality can
chatbots contribute to enhancing efficiency in the educational pro- significantly impact their perceptions and adoption decisions (Aslam
cess. It encompasses aspects such as time savings when using the & Nisar, 2023). A reliable system also supports connectivism by
chatbot, reduction in user effort required to complete tasks, ensuring continuous access to learning networks and resources.
promptness of responses, and the chatbot’s ability to automate re- • Information Quality: Users assess the accuracy of chatbot responses,
petitive and time-consuming tasks. Research has emphasized the the relevance of information to their queries, and the perceived
significance of perceived efficiency in shaping users’ attitudes and credibility and trustworthiness of information sources. Additionally,
behaviors towards chatbots, as it directly impacts the overall satis- they consider the depth and comprehensiveness of the information
faction (El Azhari et al., 2022b; Wang, Lund, et al., 2023). Cognitive provided and the chatbot’s ability to offer up-to-date and current
load theory supports this by highlighting the importance of reducing information. In essence, information quality strongly influences user
extraneous cognitive load to optimize learning efficiency. satisfaction and their willingness to use chatbots for information-
seeking tasks (Merkouris et al., 2022). High-quality information
4.2. Trust and privacy factor delivery aligns with cognitive load theory by minimizing extraneous
cognitive load and enhancing learning efficiency.
This factor plays a fundamental role in shaping the perceptions of
users and their decisions related to the adoption of AI-based chatbots in
higher education. Building and maintaining trust while addressing pri- 4.4. Social and contextual factor
vacy concerns are essential steps in ensuring the effective integration of
chatbots into educational contexts. The elements of this factor shape the environment in which AI-
chatbots are introduced in higher education. Understanding and
• Trust in AI-based Chatbots: This variable is a cornerstone in the addressing these elements are critical for promoting the successful
acceptance and continued use of AI-based chatbots. It entails integration of chatbots into educational settings.
assessing the user’s trust in the chatbot’s accuracy, reliability, and
consistency in providing responses. Users also consider the trans- • Social Influence: This component involves various aspects of social
parency of the chatbot’s functioning and decision-making processes, influence. It explores how peers and educators influence chatbot use
as well as their overall confidence in how the chatbot handles their among students, the broader social acceptance of chatbot technology
personal data. Research findings suggest that trust in chatbots within the educational context, the prevalence of chatbot utilization
significantly influences users’ willingness to rely on them for various in educational settings, and how professional networks view the
tasks and interactions (Bilquise et al., 2023; Pesonen, 2021). This integration of chatbots in education. Social influence can be instru-
concept is integral to SDT as trust significantly contributes to foster mental in shaping individuals’ attitudes and behaviors towards
intrinsic motivation and engagement. technology acceptance (Bilquise et al., 2023; Foroughi et al., 2023).
• Privacy Concerns: Privacy is a paramount concern in today’s digital Connectivism supports the role of social networks in influencing
age. Within this category, various dimensions of privacy concerns learning and technology adoption.
related to chatbot usage are investigated. This includes users’ com- • Facilitating Conditions: This category encompasses several key ele-
fort levels with data sharing, their trust in the security measures ments. It evaluates users’ perception of organizational support for
implemented to protect their data, concerns about potential misuse chatbot use within their educational institutions. It also considers the
of their data, awareness of data collection policies, and the degree of adequacy of the technical infrastructure to support chatbot func-
control users have over the data they share with the chatbot. Privacy tionality, the availability of training and resources for users, the
concerns can be a significant barrier to the adoption of chatbots, and effectiveness of technical support services, and the availability and
tackling these concerns is essential to fostering user trust and usefulness of user guides or tutorials. The presence of facilitating
acceptance (Pesonen, 2021). conditions is critical to ensuring that users can effectively utilize
chatbots and maximize their benefits (Romero-Rodríguez et al.,
4.3. Interaction and system factor 2023). These conditions align with SDT by ensuring that users have
the support and resources needed to feel competent and autonomous.
The interaction and system elements collectively contribute to users’ • Accessibility: Ensuring that chatbot technology is accessible to a
overall experiences with AI-based chatbots in higher education. diverse user base is of paramount importance. This includes assessing
Ensuring high levels of interactivity, system reliability, and information equal access for users with diverse needs, such as those with dis-
quality is essential for fostering positive user perceptions and encour- abilities, and the availability of accommodations to support their
aging the adoption of chatbot technology. usage. It also involves evaluating the compatibility of chatbots with
various devices and platforms, ensuring their smooth functioning
• Perceived Interactivity: This variable encompasses several critical across different environments, and their ability to provide support in
aspects of user interactions. It includes evaluating how responsive multiple languages. These considerations are essential to the inclu-
chatbots are during conversations, their capacity to engage in sivity of chatbots, ensuring they can accommodate the needs of all
meaningful dialogues, and their ability to tailor responses according potential users (Alasadi & Baiz, 2023). Accessibility considerations
to user preferences and requirements. The provision of real-time are critical to both constructivist learning theory and connectivism,
feedback during tasks and the chatbot’s adaptability to the user’s as it ensures that all students can actively engage with and benefit
level of expertise or familiarity with subjects are also essential from the technology.
components. Enhanced interactivity positively correlates with user • Ethical Considerations: This category includes various ethical di-
satisfaction and adoption intentions (Rahim et al., 2022; Wang, mensions such as the transparency of chatbots in disclosing their
Wang, et al., 2023). The principles of constructivist learning theory capabilities and limitations to users, the fair treatment of users
support the importance of interactivity as it fosters active learning without bias in chatbot interactions, the responsible management of
and engagement. user data, and adherence to ethical and regulatory standards asso-
• System Quality: This includes the chatbot’s ability to maintain un- ciated with data privacy and use. In an era of growing concerns about
interrupted service, its performance in effectively handling user data privacy and fairness, addressing ethical considerations is crucial
9
for building trust and ensuring responsible chatbot adoption (Bah- expectations, and experiences compared to educators, leading to
roun et al., 2023; Fuchs & Aguilos, 2023). varying adoption patterns (Limna et al., 2023).
• Demographics: Variables like gender, age, and educational level can
influence the adoption and perception of technology. Different de-
4.5. Outcome variables
mographic groups may have distinct preferences, needs, and barriers
that influence their acceptance level and usage patterns of AI-
The outcome variables are critical in evaluating the impact of AI-
chatbots (Ragheb et al., 2022).
driven chatbot adoption in the higher education landscape. They pro-
• Previous Experience with AI-Chatbots: Individuals with prior expe-
vide insights on the degree of acceptance of this technology, its effec-
rience using AI and chatbot technologies may have different per-
tiveness, and its overall success in enhancing the learning experience.
ceptions and usage patterns compared to those without prior
exposure. Past experiences can shape attitudes, expectations, and
• Adoption of AI-Chatbots: This variable is measured by various in-
confidence in utilizing these technologies (Essel et al., 2022).
dicators including the frequency of use, the duration of engagement,
• Institution Type: The type of educational institution, such as public
the consistency of usage over time, the degree to which chatbots are
versus private, may influence how AI and chatbots are perceived and
integrated into learning activities, users’ intent to continue using
used. Differences in resources, organizational culture, and infra-
chatbots, and the level of reliance on chatbots for academic support.
structure can impact the adoption and acceptance of these technol-
Prior research has emphasized the significance of adoption as a key
ogies (Rahim et al., 2022).
outcome variable (Foroughi et al., 2023; Tiwari et al., 2023).
• Study Field: The specific academic discipline or field of study can
• User Satisfaction: This variable encompasses several dimensions
also act as a moderating variable impacting the acceptance of AI-
including user satisfaction ratings, the overall quality of interactions,
based chatbots in the context of education. Different fields may
the net promoter score (NPS) that measures user’s loyalty and
have varying degrees of receptivity to these technologies influenced
advocacy, and the extent to which users feel empowered by chatbot
by the discipline’s culture, technological integration level, and other
interactions. High levels of users’ satisfaction are indicative of a
specific educational needs (Wang, King, et al., 2023).
positive user experience and are associated with higher technology
adoption rates (Suárez et al., 2022). User satisfaction is closely tied to
5. Preliminary second order model
SDT which emphasizes the role of competence and relatedness in
enhancing intrinsic motivation and satisfaction.
Higher-order models, alternatively named higher component models
• Learning Outcomes: They include various dimensions such as aca-
(HCM), include multi-order structures with diverse levels of compo-
demic performance, knowledge acquisition, skill development, the
nents. In addressing complex constructs, researchers can conceptualize
perceived improvement in learning effectiveness attributed to chat-
them at higher abstraction levels. In this scenario, the parent construct is
bot use, and the practical application of knowledge gained through
depicted by the corresponding components capturing its characteristics
chatbot interactions. These outcomes offer relevant insights on the
(Hair Jr et al., 2021). This method simplifies the model and enhances its
educational advantages and efficacy of chatbots related to enhancing
parsimony (Cheah et al., 2019). That is, instead of modeling the attri-
student learning experiences (Alshahrani, 2023; Deng & Yu, 2023).
butes as explicit determinants of other variables of the model on a single
Positive learning outcomes support constructivist and cognitive load
construct layer, lower-order constructs (LOC) are consolidated into a
theories which emphasize active learning and efficient cognitive
multidimensional higher-order component (HOC). Researchers typically
processing.
employ the repeated indicators strategy for specifying the HOC’s mea-
• Personalized Learning Experience: Personalization is a key feature of
surement model by associating related indicators with the fundamental
chatbots, and its impact on the learning experience is a crucial
LOCs (Saihi et al., 2023).
outcome variable. This variable examines the extent to which chat-
To construct the preliminary model, the variables and indicators
bots provide tailored content, customized recommendations, feed-
identified in Section 4 are adopted. As shown in Fig. 2, the model en-
back mechanisms for progress tracking, the ability to adjust the
compasses 77 indicators that are reflectively measuring 16 LOCs. Out of
learning pace to individual needs, and the use of adaptive assess-
these, 12 are formatively measuring four HOCs, namely trust and pri-
ments. Personalized learning experiences have proven effective to-
vacy factor, interaction and system factor, social and contextual factor,
wards improving engagement and learning outcomes (Alshahrani,
and perception factor. As described in Table 1, the four HOCs and their
2023). Personalization aligns with SDT by addressing individual
corresponding LOCs are structured as follows.
learning needs and enhancing motivation.
• HOC1 (“Trust and Privacy Factor”): Its corresponding LOCs are

4.6. Moderating variables “Trust in AI-Chatbots” and “Privacy Concerns”.
• HOC2 (“Interaction and System Factor”): Its corresponding LOCs are
There are moderating variables that influence the relationships “Perceived Interactivity”, “System Quality” and “Information
among the components of the first four factors and the outcome vari- Quality”.
ables. These moderating elements should be considered when examining • HOC3 (“Social and Contextual Factor”): Its corresponding LOCs are
the relationships between the influencing variables and the adoption of “Social Influence”, “Facilitating Conditions”, “Accessibility” and
AI-based chatbots in the context of higher education. Accounting for “Ethical Considerations”.
these factors allows the study to capture the variations that exist within • HOC4 (“Perception Factor”): Its corresponding LOCs are “Perceived
different user groups and educational contexts. Ease of Use”, “Perceived Usefulness” and “Perceived Efficiency”.
• Tech-Savviness: The level of comfort, experience, and proficiency The indicators of the HOCs are measured using the repeated indi-
individuals have with technology. Tech-savviness can influence the cator approach. This means each parent HOC is measured using all the
perceptions and use of AI and chatbots, with more tech-savvy in- indicators of its corresponding LOCs. For model parsimony and clarity,
dividuals potentially having higher acceptance and utilization rates the indicators of the HOCs are not duplicated in Fig. 2 as they are already
(Sharma, 2023). presented in the corresponding LOCs. Regarding the arrows linking
• Role (Student/Educator): The perspectives and roles of individuals as variables, all the arrows between HOCs and their LOCs represent
students or educators may impact their perceptions and attitudes formative relationships. The remaining arrows represent relationships of
toward AI and chatbots. Students may have different needs, the path model. The generic model is presented as a Reflective-
10
Fig. 2. Preliminary second order model for AI-chatbot adoption.
Formative (Type II) second-order model and it also depicts the hy- 6.1.1. Evaluation of the reflective measurement model
pothesized relationships among its variables. Reliability and validity tests of the Mode A model are performed as
described in Fig. 3. That is, the initial assessment involves internal
6. Data analysis and discussion of the findings consistency reliability, with Cronbach’s alpha commonly being
employed to estimate this aspect. Another measure of internal consis-
A systematic approach is employed to assess PLS-SEM results, aiming tency, composite reliability (ρc ), is based on the outer loadings of in-
to explain the maximum amount of variation in the endogenous vari- dicators. In accordance with Hair Jr et al. (2021), both metrics have a
ables through the PLS path model. This method involves evaluating the tendency to somewhat underestimate or overestimate the real reliability
effectiveness of both measurement and structural models using specific measures. Therefore, it is advisable that the reliability measure (ρa ) falls
metrics that evaluate their predictive capabilities. Key evaluation between them (lower limit and upper limit). Measures ranging from 0.7
criteria include reliability and validity for measurement models; yet for to 0.9 are deemed satisfactory for these criteria, while in exploratory
the structural model, the focus is on explained variance (R2 ), effect size research, values within the range of 0.6–0.7 are considered acceptable.
(f 2 ), predictive relevance (Q2 ), and the statistical significance of path Subsequently, convergent validity examines how well an indicator
coefficients. The assessment process initiates with an examination of the correlates positively with other indicators belonging to the same
measurement models’ quality (both reflective and formative). After construct. For reflective constructs, the evaluation of convergent val-
confirming the adequacy of the measurement characteristics of the idity considers both the outer loadings of indicators, also known as in-
constructs, attention shifts to examining the estimates of the path model dicator reliability, as well as the average variance extracted (AVE) that
(Hair Jr et al., 2021). indicates the average variability in the elements explained by the latent
component under consideration. It is recommended that indicator
loadings surpass 0.7, with a careful examination of the impact on other
6.1. Evaluation of the measurement models metrics when removing those with loadings between 0.4 and 0.7.
Additionally, the AVE value should exceed 0.5. Finally, discrim-
Two types of measurement models are involved in the proposed inant/divergent validity ensures that measures diverge from those of
model. Firstly, LOCs are measured reflectively via their corresponding conceptually unrelated constructs. This validation indicates that a
indicators which defines a measurement model of reflective type, construct represents a distinct concept not duplicated by others in the
referred to as Mode A. Secondly, four HOCs are formatively measured model. The evaluation involves: (i) Cross-loadings, where an indicator’s
through their corresponding LOCs, thereby involving a formative mea- outer loading with its construct should exceed its correlation with other
surement model (Mode B). Different approaches are adopted by re- constructs; (ii) Fornell-Larcker criterion (FLC), requiring a construct’s
searchers for the aim of assessing the quality of these two types of square root of AVE to exceed its strongest correlation to other compo-
measurement models as summarized in the flowchart of Fig. 3. nents of the model; (iii) Heterotrait-monotrait (HTMT) assessing the
11
Fig. 3. Approaches for measurement models evaluation.
disattenuated correlation among two components, where values above with AVE measures surpassing the recommended threshold. Discrimi-
0.85 suggest a lack of discriminant validity. nant validity is evaluated using the criterion FLC for individual con-
To begin with, the estimation procedure involves an assessment of structs, ensuring that it surpasses its maximum correlation with other
the reflective outer model’s quality. Relevant adjustments to this mea- components, as depicted in Table 4. Additionally, as evidenced by
surement model are iteratively carried out until the characteristics of Table 5, the HTMT ratio further confirms the established discriminant
constructs’ measurements reach adequate levels. The initial model, validity, with values below the 0.85 threshold.
shown in Fig. 1, incorporates the 77 indicators presented in Table 1.
Following the execution of the PLS-SEM algorithm’s factor model and a 6.1.2. Evaluation of the formative measurement model
thorough examination of the obtained results, considering the explor- Research has recently highlighted the increasing use of HCMs in PLS-
atory nature of the research, the decision is made to set the criterion for SEM, driven by their advantages (Sarstedt et al., 2019). However,
removing an indicator at an outer loading threshold of 0.6. Afterwards, confusion arises among researchers regarding HOCs’ specification,
potential indicators for removal are identified, and their elimination’s estimation, and validation. Some neglect proper evaluation of the reli-
impact on other reliability and validity measures is examined and ability of HOCs and their validity or misinterpret their relationships with
summarized in Table 2. Ultimately, six indicators are excluded, resulting the LOCs as path model relationships, where in practice, LOCs are ele-
in a revised model comprising 71 indicators in the initial layer, as ments of HOCs’ measurement models. To address these challenges, re-
depicted in Fig. 4 which also shows the obtained outer loadings in the searchers have proposed various approaches, including the extended
modified measurement model. Evaluation of reliability, as indicated in repeated indicator approach, embedded two-stage approach, disjoint
Table 3, is determined using Cronbach’s alpha and ρc , both exceeding two-stage approach, and improved repeated indicator approach (Cheah
recommended values. The measure ρa , positioned between Cronbach’s et al., 2019). This study adopts the embedded two-stage approach for
alpha and ρc , surpasses 0.7, implying satisfactory reliability. Moreover, HOC estimation and validation. The procedure involves running a
indicators’ reliability is verified and convergent validity is confirmed, standard repeated indicator approach in the initial phase and
Table 2
Impact resulting from the exclusion of indicators.
Construct Indicators to exclude Cronbach’s alpha ρc AVE Impact
Privacy Concerns Misuse_Concerns and Comfort_Sharing 0.679 0.797 0.481 Cronbach alpha, ρc and AVE improved significantly
0.838 0.901 0.752
Social Influence Peer_Influence 0.683 0.804 0.515
0.697 0.832 0.624
Facilitating Conditions Technical_Infrastructure 0.859 0.9 0.648
0.885 0.921 0.745
Perceived Efficiency Response_Promptness 0.791 0.871 0.649
0.901 0.938 0.835
Perceived Ease of Use Simplicity 0.712 0.807 0.510
0.733 0.827 0.548
12
Fig. 4. Updated model for AI-chatbot adoption.
outer weights, illustrating their contribution to forming the construct.

Table 3
The process of evaluating HOCs’ measurement models using the
Evaluating Reliability and convergent validity of the modified model’s
embedded multistage approach begins with the execution of the first
constructs.
stage, wherein the saved LV scores are designated as indicators for the
Construct Cronbach’s alpha AVE
ρa ρc
HOCs. Subsequently, the runs and corresponding tests of stage 2 are
Perceived Ease of Use 0.733 0.759 0.827 0.548 carried out as illustrated in Fig. 5. As presented in Fig. 6 which shows the
Perceived Usefulness 0.929 0.933 0.945 0.741 redundancy analysis tests, the intensity of the path coefficients con-
Perceived Efficiency 0.901 0.901 0.938 0.835
Trust in AI-Chatbots 0.812 0.815 0.877 0.641
necting the constructs to their respective global single-item measures is
Privacy Concerns 0.838 0.868 0.901 0.752 0.848 for "Trust and Privacy Factor", 0.736 for "Interaction and System
Perceived Interactivity 0.859 0.866 0.896 0.590 Factor", 0.853 for "Social and Contextual Factor", and 0.907 for
System Quality 0.695 0.713 0.831 0.622 "Perception Factor", resulting in R2 values of 0.720, 0.542, 0.727, and
Information Quality 0.829 0.831 0.880 0.595
0.823, respectively. Consequently, convergent validity for these HOCs is
Social Influence 0.697 0.713 0.832 0.624
Facilitating Conditions 0.885 0.886 0.921 0.745 established. Moving forward, examination of collinearity among the
Accessibility 0.730 0.732 0.822 0.491 formative indicators is conducted through the VIF criterion, with results
Ethical Considerations 0.775 0.784 0.857 0.603 presented in Table 6. The VIF values remain below the threshold of five,
Learning Outcomes 0.918 0.934 0.939 0.756 indicating the absence of collinearity issues. Subsequently, the signifi-
Personalized Learning Experience 0.903 0.916 0.928 0.720
cance and relevance of the outer weights are analyzed. A bootstrapping
User Satisfaction 0.895 0.900 0.923 0.707
Adoption of AI-chatbots 0.913 0.921 0.933 0.700 procedure, conducted with 5000 samples and a significance level of
0.05, indicates the significance of all formative indicators, as reflected
by the p-values presented in Table 7. Additionally, their outer loadings
incorporating latent variable (LV) scores as additional variables. In are reported alongside the results. Based on these findings, the formative
following stages, the LV scores serve as formative indicators for the HOCs demonstrate satisfactory quality levels, paving the way for the
parent constructs’ measurement models. To validate formative HOCs, path model evaluation.
the three-step approach described in Fig. 3 is followed. First, convergent
validity is evaluated through a redundancy analysis test, linking the
HOC to a distinct global single-item measure, with a path coefficient 6.2. Evaluation of the structural model
exceeding 0.7 supports convergent validity. Then, potential collinearity
issues in LOCs are examined using the Variance Inflation Factor (VIF), After establishing the reliability and validity of the constructs, the
indicating a problem in case the VIF value exceeds 5. Finally, the sig- next steps involve examining the structural model estimates. This
nificance and relevance of formative indicators are evaluated using assessment is crucial for understanding the predictive capabilities of the
model and the relationships among its components. As with measure-
13
Table 4
Assessment of discriminant validity based on FLC.
FLC 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1 Accessibility 0.69
2 Adoption of AI-chatbots 0.58 0.84
3 Ethical Considerations 0.52 0.34 0.78
4 Facilitating Conditions 0.56 0.59 0.60 0.86
5 Information Quality 0.47 0.61 0.65 0.56 0.77
6 Learning Outcomes 0.53 0.83 0.38 0.60 0.62 0.87
7 Perceived Ease of Use 0.39 0.24 0.35 0.22 0.34 0.32 0.71
8 Perceived Efficiency 0.49 0.74 0.32 0.41 0.43 0.61 0.42 0.91
9 Perceived Interactivity 0.48 0.57 0.60 0.68 0.67 0.54 0.49 0.54 0.77
10 Perceived Usefulness 0.66 0.72 0.48 0.43 0.66 0.66 0.49 0.74 0.64 0.86
11 Personalized Learn Exp 0.50 0.68 0.58 0.76 0.66 0.79 0.22 0.44 0.58 0.61 0.85
12 Privacy Concerns 0.35 0.48 0.55 0.70 0.55 0.48 0.12 0.38 0.60 0.45 0.65 0.87
13 Social Influence 0.59 0.73 0.48 0.57 0.51 0.60 0.40 0.64 0.55 0.64 0.50 0.41 0.79
14 System Quality 0.54 0.43 0.36 0.32 0.40 0.34 0.35 0.36 0.44 0.52 0.35 0.36 0.23 0.79
15 Trust in AI-Chatbots 0.45 0.68 0.46 0.62 0.54 0.56 0.13 0.60 0.61 0.61 0.62 0.70 0.52 0.53 0.80
16 User Satisfaction 0.60 0.81 0.25 0.47 0.55 0.84 0.36 0.64 0.50 0.66 0.59 0.33 0.68 0.32 0.54 0.84
Table 5
Discriminant validity based on HTMT ratio.
HTMT 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
2 Adoption of AI-chatbots 0.70

3 Ethical Considerations 0.68 0.40
4 Facilitating Conditions 0.65 0.67 0.72
5 Information Quality 0.60 0.70 0.81 0.68
6 Learning Outcomes 0.65 0.83 0.43 0.69 0.70
7 Perceived Ease of Use 0.49 0.27 0.40 0.30 0.40 0.36
8 Perceived Efficiency 0.66 0.82 0.39 0.47 0.55 0.70 0.69
9 Perceived Interactivity 0.61 0.64 0.72 0.81 0.79 0.61 0.57 0.70
10 Perceived Usefulness 0.80 0.77 0.56 0.45 0.75 0.70 0.51 0.83 0.71
11 Personalized Learn Exp 0.60 0.73 0.69 0.82 0.74 0.84 0.25 0.50 0.65 0.65
12 Privacy Concerns 0.64 0.62 0.82 0.81 0.80 0.62 0.36 0.54 0.81 0.61 0.80
13 Social Influence 0.79 0.83 0.64 0.68 0.65 0.80 0.48 0.81 0.67 0.80 0.64 0.64
14 System Quality 0.76 0.53 0.50 0.43 0.53 0.42 0.46 0.54 0.58 0.64 0.44 0.76 0.41
15 Trust in AI-Chatbots 0.59 0.78 0.58 0.73 0.66 0.66 0.28 0.74 0.75 0.71 0.72 0.82 0.75 0.73
16 User Satisfaction 0.74 0.81 0.34 0.53 0.66 0.83 0.39 0.77 0.57 0.71 0.63 0.48 0.78 0.40 0.65
Fig. 5. Stage 2 runs using the LV scores.
ment models, a systematic approach is applied to evaluate path models evaluated using the coefficient of determination R2 , indicating the
as summarized in Fig. 7. First, collinearity issues are checked using VIF explained variance in endogenous variables by linked exogenous vari-
in order to ensure unbiased coefficients. Path coefficients are then ables. Effect size f 2 quantifies the impact of excluding an exogenous
estimated, with researchers commonly analyzing their statistical sig- variable reflected through the change in R2 . For interpreting effect size
nificance using p-values and bootstrap confidence intervals. The rele- magnitudes, Hair Jr et al. (2021) provide thresholds of 0.02 (small), 0.15
vance of the significant relationships are investigated considering both (medium), and 0.35 (large), with values below 0.02 indicating no effect.
direct and indirect effects. The predictive power of the model is The model’s predictive relevance is evaluated through Stone-Geisser’s
14
Fig. 6. Redundancy analysis tests for convergent validity.
dictor constructs of each subpart of the structural model is conducted

Table 6
using the VIF criterion. The obtained findings indicate the absence of
VIF measures.
collinearity issues in the path model. With this information, the assess-
Indicator VIF Indicator VIF ment of other criteria can be undertaken. The examination of the R2
LV_Accessibility 1.814 <5 LV_Perceived Interactivity 1.938 <5 values for the endogenous constructs aims to determine the extent of
LV_Ethical Considerations 1.711 <5 LV_Perceived Usefulness 2.402 <5 variability explained by the exogenous variables associated with them,
LV_Facilitating Conditions 1.961 <5 LV_Privacy Concerns 1.954 <5 as outlined in Table 8. In accordance with the guidelines established by
LV_Information Quality 1.865 <5 LV_Social Influence 1.771 <5
LV_Perceived Ease of Use 1.328 <5 LV_System Quality 1.267 <5 (Hair Jr et al., 2021), the R2 values for "Perceived Ease of Use" (0.356),
LV_Perceived Efficiency 2.220 <5 LV_Trust in AI-Chatbots 1.954 <5 “Perceived Efficiency” (0.393) and “Personalized Learning Experience”
(0.452) are within the weak to moderate range, the corresponding
values for "Perceived Usefulness" (0.579), “Learning Outcomes” (0.571)
Table 7 and “User Satisfaction” (0.604) fall within the moderate to high range,
Outer weights significance and relevance. while the R2 value for "Adoption of AI-Chatbots" (0.796) is deemed
substantial. Table 8 also includes the effect size matrix and predictive
Original Sample P Outer
sample mean values weights relevance corresponding to the mentioned variables. Following the same
guidelines for f 2 assessment, it is observed that the variables “Trust and
LV_Privacy Concerns→Trust and 0.330 0.317 0.031 0.330
Privacy_Factor Privacy Factor” and “Personalized Learning Experience” do not have
LV_Trust in AI-Chatbots→Trust 0.741 0.745 <0.001 0.742 effects on “Perceived Ease of Use” and “Adoption of AI-Chatbots”,
and Privacy_Factor respectively. The effects of the initial formative constructs, namely
LV_Information 0.640 0.632 <0.001 0.638 “Trust and Privacy Factor”, “Interaction and System Factor” and “Social
Quality→Interaction and
System Factor
and Contextual Factors” on the endogenous variables forming the
LV_Perceived 0.354 0.343 0.017 0.355 “Perception Factor” range from small to medium effects. The effects of
Interactivity→Interaction and the variables “Perception Factor”, “User Satisfaction” and “Learning
System Factor Outcomes” on their respective endogenous variables predominantly
LV_System Quality→Interaction 0.165 0.167 0.042 0.165
and System Factor
exhibit large effects. The Q2 values, all being strictly positive, confirm
LV_Accessibility→Social and 0.262 0.264 0.001 0.261 the endogenous constructs’ predictive relevance. Moreover, with values
Contextual Factor surpassing 0.35, the overall predictive relevance is notably strong.
LV_Ethical 0.140 0.143 0.043 0.140 Lastly, Table 9 presents path coefficients along with corresponding
Considerations→Social and
p-values, indicating the significance of all hypothesized paths except H3
Contextual Factor
LV_Facilitating 0.475 0.479 <0.001 0.475 and H15.
Conditions→Social and
Contextual Factor
LV_Social Influence→Social and 0.533 0.524 <0.001 0.534 6.3. Moderation analysis
Contextual Factor
LV_Perceived Ease of 0.174 0.176 0.022 0.174
Use→Perception Factor Moderation refers to a scenario where the association among two
LV_Perceived 0.396 0.387 <0.001 0.396 components is not constant but relies on a third variable, identified as a
Efficiency→Perception Factor moderator. This third variable can alter the intensity or direction of the
LV_Perceived 0.749 0.753 <0.001 0.749 relationship between the two constructs within the model. These
Usefulness→Perception Factor
moderating relationships are anticipated and explicitly examined by the
researcher (Hair Jr et al., 2021). While the model illustrated in Figs. 2
Q2 , obtained via the blindfolding procedure. Positive Q2 values support and 4 includes only the moderating role of “Tech-Savviness”, “Role” and
the predictive relevance of exogenous constructs for the considered “Study_Field” for schematic simplicity, it is important to highlight that
endogenous variable. Finally, statistical significance of the hypothesized additional hypothesized moderating relationships related to the vari-
relationships is assessed using the bootstrapping method. ables “Gender”, “Age”, “Education_Level” and “Institution_Type” are
To start with, an analysis of potential collinearity issues in the pre- also investigated. It is also important to note that categorical variables
need to undergo dummy coding for inclusion in the PLS path model.
15
Fig. 7. Approach for path model evaluation.
Table 8
Evaluation criteria of the structural model.
Assessment criteria Perceived Ease of Perceived Perceived User Learning Personalized Adoption of AI-
Use Usefulness Efficiency Satisfaction Outcomes Lear_Exp Chatbots
R2 0.356 0.579 0.393 0.604 0.571 0.452 0.796

Q2 predict 0.361 0.550 0.355 0.462 0.452 0.415 0.479
Effect size Trust and Privacy Factor 0.007 (No effect) 0.048 (Small) 0.165 (Medium) NA NA NA NA
f2 Interaction and System 0.194 (Medium) 0.160 0.183 (Medium) NA NA NA NA
Factor (Medium)
Social and Contextual 0.054 (Small) 0.044 (Small) 0.059 (Small) NA NA NA NA
Factor
Perception Factor NA NA NA 0.542 0.491 0.302 0.840 (large)
(large) (large) (Medium)
User Satisfaction NA NA NA NA NA NA 0.480 (large)
Personalized Learning NA NA NA NA NA NA 0.010 (No effect)
Exp
Learning Outcomes NA NA NA NA NA NA 0.375 (large)
Additionally, the primary goal of the research is to assess both moder- tion of AI-chatbots," 0.571 for "Learning Outcomes," 0.452 for "Person-
ating effect and direct effect subject to moderation. In this situation, alized Learning Experience," and 0.604 for "User satisfaction". However,
researchers are encouraged to examine the direct impact in a founda- upon incorporating the interaction terms, the R2 values increased to
tional model that lacks an interaction term and subsequently explore the 0.882, 0.654, 0.568, and 0.623, respectively. This indicates an increase
comprehensive model, incorporating one moderator at a time (Becker, of 8.6%, 8.3%, 11.6%, and 1.9% in the explained variances of the
Cheah, et al., 2023). respective dependent variables. Additionally, significance of the hy-
In PLS-SEM literature, three methods are discussed for constructing pothesized moderating effects was examined, and the summary of the
the interaction term which shows the joint impact of independent and moderation analysis is provided in Table 10. The findings revealed
moderating constructs on the outcome variable. Traditionally, the significant moderating impacts for M2, M4, M6, M7, M8, M13, M16, and
product indicator approach, involving cross-multiplying the indicators M18. Moreover, a slope analysis is included to gain a deeper under-
of the moderating variable with those of the independent construct, was standing of the nature of these significant moderating effects. Only three
common. However, recent research indicates its drawbacks related to interaction plots are depicted in Fig. 8, while the remaining plots can be
statistical power and precision in parameters estimates (Becker et al., found in Fi gure A.1 in the appendix.
2018). Instead, researchers are advised to use the two-stage approach Given that the student role is coded as 1 and the educator role as 0, as
which is adopted in this study. In this approach, scores derived from a illustrated in the initial interaction plot of Fig. 8, the student line ex-
model estimation that initially excludes the interaction term are utilized hibits a steeper slope. This suggests a more pronounced impact of the
in the first stage. These scores then serve as the basis to calculate the "Perception Factor" on "Learning Outcomes" for students compared to
interaction term during the second stage. educators. In the second interaction plot, the line is markedly steeper for
The research investigated the moderating effect of the defined vari- users with high "Tech-Savviness," indicating a stronger impact on
ables on the relationships between the "Perception Factor" variable and “Learning Outcomes” at high levels of technological proficiency
the outcome variables (i.e., "Learning Outcomes," "Personalized Learn- compared to lower levels. However, at lower levels of "Tech-Savviness,"
Exp," "User Satisfaction," and "Adoption of AI-chatbots"). When the increase in the "Perception Factor" does not lead to a similar change
excluding the moderating effects, the R2 values were 0.796 for "Adop- in "Learning Outcomes". Therefore, increased tech proficiency
16
Table 9 Table 10
Path model significance. Significance of moderating effects.
Path STDEV P Relationship Coefficient STDEV P-value
coefficient values
M1: Moderating effect (Role x − 0.035 0.093 0.706
H1: Trust and Privacy Factor → Perceived 0.258 0.108 0.017 Perception_Factor) → Adoption of AI-
Efficiency chatbots
H2: Trust and Privacy Factor → Perceived 0.491 0.120 <0.001 M2: Moderating effect (Role x Perception 0.294 0.110 0.007
Usefulness Factor) → Learning Outcomes
H3: Trust and Privacy Factor → Perceived 0.082 0.080 0.304 M3: Moderating effect (Role x Perception 0.117 0.103 0.257
Ease of Use Factor) → Personalized Learn-Exp
H4: Interaction and System Factor → 0.613 0.116 <0.001 M4: Moderating effect (Role x Perception 0.322 0.112 0.004
Perceived Ease of Use Factor) → User Satisfaction
H5: Interaction and System Factor → 0.171 0.112 0.013 M5: Moderating effect (Tech savviness x 0.004 0.033 0.894
Perceived Efficiency Perception Factor) → Adoption of AI-chatbots
H6: Interaction and System Factor → 0.482 0.094 <0.001 M6: Moderating effect (Tech savviness x − 0.218 0.043 <0.001
Perceived Usefulness Perception_Factor) → Learning Outcomes
H7: Social and Contextual Factor → 0.301 0.103 0.003 M7: Moderating effect (Tech savviness x − 0.232 0.041 <0.001
Perceived Ease of Use Perception Factor) → Personalized Learn-Exp
H8: Social and Contextual Factor → 0.265 0.097 0.006 M8: Moderating effect (Tech savviness x − 0.252 0.042 <0.001
Perceived Efficiency Perception Factor) → User Satisfaction
H9: Social and Contextual Factor → 0.255 0.091 0.005 M9: Moderating effect (Study Field x Perception − 0.040 0.078 0.605
Perceived Usefulness Factor) →Adoption of AI-chatbots
H10: Perception Factor → Adoption of AI- 0.653 0.103 <0.001 M10: Moderating effect (Study Field x 0.136 0.117 0.243
chatbots Perception Factor) → Learning Outcomes
H11: Perception Factor → Learning 0.738 0.094 <0.001 M11: Moderating effect (Study Field x 0.205 0.128 0.168
Outcomes Perception Factor) → Personalized Learn-Exp
H12: Perception Factor → Personalized 0.276 0.067 <0.001 M12: Moderating effect (Study Field x 0.210 0.131 0.108
Learning Experience Perception Factor) → User Satisfaction
H13: Perception Factor → User Satisfaction 0.744 0.072 <0.001 M13: Moderating effect (Gender x Perception 0.233 0.111 0.036
H14: Learning Outcomes → Adoption of AI- 0.490 0.081 <0.001 Factor) → Adoption of AI-chatbots
chatbots M14: Moderating effect (Gender x Perception − 0.202 0.153 0.188
H15: Personalized Learning Experience → 0.030 0.077 0.702 Factor) → Learning Outcomes
Adoption of AI-chatbots M15: Moderating effect (Gender x Perception 0.064 0.135 0.633
H15: User Satisfaction → Adoption of AI- 0.214 0.067 0.001 Factor) → Personalized Learn-Exp
chatbots M16: Moderating effect (Gender x Perception − 0.346 0.148 0.019
Factor) → User Satisfaction
M17: Moderating effect (Age x Perception 0.535 0.315 0.089
strengthens the impact of perception dimensions on learning outcomes Factor) → Adoption of AI-chatbots
M18: Moderating effect (Age x Perception 0.600 0.295 0.042
elements. In the third plot, with the male category coded as 0 and female
Factor) → Learning Outcomes
as 1, the female line is distinctly steeper, indicating a more substantial M19: Moderating effect (Age x Perception 0.268 0.200 0.181
impact of the "Perception Factor" on the "Adoption of AI-Chatbots" for Factor) → Personalized Learn-Exp
female users compared to males. M20: Moderating effect (Age x Perception 0.236 0.231 0.308
Factor) → User Satisfaction
M21: Moderating effect (Education Level x − 0.223 0.224 0.320
6.4. Discussion and implications of the findings Perception Factor) → Adoption of AI-chatbots
M22: Moderating effect (Education Level x − 0.036 0.282 0.899
Perception Factor) → Learning Outcomes
This research presents a thorough exploration into the perceptions of M23: Moderating effect (Education Level x 0.121 0.202 0.550
students and educators towards AI-based chatbots in higher education, Perception Factor) → Personalized Learn-Exp
revealing a multifaceted spectrum of acceptance, concerns, and expec- M24: Moderating effect (Education Level x 0.191 0.285 0.512
tations. The findings provide valuable insights into how various aspects Perception Factor) → User Satisfaction
related to these innovative technologies are perceived and the pivotal

factors influencing their adoption. This section explores the broader trust and perceived competence are crucial for fostering intrinsic moti-
implications of the results and offers interpretations of the main trends vation and engagement.
identified through the research.
6.4.2. Model validation and educational implications
6.4.1. Positive perceptions and emerging concerns Next, the validation of the developed model affirms its relevance and
The research demonstrates a generally positive inclination towards applicability in the context of higher education, particularly regarding
AI-chatbots, emphasizing their potential to revolutionize educational the adoption of AI-based chatbots among students and educators. The
experiences. Both educators and students acknowledge the efficiency, meticulous analysis of data gathered from various respondents within
usefulness, and convenience offered by AI-chatbots. This aligns with this sector demonstrates the internal consistency, reliability, and val-
prior research indicating the transformative role of AI in educational idity of the variables under study. This validation process confirms the
settings (Chan & Hu, 2023; Gill et al., 2024). The model’s path co- significance and relevance of the majority of the hypothesized re-
efficients explicitly support this inclination, with "Perceived Usefulness" lationships and emphasizes the strong predictive relevance of the model
and "Perceived Efficiency" showing strong positive effects on the adop- as established in Sections 6.1 and 6.2. These results indicate that the
tion intention. These findings are consistent with constructivist learning model serves as a robust framework for understanding and evaluating
theory, which emphasizes the importance of tools that enhance the factors shaping the perception of users and influencing the adoption
engagement and facilitate active learning. However, distinct concerns and use of AI-chatbots in educational settings.
pertaining to aspects such as "trust and privacy", "interactivity", and Furthermore, the indicators identified and their corresponding
"information quality" emerged as significant themes. For instance, these weights provide valuable insights for educational administrators and
concerns are reflected in the model through the "Trust and Privacy" policymakers in prioritizing initiatives related to AI-chatbot integration.
factor’s significant influence on the "Perception Factor", highlighting the For instance, substantial impact observed for factors such as "Perceived
critical role of trust in shaping user acceptance. SDT supports this as
17
Fig. 8. Slope analysis – Interaction plots.
Usefulness" and "Perceived Efficiency" suggests that these should be are essential (Chauncey & McKenna, 2023). These guidelines should
focal points in efforts to enhance AI-chatbot adoption in education. Also, address potential biases and ensure that the technology is used to
focusing on developing chatbots that are not only technologically complement human interaction in the education process, in lieu of
advanced but also user-friendly, pedagogically relevant, accessible and replacing it.
available will likely encourage greater acceptance and usage among For technology developers, the findings emphasize the need to focus
students and educators. Constructivist learning theory and cognitive on advanced NLP capabilities, ensuring that chatbots can accurately and
load theory both support the importance of user-friendly and efficient effectively function in academic contexts. The user-centric design should
tools in enhancing learning experiences and reducing cognitive load, be a priority, with chatbots designed to meet the diverse needs and
thus improving overall learning outcomes. Moreover, the findings preferences of distinct users. This involves considering the technical
highlight the significance of trust and privacy aspects, ethical consid- aspects of the chatbots and also their pedagogical relevance and
erations, along with the system’s quality and interactivity. The model adaptability to various learning environments. Furthermore, the
clearly indicates that addressing these factors directly correlates with model’s insights regarding the role of certain variables like tech-
improved perceptions and adoption intentions. Therefore, these aspects savviness, user role and user gender in moderating certain relation-
are crucial towards ensuring a positive user’s experience and in fostering ships within the adoption model highlight the need for a tailored
a supportive environment for the integration of AI technologies in ed- approach in implementing these technologies. Taking into account these
ucation. The right attention to these factors will be essential in aligning differences among potential users is essential in the design and
AI-chatbot capabilities with the evolving needs of modern education, deployment of AI-chatbots to ensure they are accessible and beneficial to
thereby contributing significantly to the enhancement of teaching and all user groups.
learning processes, user satisfaction, and learning outcomes. Finally, the study reveals the need for ongoing research and devel-
opment in the field of AI in education. As AI technology continues to
6.4.3. Unexpected findings on personalized learning evolve, so too must the understanding of its potential impacts, chal-
A surprising revelation from the model is that the personalized lenges, and ethical considerations in educational contexts. The predic-
learning experience does not have a significant impact on technology tive power of the model for critical outcomes such as “Learning
adoption contrary to what might be expected. This finding prompts a Outcomes” and “User Satisfaction” emphasizes the potential of AI
deeper reflection on the current state and future trajectory of AI-chatbot technologies to significantly impact educational experiences which
integration in higher education. It appears that, at this stage, users are warrants further investigation into their long-term effects and optimi-
prioritizing the efficiency and usefulness of AI-chatbots over their zation strategies.
capability to deliver personalized learning experiences. This could be
attributed to the nascent stage of AI technology in education, where 7. Conclusion and future research directions
immediate practical benefits outweigh the perceived value of person-
alization. However, this does not diminish the potential importance of The primary aim of this research was to explore how students and
personalized learning experiences in the long run. As AI technology educators perceive AI-based chatbots in higher education, and to un-
continues to evolve and mature, it is anticipated that personalization derstand the factors driving their adoption. Employing a PLS-SEM
aspects will become increasingly significant in influencing adoption and approach, the study aimed to establish and validate a model that
acceptance. This insight aligns with SDT which suggests that personal- could explain the complex web of relationships influencing the use of
ized support can enhance students’ sense of autonomy and competence these AI tools in academic settings. A rigorous methodology under-
over time, potentially increasing their engagement and satisfaction. pinned the research, beginning with an exhaustive literature review that
informed the development of a foundational framework, a preliminary
6.4.4. Practical implications for implementation and research hierarchical model and a comprehensive questionnaire. This survey in-
The findings of this study have a wide range of implications. For strument, administered to a diverse group of participants within the
educational institutions, there is a clear indication of the need to balance higher education sector, facilitated the collection of rich data from 192
technology integration with the preservation of human-centric educa- respondents. Advanced SEM techniques were then applied to analyze
tional experiences as emphasized by Rane (2023). While chatbots can this data, leading to establishing the model’s reliability and validity,
enhance efficiency and provide personalized support, they should not assessing the significance of the hypothesized relationships, evaluating
replace the invaluable human elements of empathy, mentorship, and the path model, and investigating the potential moderating role of
personal interaction in education. Furthermore, the concerns raised certain variables.
regarding data privacy and security necessitate the establishment of The study’s key findings depict a generally positive attitude towards
clear policies and practices. These measures should ensure the protec- AI-chatbots, emphasizing their potential in enhancing educational ex-
tion of sensitive information and maintain user’s trust in the technology. periences. However, it also pinpointed critical concerns about trust and
Additionally, establishing guidelines for the ethical use of AI-chatbots privacy, as well as the quality and reliability of information, which could
18
impact their wider acceptance. Finally, one of the study’s major con- agencies in the public, commercial, or not-for-profit sectors.
tributions is the establishment of a detailed framework that deepens the
understanding of the determinants of AI-chatbot adoption in educa- Declarations of ethical approval
tional contexts and also provides actionable guidance for educational
leaders, policy-makers, and technology developers. The study was approved by the Institutional Review Board (IRB) of
This study paves the way for new pathways in this rich area of the American University of Sharjah (AUS) with Protocol ID: 24-023.
research. First, an important direction for future works is the under- Informed consent was obtained from all participants, and their privacy
taking of longitudinal research. Such studies would track changes in rights were strictly observed.
perceptions and adoption patterns of AI-chatbots over time and offer
valuable insights into the long-term effects and sustainability of these CRediT authorship contribution statement
technologies in educational environments. Another critical area of
investigation is the impact of AI-chatbots on learning outcomes and Afef Saihi: Writing – review & editing, Writing – original draft,
student performance. This research could provide concrete evidence of Visualization, Validation, Software, Resources, Project administration,
the effectiveness of AI-chatbots as educational tools, highlighting their Methodology, Investigation, Formal analysis, Data curation, Conceptu-
strengths and areas for improvement. Furthermore, the exploration of alization. Mohamed Ben-Daya: Writing – review & editing, Supervi-
ethical concerns and privacy issues linked to the use of AI-chatbots in sion, Project administration, Investigation. Moncer Hariga: Writing –
education remains a paramount area of study. Further research is rec- review & editing, Supervision, Project administration, Investigation.
ommended to explore these aspects, with the aim of establishing Rami As’ad: Writing – review & editing, Project administration.
responsible usage guidelines and practices that protect users’ privacy
while maximizing the benefits of AI technology. Lastly, identifying and Declaration of competing interest
analyzing the barriers to AI-chatbot adoption can highlight some of the
challenges faced by users and institutions. Understanding these barriers The authors declare that they have no known competing financial
is crucial in developing effective strategies towards facilitating wider interests or personal relationships that could have appeared to influence
acceptance and integration of AI technologies in higher education. the work reported in this paper.
Availability of data and materials Acknowledgements
Data and materials supporting the findings of this study are available The authors sincerely thank the respondents to the survey for their
from the corresponding author on reasonable request. contribution to the success of this research. The authors acknowledge
the support of the American University of Sharjah under the Open Ac-
Funding cess Program. This paper represents the opinions of the authors and does
not mean to represent the position or opinions of the American Uni-
This research did not receive any specific grant from funding versity of Sharjah.
List of Acronyms
AI Artificial Intelligence
AIEd AI in Education
AVE Average Variance Extracted
CB-SEM Covariance-Based Structural Equation Modeling
FLC Fornell-Larcker Criterion
GPT Generative Pretrained Transformer
HCM Higher Component Models
HOC Higher-Order Component
HTMT Heterotrait-Monotrait Ratio
IRB Institutional Review Board
LOC Lower-Order Construct
LV Latent Variable
NLP Natural Language Processing
NPS Net Promoter Score
PLS-SEM Partial Least Square Structural Equation Modeling
SDT Self-Determination Theory
SEM Structural Equation Modeling
TAM Technology Acceptance Model
UTAUT Unified Theory of Acceptance and Use of Technology
VIF Variance Inflation Factor
19
Appendix
Fig. A.1. Slope analyses – Interaction plots.
References Alias, S., Sainin, M. S., Fun, T. S., Daut, N., & Sheng, T. L. (2021). Unsupervised text
feature extraction for academic chatbot using constrained FP-growth. ASM Science
Journal, 14, 1–11.
Adams, K. (2023). Babylon health’s CEO ‘should spend some time with elizabeth holmes.
Alipio, M., Lantajo, G. M., & Pregoner, J. D. (2023). On the use of ChatGPT in health
MedCity News.
science education: Opportunities and obstacles. IMCC Journal of Science, 3(2), 1–7.
Adiguzel, T., Kaya, M. H., & Cansu, F. K. (2023). Revolutionizing education with AI:
Allagui, B. (2023). Chatbot feedback on students’ writing: Typology of comments and
Exploring the transformative potential of ChatGPT. Contemporary Educational
effectiveness. Lecture Notes in Computer Science.
Technology, 15(3), ep429. https://doi.org/10.30935/cedtech/13152
Aloqayli, A., & Abdelhafez, H. (2023). Intelligent chatbot for admission in higher
Agarwal, S., Linh, N. T. D., & Aponte, G. J. R. (2022). Student perception regarding
education. International Journal of Information and Education Technology, 13(9),
chatbot for counselling in higher education. Intelligent Systems Reference Library, 215,
1348–1357. https://doi.org/10.18178/ijiet.2023.13.9.1937
263–281. https://doi.org/10.1007/978-3-030-90119-6_22
Alshahrani, A. (2023). The impact of ChatGPT on blended learning: Current trends and
Aggarwal, A., Tam, C. C., Wu, D., Li, X., & Qiao, S. (2023). Artificial intelligence–based
future research directions. International Journal of Data and Network Science, 7(4),
chatbots for promoting health behavioral changes: Systematic review. Journal of
2029–2040.
Medical Internet Research, 25, Article e40789.
An, X., Sing Chai, C., Lia, Y., Zhou, Y., & Yang, B. (2023). Modeling students’ perceptions
Aguirre, M., Serras, M., García-Sardiña, L., López-Fernández, J., Méndez, A., & del
of artificial intelligence assisted language learning. Computer Assisted Language
Pozo, A. (2022). Exploiting in-domain bilingual corpora for zero-shot transfer
Learning, 22(1). https://doi.org/10.1080/09588221.2023.2246519
learning in NLU of intra-sentential code-switching chatbot interactions. Emnlp 2022 -
Antony, S., & Ramnath, R. (2023). A phenomenological exploration of students’
proceedings of the 2022 conference on empirical methods in natural Language Processing:
perceptions of AI chatbots in higher education. IAFOR Journal of Education, 11(2).
Industry track.
Aslam, M. S., & Nisar, S. (2023). Artificial intelligence applications using ChatGPT in
Agustini, N. P. O. (2023). Examining the role of ChatGPT as a learning tool in promoting
education: Case studies and practices: Case studies and practices. IGI Global.
students’ English Language learning autonomy relevant to kurikulum merdeka
Babington-Ashaye, A., de Moerloose, P., Diop, S., & Geissbuhler, A. (2023). Design,
belajar. Edukasia: Jurnal Pendidikan Dan Pembelajaran, 4(2), 921–934.
development and usability of an educational AI chatbot for People with Haemophilia
Ahmad, S. F., Alam, M. M., Rahmat, M. K., Mubarik, M. S., & Hyder, S. I. (2022).
in Senegal. Haemophilia, 29(4), 1063–1073.
Academic and administrative role of artificial intelligence in education.
Bahroun, Z., Anane, C., Ahmed, V., & Zacca, A. (2023). Transforming education: A
Sustainability, 14(3), 1101.
comprehensive review of generative artificial intelligence in educational settings
Ahmed, A. A. A., & Ganapathy, A. (2021). Creation of automated content with embedded
through bibliometric and content analysis. Sustainability, 15(17), 12983. https://doi.
artificial intelligence: A study on learning management system for educational
org/10.3390/su151712983
entrepreneurship. Academy of Entrepreneurship Journal, 27(3), 1–10.
Baidoo-Anu, D., & Ansah, L. O. (2023). Education in the era of generative artificial
Aithal, S., & Aithal, P. (2023). Effects of AI-based ChatGPT on higher education libraries.
intelligence (AI): Understanding the potential benefits of ChatGPT in promoting
International Journal of Management, Technology, and Social Sciences (IJMTS), 8(2),
teaching and learning. Journal of AIDS, 7(1), 52–62.
95–108.
Bao, S., He, H., Wang, F., Wu, H., Wang, H., Wu, W., … Xu, X. (2021). PLATO-2: Towards
Alam, A., & Mohanty, A. (2023). Educational technology: Exploring the convergence of
building an open-domain chatbot via curriculum learning. Findings of the Association
technology and pedagogy through mobility, interactivity, AI, and learning tools.
for Computational Linguistics: ACL-IJCNLP, 2021.
Cogent Engineering, 10(2), Article 2283282.
Baskara, F. R. (2023). Chatbots and flipped learning: Enhancing student engagement and
Alasadi, E. A., & Baiz, C. R. (2023). Generative AI in education and research:
learning outcomes through personalised support and collaboration. IJORER:
Opportunities, concerns, and solutions. Journal of Chemical Education, 100(8),
International Journal of Renewable Energy Resources, 4(2), 223–238.
2965–2971. https://doi.org/10.1021/acs.jchemed.3c00323
Becker, J.-M., Cheah, J.-H., Gholamzade, R., Ringle, C. M., & Sarstedt, M. (2023). PLS-
Aleedy, M., Atwell, E., & Meshoul, S. (2022). Towards deep learning-powered chatbot for
SEM’s most wanted guidance. International Journal of Contemporary Hospitality
translation learning. Lecture Notes in Computer Science.
Management, 35(1), 321–346.
Ali, M. S., Azam, F., Safdar, A., & Anwar, M. W. (2022). Intelligent agents in educational
Becker, B. A., Denny, P., Finnie-Ansley, J., Luxton-Reilly, A., Prather, J., & Santos, E. A.
institutions: NEdBOT - NLP-based chatbot for administrative support using
(2023a). Programming is hard - or at Least it used to Be: Educational opportunities
DialogFlow. In Proceedings - 2022 IEEE international conference on agents (Vol. 2022)
and challenges of AI code generation. Sigcse 2023 - proceedings of the 54th ACM
ICA.
technical symposium on computer science education.
20
Becker, J.-M., Ringle, C. M., & Sarstedt, M. (2018). Estimating moterating effects in PLS- Conyette, M. (2023). Understanding business faculty adoption of technology, its
SEM andPLSc-SEM: Interaction term gerneration* data treatment. Journal of Applied complexities and nuances. Edulearn23 Proceedings.
Structural Equation Modeling, 2, 1–21. Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring
Belda-Medina, J., & Calvo-Ferrer, J. R. (2022). Using chatbots as AI conversational academic integrity in the era of ChatGPT. Innovations in Education & Teaching
partners in language learning. Applied Sciences, 12(17), 8427. International. https://doi.org/10.1080/14703297.2023.2190148
Belen-Saglam, R., Nurse, J. R., & Hodges, D. (2022). An investigation into the sensitivity Cox, C., & Tzoc, E. (2023). ChatGPT Implications for academic libraries. College &
of personal information and implications for disclosure: A UK perspective. Frontiers Research Libraries News, 84(3), 99–102. https://doi.org/10.5860/crln.84.3.99
of Computer Science, 4, Article 908245. Crawford, J., Cowling, M., Ashton-Hay, S., Kelder, J.-A., Middleton, R., & Wilson, G. S.
Bhutoria, A. (2022). Personalized education and artificial intelligence in the United (2023). Artificial intelligence and authorship editor policy: ChatGPT, bard bing AI,
States, China, and India: A systematic review using a human-in-the-loop model. and beyond. Journal of University Teaching and Learning Practice, 20(5), 1.
Computers & Education: Artificial Intelligence, 3, Article 100068. Currie, G., & Barry, K. (2023). ChatGPT in nuclear medicine education. Journal of Nuclear
Bilquise, G., Ibrahim, S., & Salhieh, S. E. M. (2023). Investigating student acceptance of Medicine Technology, 51(3), 247–254. https://doi.org/10.2967/jnmt.123.265844
an academic advising chatbot in higher education institutions. Education and Currie, G., Singh, C., Nelson, T., Nabasenja, C., Al-Hayek, Y., & Spuur, K. (2023).
Information Technologies. https://doi.org/10.1007/s10639-023-12076-x ChatGPT in medical imaging higher education. Radiography, 29(4), 792–799.
Bilquise, G., & Shaalan, K. (2022). AI-based academic advising framework: A knowledge https://doi.org/10.1016/j.radi.2023.05.011
management perspective. International Journal of Advanced Computer Science and Dalalah, D., & Dalalah, O. M. A. (2023). The false positives and false negatives of
Applications, 13(8). generative AI detection tools in education and academic research: The case of
Bin-Hady, W. R. A., Al-Kadi, A., Hazaea, A., & Ali, J. K. M. (2023). Exploring the ChatGPT. International Journal of Management in Education, 21(2), Article 100822.
dimensions of ChatGPT in English Language learning: A global perspective. Library https://doi.org/10.1016/j.ijme.2023.100822
Hi Tech. https://doi.org/10.1108/LHT-05-2023-0200 Davis, F. D. (1989). Technology acceptance model: TAM. Al-suqri, MN, Al-aufi, AS:
Bin-Nashwan, S. A., Sadallah, M., & Bouteraa, M. (2023). Use of ChatGPT in academia: Information seeking Behavior and technology adoption.
Academic integrity hangs in the balance. Technology in Society, Article 102370. De Andrés-Sánchez, J., & Gené-Albesa, J. (2023). Explaining policyholders’ chatbot
Biswas, S. (2023). Evaluating errors and improving performance of ChatGPT. Int. J. Clin. acceptance with an unified technology acceptance and use of technology-based
Med. Educ. Res., 2(6), 182–188. model. Journal of Theoretical and Applied Electronic Commerce Research, 18(3),
Božić, V., & Poola, I. (2023). Chat GPT and education. Preprint. 1217–1237.
Breeding, T., Martinez, B., Patel, H., Nasef, H., Arif, H., Nakayama, D., & Elkbuli, A. Deng, X., & Yu, Z. (2023). A meta-analysis and systematic review of the effect of chatbot
(2023). The utilization of ChatGPT in reshaping future medical education and technology use in sustainable education. Sustainability, 15(4), 2940. https://doi.org/
learning perspectives: A curse or a blessing? The American Surgeon. https://doi.org/ 10.3390/su15042940
10.1177/00031348231180950 Dhanasekar, V., Preethi, Y., Vishali, S., Praveen Joe, I. R., & Booma Poolan, M. (2021).
Calle, M., Narváez, E., & Maldonado-Mahauad, J. (2021). Proposal for the design and A chatbot to promote students mental health through emotion recognition. In
implementation of miranda: A chatbot-type recommender for supporting self- Proceedings of the 3rd international conference on inventive research in computing
regulated learning in online environments. CEUR Workshop Proceedings. applications (Vol. 2021)ICIRCA.
Carr, B. (2023). Revolutionizing education: Unleashing the power of chat GPT/AI to Díaz-Rodríguez, N., Del Ser, J., Coeckelbergh, M., de Prado, M. L., Herrera-Viedma, E., &
empower educators. Technology and the Curriculum: Summer, 2023. Herrera, F. (2023). Connecting the dots in trustworthy Artificial Intelligence: From
Chaiprasurt, C., Amornchewin, R., & Kunpitak, P. (2022). Using motivation to improve AI principles, ethics, and key requirements to responsible AI systems and regulation.
learning achievement with a chatbot in blended learning. World Journal on Information Fusion, Article 101896.
Educational Technology: Current Issues, 14(4), 1133–1151. https://doi.org/10.18844/ El Azhari, K., Hilal, I., Daoudi, N., & Ajhoun, R. (2022a). “AskBot”: The AI chatbot that
wjet.v14i4.6592 enhances the learning process. In International conference on digital technologies and
Chan, C. K. Y., & Hu, W. (2023). Students’ voices on generative AI: Perceptions, benefits, applications (pp. 388–396).
and challenges in higher education. International Journal of Educational Technology in El Azhari, K., Hilal, I., Daoudi, N., & Ajhoun, R. (2022b). Smart learning using
Higher Education, 20(1), 43. https://doi.org/10.1186/s41239-023-00411-8 autonomous chatbot based on NLP techniques. In The international conference on
Chand, R. R., & Sharma, N. A. (2023). Development of bilingual chatbot for university artificial intelligence and smart environment (pp. 723–728).
related FAQs using natural Language Processing and deep learning. Lecture Notes in Elbanna, S., & Armstrong, L. (2024). Exploring the integration of ChatGPT in education:
Computer Science. Adapting for the future. Management & Sustainability: An Arab Review, 3(1), 16–29.
Chang, C. Y., Hwang, G. J., & Gau, M. L. (2022). Promoting students’ learning Essel, H. B., Vlachopoulos, D., Tachie-Menson, A., Johnson, E. E., & Baah, P. K. (2022).
achievement and self-efficacy: A mobile chatbot approach for nursing training. The impact of a virtual teaching assistant (chatbot) on students’ learning in
British Journal of Educational Technology, 53(1), 171–188. https://doi.org/10.1111/ Ghanaian higher education. International Journal of Educational Technology in Higher
bjet.13158 Education, 19(1), 57. https://doi.org/10.1186/s41239-022-00362-6
Chang, D. H., Lin, M. P. C., Hajian, S., & Wang, Q. Q. (2023). Educational design Farrokhnia, M., Banihashem, S. K., Noroozi, O., & Wals, A. (2024). A SWOT analysis of
principles of using AI chatbot that supports self-regulated learning in education: Goal ChatGPT: Implications for educational practice and research. Innovations in Education
setting, feedback, and personalization. Sustainability, 15(17), 12921. https://doi. & Teaching International, 61(3), 460–474. https://doi.org/10.1080/
org/10.3390/su151712921 14703297.2023.2195846
Chauncey, S. A., & McKenna, H. P. (2023). A framework and exemplars for ethical and Ferrara, E. (2023). Fairness and bias in artificial intelligence: A brief survey of sources,
responsible use of AI Chatbot technology to support teaching and learning. impacts, and mitigation strategies. arXiv preprint arXiv:2304.07683.
Computers & Education: Artificial Intelligence, 5, Article 100182. Firat, M. (2023). What ChatGPT means for universities: Perceptions of scholars and
Cheah, J. H., Magno, F., & Cassia, F. (2024). Reviewing the SmartPLS 4 software: The latest students. Journal of Applied Learning and Teaching, 6(1), 57–63. https://doi.org/
features and enhancements. 10.37074/jalt.2023.6.1.22
Cheah, J.-H., Ting, H., Ramayah, T., Memon, M. A., Cham, T.-H., & Ciavolino, E. (2019). Fitria, T. N. (2023). Artificial intelligence (AI) technology in OpenAI ChatGPT
A comparison of five reflective–formative estimation approaches: Reconsideration application: A review of ChatGPT in writing English essay. ELT Forum: Journal of
and recommendations for tourism research. Quality and Quantity, 53(3), 1421–1458. English Language Teaching, 12(1), 44–58.
Chen, Y., Jensen, S., Albert, L. J., Gupta, S., & Lee, T. (2023). Artificial intelligence (AI) Følstad, A., & Bjerkreim-Hanssen, N. (2023). User interactions with a municipality
student assistants in the classroom: Designing chatbots to support student success. chatbot—lessons learnt from dialogue analysis. International Journal of Human-
Information Systems Frontiers, 25(1), 161–182. Computer Interaction, 1–14.
Chen, X., Xie, H., & Hwang, G. J. (2020). A multi-perspective study on artificial Foroughi, B., Senali, M. G., Iranmanesh, M., Khanfar, A., Ghobakhloo, M., Annamalai, N.,
intelligence in education: Grants, conferences, journals, software tools, institutions, & Naghmeh-Abbaspour, B. (2023). Determinants of intention to use ChatGPT for
and researchers. Computers & Education: Artificial Intelligence, 1, Article 100005. educational purposes: Findings from PLS-SEM and fsQCA. International Journal of
Cheong-Trillo, N. (2023). Benefits and Barriers of chatbot Use in education. Technology and Human-Computer Interaction, 1–20.
the curriculum: Summer 2023. Fuchs, K. (2023). Exploring the opportunities and challenges of NLP models in higher
Chetradevee, S. L., Anushka Xavier, K., & Jayapandian, N. (2022). Artificial intelligence education: Is chat GPT a blessing or a curse? Frontiers in Education, 8, Article
technological revolution in education and space for next generation. Lecture Notes in 1166682.
Networks and Systems. Fuchs, K., & Aguilos, V. (2023). Integrating artificial intelligence in higher education:
Cheung, B. H. H., Lau, G. K. K., Wong, G. T. C., Lee, E. Y. P., Kulkarni, D., Seow, C. S., … Empirical insights from students about using ChatGPT. International Journal of
Co, M. T. H. (2023). ChatGPT versus human in generating medical graduate exam Information and Education Technology, 13(9), 1365–1371. https://doi.org/10.18178/
multiple choice questions—a multinational prospective study (Hong Kong S. A.R., ijiet.2023.13.9.1939
Singapore, Ireland, and the United Kingdom). PLoS One, 18(8 August), Article Getchell, K. M., Carradini, S., Cardon, P. W., Fleischmann, C., Ma, H., Aritz, J., &
e0290691. https://doi.org/10.1371/journal.pone.0290691 Stapp, J. (2022). Artificial intelligence in business communication: The changing
Chu, M.-n. (2023). Assessing the benefits of ChatGPT for business: An empirical study on landscape of research and teaching. Business and Professional Communication
organizational performance. IEEE Access. https://doi.org/10.1109/ Quarterly, 85(1), 7–33.
ACCESS.2023.3297447 Gill, S. S., Xu, M., Patros, P., Wu, H., Kaur, R., Kaur, K., … Buyya, R. (2024).
Chukwuere, J. E. (2023). ChatGPT: The game changer for higher education institutions. Transformative effects of ChatGPT on modern education: Emerging era of AI
Jozac Academic Voice, 3, 22–27. chatbots. Internet of Things and Cyber-Physical Systems, 4, 19–23.
Contreras, C. R., & Valette-Florence, P. (2023). Toward a theoretical model of branded Grassini, S. (2023). Shaping the future of education: Exploring the potential and
chatbot adoption, a bibliometric and machine learning perspective: An abstract. In consequences of AI and ChatGPT in educational settings. Education Sciences, 13(7),
Developments in marketing science: Proceedings of the academy of marketing science (pp. 692.
181–182).
21
Guo, K., Zhong, Y., Li, D., & Chu, S. K. W. (2023). Effects of chatbot-assisted in-class Journal of Emerging Technologies in Learning, 18(14), 188–199. https://doi.org/
debates on students’ argumentation skills and task motivation. Computers & 10.3991/ijet.v18i14.40935
Education, 203, Article 104862. Kim, M., & Adlof, L. (2024). Adapting to the future: ChatGPT as a means for supporting
Gupta, S., Bag, S., Modgil, S., de Sousa Jabbour, A. B. L., & Kumar, A. (2022). Examining constructivist learning environments. TechTrends, 68(1), 37–46.
the influence of big data analytics and additive manufacturing on supply chain risk Kiraz, A., Canpolat, O., Özkurt, C., & Taşkın, H. (2020). Analysis of the factors affecting
control and resilience: An empirical study. Computers & Industrial Engineering, 172, the Industry 4.0 tendency with the structural equation model and an application.
Article 108629. Computers & Industrial Engineering, 150, Article 106911.
Hair Jr, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2021). A primer on partial least Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). Exploring generative artificial
squares structural equation modeling (PLS-SEM). Sage publications. intelligence preparedness among university language instructors: A case study.
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011). PLS-SEM: Indeed a silver bullet. Journal Computers & Education: Artificial Intelligence, 5, Article 100156. https://doi.org/
of Marketing Theory and Practice, 19(2), 139–152. 10.1016/j.caeai.2023.100156
Han, J., Yoo, H., Myung, J., Kim, M., Lee, T. Y., Ahn, S.-Y., & Oh, A. (2023). ChEDDAR: Kooli, C. (2023). Chatbots in education and research: A critical examination of ethical
Student-ChatGPT dialogue in EFL writing education. arXiv preprint arXiv:2309.13243. implications and solutions. Sustainability, 15(7), 5614.
Hettiarachchige, H. H., & Jahankhani, H. (2021). Holistic authentication framework for Kurni, M., Mohammed, M. S., & Srinivasa, K. (2023). Chatbots for education. In
virtual agents; UK banking industry. In Challenges in the IoT and smart environments: A A beginner’s guide to introduce artificial intelligence in teaching and learning (pp.
practitioners’ guide to security, ethics and criminal threats (pp. 245–286). Springer. 173–198). Springer.
Hew, K. F., Huang, W., Du, J., & Jia, C. (2023). Using chatbots to support student goal Lang, R., McLay, L., & Rispoli, M. (2023). Advanced language models: Potential to
setting and social presence in fully online activities: Learner engagement and improve augmentative and alternative communication for individuals with
perceptions. Journal of Computing in Higher Education, 35(1), 40–68. intellectual and developmental disabilities. Advances in Neurodevelopmental
Houston, A. B., & Corrado, E. M. (2023). Embracing ChatGPT: Implications of emergent Disorders, 1–4.
language models for academia and libraries. Technical Services Quarterly, 40(2), Le Nguyen, T., & Do, T. T. H. (2019). Artificial intelligence in healthcare: A new
76–91. https://doi.org/10.1080/07317131.2023.2187110 technology benefit for both patients and doctors. In 2019 portland international
Huang, W., Hew, K. F., & Fryer, L. K. (2022). Chatbots for language learning—are they conference on management of engineering and technology (PICMET) (pp. 1–15). IEEE.
really useful? A systematic review of chatbot-supported language learning. Journal of Lee, S. B. (2020). Chatbots and communication: The growing role of artificial intelligence
Computer Assisted Learning, 38(1), 237–257. https://doi.org/10.1111/jcal.12610 in addressing and shaping customer needs. Business Communication Research and
Huschens, M., Briesch, M., Sobania, D., & Rothlauf, F. (2023). Do you trust ChatGPT?– Practice, 3(2), 103–111.
perceived credibility of human and AI-generated content. arXiv preprint arXiv: Li, L., Ma, Z., Fan, L., Lee, S., Yu, H., & Hemphill, L. (2023). ChatGPT in education: A
2309.02524. discourse analysis of worries and concerns on social media. arXiv preprint arXiv:
Hwang, G.-J., & Chang, C.-Y. (2023). A review of opportunities and challenges of 2305.02201.
chatbots in education. Interactive Learning Environments, 31(7), 4099–4112. Liesa-Orús, M., Latorre-Cosculluela, C., Sierra-Sánchez, V., & Vázquez-Toledo, S. (2023).
Ifelebuegu, A. O. (2023). Rethinking online assessment strategies: Authenticity versus AI Links between ease of use, perceived usefulness and attitudes towards technology in
chatbot intervention. Journal of Applied Learning and Teaching, 6(2). older people in university: A structural equation modelling approach. Education and
Ifelebuegu, A. O., Kulume, P., & Cherukut, P. (2023). Chatbots and AI in Education Information Technologies, 28(3), 2419–2436.
(AIEd) tools: The good, the bad, and the ugly. Journal of Applied Learning and Lim, M. S., Ho, S. B., & Chai, I. (2021). Design and functionality of a university academic
Teaching, 6(2). advisor chatbot as an early intervention to improve students’ academic performance.
Ilieva, G., Yankova, T., Klisarova-Belcheva, S., Dimitrov, A., Bratkov, M., & Angelov, D. In Computational science and technology: 7th ICCST 2020, pattaya, Thailand, 29–30
(2023). Effects of generative chatbots in higher education. Information, 14(9), 492. august, 2020 (pp. 167–178). Singapore: Springer Singapore.
İpek, Z. H., Gözüm, A.İ. C., Papadakis, S., & Kallogiannakis, M. (2023). Educational Limna, P., Kraiwanit, T., Jangjarat, K., Klayklung, P., & Chocksathaporn, P. (2023). The
applications of the ChatGPT AI system: A systematic review research. Educational use of ChatGPT in the digital era: Perspectives on chatbot implementation. Journal of
Process: International Journal, 12(3), 26–55. Applied Learning and Teaching, 6(1).
Iskandar, A., Kartowagiran, B., Amiruddin, E. G., & Said, A. (2023). Implementation of Limo, F. A. F., Tiza, D. R. H., Roque, M. M., Herrera, E. E., Murillo, J. P. M., Huallpa, J. J.,
natural Language Processing and machine learning in the chatbot customer service … Gonzáles, J. L. A. (2023). Personalized tutoring: ChatGPT as a virtual tutor for
application for new student admission. AIP Conference Proceedings. personalized learning experiences. Przestrzen Spoleczna, 23(1), 292–312.
Jain, V., Singh, I., Syed, M., Mondal, S., & Palai, D. R. (2024). Enhancing educational Lin, Y. (2023). The impact of CHAT GPT on English online courses. Advances in
interactions: A comprehensive review of AI chatbots in learning environments. In Educational Technology and Psychology, 7(10), 55–59.
2024 11th international conference on reliability, infocom technologies and optimization Lin, Z. (2023). Why and how to embrace AI such as ChatGPT in your academic life. Royal
(trends and future directions)(ICRITO) (pp. 1–5). IEEE. Society Open Science, 10(8), Article 230658. https://doi.org/10.1098/rsos.230658
Jan, I. U., Ji, S., & Kim, C. (2023). What (de) motivates customers to use AI-powered Ling, L., & Tan, C. W. (2020). Social learning network and its applications in large scale
conversational agents for shopping? The extended behavioral reasoning perspective. online education through chatbot. In Online social networks: Perspectives, applications
Journal of Retailing and Consumer Services, 75, Article 103440. and developments (pp. 165–183).
Jeyaraman, M., Jeyaraman, N., Nallakumarasamy, A., Yadav, S., & Bondili, S. K. (2023). Liu, Q., Huang, J., Wu, L., Zhu, K., & Ba, S. (2020). Cbet: Design and evaluation of a
ChatGPT in medical education and research: A boon or a bane? Cureus, 15(8). domain-specific chatbot for mobile learning. Universal Access in the Information
Johnson, D., Goodman, R., Patrinely, J., Stone, C., Zimmerman, E., Donald, R., , … Society, 19(3), 655–673. https://doi.org/10.1007/s10209-019-00666-x
Choe, J., et al. (2023). Assessing the accuracy and reliability of AI-generated medical Liu, G., & Ma, C. (2023). Measuring EFL learners’ use of ChatGPT in informal digital
responses: An evaluation of the chat-GPT model. Research Square. learning of English based on the technology acceptance model. Innovation in
Kaiss, W., Mansouri, K., & Poirier, F. (2023). Effectiveness of an adaptive learning Language Learning and Teaching, 1–14.
chatbot on students’ learning outcomes based on learning styles. International Lundberg, F. (2023). AI and business decision making: Striking a balance between
Journal of Emerging Technologies in Learning, 18(13), 250–261. https://doi.org/ efficiency and responsibility. In Karlstad business school.
10.3991/ijet.v18i13.39329 Malik, S. I., Ashfque, M. W., Tawafak, R. M., Al-Farsi, G., Usmani, N. A., &
Kalla, D., & Smith, N. (2023). Study and analysis of chat GPT and its impact on different Khudayer, B. H. (2022). A chatbot to facilitate student learning in a programming 1
fields of study. International Journal of Innovative Science and Research Technology, 8 course: A gendered analysis. International Journal of Virtual and Personal Learning
(3). Environments, 12(1). https://doi.org/10.4018/IJVPLE.310007
Kamalov, F., Santandreu Calonge, D., & Gurrib, I. (2023). New era of Artificial Martiniello, N., Asuncion, J., Fichten, C., Jorgensen, M., Havel, A., Harvison, M., …
intelligence in education: Towards a sustainable multifaceted revolution. Vo, C. (2021). Artificial intelligence for students in postsecondary education: A
Sustainability, 15(16), Article 12451. world of opportunity. AI Matters, 6(3), 17–29.
Karyotaki, M., Drigas, A., & Skianis, C. (2022). Chatbots as cognitive, educational, Mateos-Sanchez, M., Melo, A. C., Blanco, L. S., & García, A. M. F. (2022). Chatbot, as
advisory & coaching systems. Technium Social Sciences Journal, 30, 109–126. educational and inclusive tool for people with intellectual disabilities. Sustainability,
Kasneci, E., Seßler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., … 14(3), 1520.
Hüllermeier, E. (2023). ChatGPT for good? On opportunities and challenges of large Mello, R. F., Freitas, E., Pereira, F. D., Cabral, L., Tedesco, P., & Ramalho, G. (2023).
language models for education. Learning and Individual Differences, 103, Article Education in the age of generative AI: Context and recent developments. arXiv preprint
102274. arXiv:2309.12332.
Kazemitabaar, M., Chow, J., Ma, C. K. T., Ericson, B. J., Weintrop, D., & Grossman, T. Merkouris, S. S., Loram, G., Abdelrazek, M., Rodda, S. N., Ibrahim, A., Bonti, A., &
(2023). Studying the effect of AI code generators on supporting novice learners in Dowling, N. A. (2022). Improving the user experience of a gambling support and
introductory programming. In Proceedings of the 2023 CHI conference on human education website using a chatbot. Universal access in the information society. https://
factors in computing systems (pp. 1–23). doi.org/10.1007/s10209-022-00932-5
Kesarwani, S., & Juneja, S. (2023). Student chatbot system: A review on educational Michel-Villarreal, R., Vilalta-Perdomo, E., Salinas-Navarro, D. E., Thierry-Aguilera, R., &
chatbot. In 2023 7th international conference on trends in electronics and informatics Gerardou, F. S. (2023). Challenges and opportunities of generative AI for higher
(ICOEI) (pp. 1578–1583). IEEE. education as explained by ChatGPT. Education Sciences, 9(13).
Khalil, M., & Rambech, M. (2022). Eduino: A telegram learning-based platform and Mogavi, R. H., Deng, C., Kim, J. J., Zhou, P., Kwon, Y. D., Metwally, A. H. S., … Gujar, S.
chatbot in higher education. Lecture Notes in Computer Science. (2023). Exploring user perspectives on chatgpt: Applications, perceptions, and implications
Khorshidi, H., Mohammadi, A., Yousem, D. M., Abolghasemi, J., Ansari, G., Mirza- for ai-integrated education. arXiv preprint arXiv:2305.13114.
Aghazadeh-Attari, M., … Ardakani, A. A. (2023). Application of ChatGPT in Mohammad, B., Supti, T., Alzubaidi, M., Shah, H., Alam, T., Shah, Z., & Househ, M.
multilingual medical education: How does ChatGPT fare in 2023’s Iranian residency (2023). The pros and cons of using ChatGPT in medical education: A scoping review.
entrance examination. Informatics in Medicine Unlocked, 41, Article 101314. Studies in Health Technology and Informatics, 305, 644–647.
Khurma, O. A., Ali, N., & Hashem, R. (2023). Critical reflections on ChatGPT in UAE
education navigating equity and governance for safe and effective use. International
22
Monserrat, M., Mas, A., Mesquida, A. L., & Clarke, P. (2022). Investigating the use of Rasul, T., Nair, S., Kalendra, D., Robin, M., Santini, F. O., Ladeira, W. J., … Heathcote, L.
artificial intelligence (AI) in educational settings: A systematic review. In European (2023). The role of ChatGPT in higher education: Benefits, challenges, and future
conference on software process improvement (pp. 3–17). research directions. Journal of Applied Learning and Teaching, 6(1), 41–56. https://
Montenegro-Rueda, M., Fernández-Cerero, J., Fernández-Batanero, J. M., & López- doi.org/10.37074/jalt.2023.6.1.29
Meneses, E. (2023). Impact of the implementation of ChatGPT in education: A Ratten, V., & Jones, P. (2023). Generative artificial intelligence (ChatGPT): Implications
systematic review. Computers, 12(8), 153. for management educators. International Journal of Management in Education, 21(3),
Mori, M., & Du, L. (2023). AI chatbots for banks: Evolving trends and critical issues. In Article 100857. https://doi.org/10.1016/j.ijme.2023.100857
The international conference on artificial intelligence and logistics engineering (pp. 3–13). Rawas, S. (2023). ChatGPT: Empowering lifelong learning in the digital age of higher
Motlagh, N. Y., Khajavi, M., Sharifi, A., & Ahmadi, M. (2023). The impact of artificial education. Education and Information Technologies, 1–14. https://doi.org/10.1007/
intelligence on the evolution of digital education: A comparative study of OpenAI text s10639-023-12114-8
generation tools including ChatGPT, bing chat, bard, and ernie. arXiv preprint arXiv: Ray, P. P. (2023). ChatGPT: A comprehensive review on background, applications, key
2309.02029. challenges, bias, ethics, limitations and future scope. Internet of Things and Cyber-
Muñoz, S. A. S., Gayoso, G. G., Huambo, A. C., Tapia, R. D. C., Incaluque, J. L., Physical Systems, 3, 121–154.
Aguila, O. E. P., … Arias-Gonzáles, J. L. (2023). Examining the impacts of ChatGPT Ringle, C. M., Wende, S., Becker, & Jan-Michael. (2024). SmartPLS 4. Bönningstedt:
on student motivation and engagement. Przestrzen Spoleczna, 23(1), 1–27. SmartPLS. Retrieved from https://www.smartpls.com.
Murgia, E., Pera, M. S., Landoni, M., & Huibers, T. (2023). Children on ChatGPT Robayo-Pinzon, O., Rojas-Berrio, S., Rincon-Novoa, J., & Ramirez-Barrera, A. (2023).
readability in an educational context: Myth or opportunity?. In Adjunct proceedings of Artificial intelligence and the value Co-creation process in higher education
the 31st ACM conference on user modeling, adaptation and personalization (pp. institutions. International Journal of Human-Computer Interaction, 1–17.
311–316). Robles Urquijo, I., Quintela Incera, A., Van Vaerenbergh, S., Inaud, D., & Lopez
Muzurura, O., Mzikamwi, T., Rebanowako, T. G., & Mpini, D. (2023). Application of Higuera, J. M. (2019). Risks and opportunities of using fibre optic sensors for long
artificial intelligence for virtual teaching assistance (case study: Introduction to term infrastructure health monitoring systems in an 18 Year old installation. In
information technology). International Research Journal of of Engineering and International conference on smart infrastructure and construction 2019 (ICSIC) driving
Technology (IRJET), 10(9). data-informed decision-making (pp. 623–630). ICE Publishing.
Mzwri, K., & Turcsányi-Szabo, M. (2023). Internet wizard for enhancing open-domain Rodriguez-Arrastia, M., Martinez-Ortigosa, A., Ruiz-Gonzalez, C., Ropero-Padilla, C.,
question-answering chatbot knowledge base in education. Applied Sciences, 13(14), Roman, P., & Sanchez-Labraca, N. (2022). Experiences and perceptions of final-year
8114. https://doi.org/10.3390/app13148114 nursing students of using a chatbot in a simulated emergency situation: A qualitative
Naidu, K., & Sevnarayan, K. (2023). ChatGPT: An ever-increasing encroachment of study. Journal of Nursing Management, 30(8), 3874–3884.
artificial intelligence in online assessment in distance education. The Online Journal Romero-Rodríguez, J. M., Ramírez-Montoya, M. S., Buenestado-Fernández, M., & Lara-
of Communication and Media Technologies, 13(3), Article e202336. https://doi.org/ Lara, F. (2023). Use of ChatGPT at university as a tool for complex thinking:
10.30935/ojcmt/13291 Students’ perceived usefulness. Cultura de los Cuidados, 12(2), 323–339. https://doi.
Nozhovnik, O., Harbuza, T., Teslenko, N., Okhrimenko, O., Zalizniuk, V., & Durdas, A. org/10.7821/naer.2023.7.1458
(2023). Chatbot gamified and automated management of L2 learning process using Roumeliotis, K. I., & Tselikas, N. D. (2023). ChatGPT and open-AI models: A preliminary
smart sender platform. International Journal of Educational Methodology, 9(3), review. Future Internet, 15(6), 192.
603–618. https://doi.org/10.12973/ijem.9.3.603 Ruiz-Rojas, L. I., Acosta-Vargas, P., De-Moreta-Llovet, J., & Gonzalez-Rodriguez, M.
Ofosu-Ampong, K., Acheampong, B., & Kevor, M.-O. (2023). Acceptance of artificial (2023). Empowering education with generative artificial intelligence tools:
intelligence (ChatGPT) in education: Trust, innovativeness and psychological need of Approach with an instructional design matrix. Sustainability, 15(15), 11524. https://
students. Information and Knowledge Management, 13(4), 37–47. doi.org/10.3390/su151511524
Okonkwo, C. W., & Ade-Ibijola, A. (2021). Chatbots applications in education: A Rukhiran, M., Phaokla, N., & Netinant, P. (2022). Adoption of environmental
systematic review. Computers & Education: Artificial Intelligence, 2, Article 100033. information chatbot services based on the internet of educational things in smart
Opara, E., Mfon-Ette Theresa, A., & Aduke, T. C. (2023). ChatGPT for teaching, learning schools: Structural equation modeling approach. Sustainability, 14(23), 15621.
and research: Prospects and challenges. Glob Acad J Humanit Soc Sci, 5. https://doi.org/10.3390/su142315621
Park, A., & Lee, S. B. (2023). Examining AI and systemic factors for improved chatbot Saadna, Y., Boudhir, A. A., & Ben Ahmed, M. (2022). An analysis of ResNet50 model and
sustainability. Journal of Computer Information Systems, 1–15. RMSprop optimizer for education platform using an intelligent chatbot system. In
Paul, J., Ueno, A., & Dennis, C. (2023). ChatGPT and consumers: Benefits, pitfalls and Networking, intelligent systems and security: Proceedings of NISS 2021 (pp. 577–590).
future research agenda. International Journal of Consumer Studies, 47(4), 1213–1225. Saaida, M. B. (2023). AI-Driven transformations in higher education: Opportunities and
Pavlik, J. V. (2023). Collaborating with ChatGPT: Considering the implications of challenges. International Journal of Educational Research and Studies, 5(1), 29–36.
generative artificial intelligence for journalism and media education. Journalism and Saihi, A., Ben-Daya, M., & As’ad, R. (2023). A hierarchical component model for
Mass Communication Educator, 78(1), 84–93. sustainable performance measurement of maintenance practices: A fourth-order PLS-
Pesonen, J. A. (2021). ‘Are you OK?’ Students’ trust in a chatbot providing support SEM approach. Computers & Industrial Engineering, 185, Article 109699.
opportunities. Lecture Notes in Computer Science, 199–215. Sakib, N., Anik, F. I., & Li, L. (2023). ChatGPT in IT education ecosystem: Unraveling
Piñeiro-Martín, A., García-Mateo, C., Docío-Fernández, L., & López-Pérez, M.d. C. (2023). long-term impacts on job market, student learning, and ethical practices. In
Ethical challenges in the development of virtual assistants powered by large Proceedings of the 24th annual conference on information technology education (pp.
language models. Electronics, 12(14), 3170. 73–78).
Plantak Vukovac, D., Horvat, A., & Čižmešija, A. (2021). Usability and user experience of Sallam, M. (2023). ChatGPT utility in healthcare education, research, and practice:
a chat application with integrated educational chatbot functionalities. Lecture Notes Systematic review on the promising perspectives and valid concerns. Healthcare, 11
in Computer Science, 216–229. (6), 887. https://doi.org/10.3390/healthcare11060887
Pradana, M., Elisa, H. P., & Syarifuddin, S. (2023). Discussing ChatGPT in education: A Sallam, M., Salim, N. A., Barakat, M., & Al-Tammemi, A. B. (2023). ChatGPT applications
literature review and bibliometric analysis. Cogent Education, 10(2), Article in medical, dental, pharmacy, and public health education: A descriptive study
2243134. https://doi.org/10.1080/2331186X.2023.2243134 highlighting the advantages and limitations. Narrative J, 3(1), e103. https://doi.org/
Qadir, J. (2023). Engineering education in the era of ChatGPT: Promise and pitfalls of 10.52225/narra.v3i1.103
generative AI for education. IEEE global engineering education conference (pp. 1–9). Sarosa, M., Kusumawardani, M., Suyono, A., & Wijaya, M. H. (2020). Developing a social
EDUCON. media-based Chatbot for English learning. In IOP conference series: Materials science
Qureshi, B. (2023). Exploring the use of chatgpt as a tool for learning and assessment in and engineering (Vol. 732)IOP Publishing, Article 012074. No. 1.
undergraduate computer science curriculum: Opportunities and challenges. arXiv preprint Sarstedt, M., & Cheah, J. H. (2019). Partial least squares structural equation modeling using
arXiv:2304.11214. SmartPLS: A software review.
Ragheb, M. A., Tantawi, P., Farouk, N., & Hatata, A. (2022). Investigating the acceptance Sarstedt, M., Hair Jr, J. F., Cheah, J.-H., Becker, J.-M., & Ringle, C. M. (2019). How to
of applying chat-bot (Artificial intelligence) technology among higher education specify, estimate, and validate higher-order constructs in PLS-SEM. Australasian
students in Egypt. International Journal of Hygiene and Environmental Medicine, 8(2). Marketing Journal, 27(3), 197–211.
Rahim, N. I. M., Iahad, N. A., Yusof, A. F., & Al-Sharafi, M. A. (2022). AI-based chatbots Schmidhuber, J., Schlögl, S., & Ploder, C. (2021). Cognitive load and productivity
adoption model for higher-education institutions: A hybrid PLS-SEM-neural network implications in human-chatbot interaction. In 2021 IEEE 2nd international conference
modelling approach. Sustainability, 14(19), Article 12726. on human-machine systems (ICHMS) (pp. 1–6). IEEE.
Rajabi, P., Taghipour, P., Cukierman, D., & Doleck, T. (2023). Exploring ChatGPT’s Schünke, L. C., Mello, B., da Costa, C. A., Antunes, R. S., Rigo, S. J., de Oliveira
impact on post-secondary education: A qualitative study. In Proceedings of the 25th Ramos, G., … Donida, B. (2022). A rapid review of machine learning approaches for
western Canadian conference on computing education (pp. 1–6). telemedicine in the scope of COVID-19. Artificial Intelligence in Medicine, 129, Article
Rajaobelina, L., Prom Tep, S., Arcand, M., & Ricard, L. (2021). Creepiness: Its 102312.
antecedents and impact on loyalty when interacting with a chatbot. Psychology and Sebastian, G. (2023). Privacy and data protection in ChatGPT and other AI chatbots:
Marketing, 38(12), 2339–2356. Strategies for securing user information. Available at: SSRN 4454761.
Ram, B., & Pratima Verma, P. V. (2023). Artificial intelligence AI-based chatbot study of Shahriar, S., & Hayawi, K. (2023). Let’s have a chat! A conversation with ChatGPT:
ChatGPT, Google AI bard and baidu AI. World Journal of Advanced Engineering Technology, applications, and limitations. arXiv preprint arXiv:2302.13817.
Technology and Sciences, 8(1), 258–261. Shamkuwar, M., & Sharma, D. (2023). Education 4.0: Artificial intelligence dimensions.
Ramachandran, V., Palanisamy, P., & Pachamuthu, B. (2023). A Tête-à-tête with Advancements in artificial intelligence, blockchain technology, and IoT in higher
ChatGPT on the impact of artificial intelligence in medical education. Medical education: Mitigating the impact of COVID-19.
Journal of Malaysia, 78(4), 547–549. Sharef, N. M., Murad, M. A. A., Mansor, E. I., Nasharuddin, N. A., Omar, M. K., &
Rane, N. (2023). Enhancing the quality of teaching and learning through ChatGPT and Rokhani, F. Z. (2021). Personalized learning based on learning analytics and chatbot.
similar large language models: Challenges, future prospects, and ethical In 2021 1st conference on online teaching for mobile education (Vol. 2021)OT4ME.
considerations in education. Future Prospects, and Ethical Considerations in Education.
23
Sharma, N. (2023). Artificial intelligence: Paving the way to a smarter education system. Wadhawan, I., Jain, T., & Galhotra, B. (2023). Usage and adoption of Chatbot in
In Artificial intelligence for societal issues (pp. 141–157). Springer. education sector. In 2023 7th international conference on intelligent computing and
Sharma, S., Aggarwal, R., & Kumar, M. (2023). Mining twitter for insights into ChatGPT control systems (ICICCS) (pp. 1097–1103). IEEE.
sentiment: A machine learning approach. 2nd IEEE international conference on Wamba, S. F., Queiroz, M. M., Jabbour, C. J. C., & Shi, C. V. (2023). Are both generative
distributed computing and electrical circuits and electronics, ICDCECE 2023. AI and ChatGPT game changers for 21st-Century operations and supply chain
Shoufan, A. (2023). Exploring students’ perceptions of ChatGPT: Thematic analysis and excellence? International Journal of Production Economics, 265, Article 109015.
follow-up survey. IEEE Access, 11, 38805–38818. https://doi.org/10.1109/ Wang, F., King, R. B., Chai, C. S., & Zhou, Y. (2023). University students’ intentions to
ACCESS.2023.3268224 learn artificial intelligence: The roles of supportive environments and
Siemens, G. (2005). Connectivism: A theory of learning for the digital age. International expectancy–value beliefs. International Journal of Educational Technology in Higher
journal of instructional technology and distance learning, 2(1), 1–15. Education, 20(1), 51.
Skavronskaya, L., Hadinejad, A., & Cotterell, D. (2023). Reversing the threat of artificial Wang, T., Lund, B. D., Marengo, A., Pagano, A., Mannuru, N. R., Teel, Z. A., & Pange, J.
intelligence to opportunity: A discussion of ChatGPT in tourism education. Journal of (2023). Exploring the potential impact of artificial intelligence (AI) on international
Teaching in Travel & Tourism, 23(2), 253–258. students in higher education: Generative AI, chatbots, analytics, and international
Smolansky, A., Cram, A., Raduescu, C., Zeivots, S., Huber, E., & Kizilcec, R. F. (2023). student success. Applied Sciences, 13(11), 6716. https://doi.org/10.3390/
Educator and student perspectives on the impact of generative AI on assessments in app13116716
higher education. In Proceedings of the tenth ACM conference on learning@ scale (pp. Wang, T., Wang, D., Li, B., Ma, J., Pang, X. S., & Wang, P. (2023). The impact of
378–382). anthropomorphism on chatgpt actual use: Roles of interactivity, perceived
Sonkar, S., Le, M., Chen, X., Liu, N., Mallick, D. B., & Baraniuk, R. G. (2023). Code enjoyment, and extraversion. Perceived Enjoyment, and Extraversion.
soliloquies for accurate calculations in large language models. arXiv preprint arXiv: Wardat, Y., Tashtoush, M. A., AlAli, R., & Jarrah, A. M. (2023). ChatGPT: A revolutionary
2309.12161. tool for teaching and learning mathematics. Eurasia Journal of Mathematics, Science
Strzelecki, A. (2023). To use or not to use ChatGPT in higher education? A study of and Technology Education, 19(7), Article em2286. https://doi.org/10.29333/ejmste/
students’ acceptance and use of technology. Interactive Learning Environments, 1–14. 13272
https://doi.org/10.1080/10494820.2023.2209881 Wu, X., Duan, R., & Ni, J. (2023). Unveiling security, privacy, and ethical concerns of
Stupina, M., & Paniotova, V. (2023). An educational chatbot in a blended learning ChatGPT. arXiv preprint arXiv:2307.14192.
environment. In 2023 3rd international conference on technology enhanced learning in Xia, Q., Chiu, T. K., Chai, C. S., & Xie, K. (2023). The mediating effects of needs
higher education (TELE) (pp. 276–279). IEEE. satisfaction on the relationships between prior knowledge and self-regulated
Su, J., & Yang, W. (2023). Unlocking the power of ChatGPT: A framework for applying learning through artificial intelligence chatbot. British Journal of Educational
generative AI in education. ECNU Review of Education, 6(3), 355–366. https://doi. Technology, 54(4), 967–986.
org/10.1177/20965311231168423 Xia, Q., Chiu, T. K., Lee, M., Sanusi, I. T., Dai, Y., & Chai, C. S. (2022). A self-
Suárez, A., Adanero, A., Díaz-Flores García, V., Freire, Y., & Algar, J. (2022). Using a determination theory (SDT) design approach for inclusive and diverse artificial
virtual patient via an artificial intelligence chatbot to develop dental students’ intelligence (AI) education. Computers & Education, 189, Article 104582.
diagnostic skills. International Journal of Environmental Research and Public Health, 19 Xiao, C., Xu, S. X., Zhang, K., Wang, Y., & Xia, L. (2023). Evaluating reading
(14), 8735. https://doi.org/10.3390/ijerph19148735 comprehension exercises generated by LLMs: A showcase of ChatGPT in education
Syed, Z. H., Trabelsi, A., Helbert, E., Bailleau, V., & Muths, C. (2021). Question applications. In Proceedings of the 18th workshop on innovative use of NLP for building
answering chatbot for troubleshooting queries based on transfer learning. Procedia educational applications (BEA 2023) (pp. 610–625).
Computer Science, 192, 941–950. Xu, Y., Bradford, N., & Garg, R. (2023). Transparency enhances positive perceptions of
Tafrova-Grigorova, А. (2016). Historical roots and development of constructivism. social artificial intelligence. Human Behavior and Emerging Technologies, 2023(1),
Chemistry: Bulgarian Journal of Science Education, 25(1), 75–106. Article 5550418.
Tapalova, O., & Zhiyenbayeva, N. (2022). Artificial intelligence in education: AIEd for Xu, R., Feng, Y., & Chen, H. (2023). Chatgpt vs. google: A comparative study of search
personalised learning pathways. Electronic Journal of E-Learning, 20(5), 639–653. performance and user experience. arXiv preprint arXiv:2307.01135.
Thosani, P., Sinkar, M., Vaghasiya, J., & Shankarmani, R. (2020). A self learning chat-bot Yadegari, M., Mohammadi, S., & Masoumi, A. H. (2022). Technology adoption: An
from user interactions and preferences. In 2020 4th international conference on analysis of the major models and theories. Technology Analysis & Strategic
intelligent computing and control systems (ICICCS) (pp. 224–229). IEEE. Management, 1–15.
Tisland, I., Sodefjed, M. L., Vassilakopoulou, P., & Pappas, I. O. (2022). The role of Yang, J., Chen, Y.-L., Por, L. Y., & Ku, C. S. (2023). A systematic literature review of
quality, trust, and empowerment in explaining satisfaction and use of chatbots in e- information security in chatbots. Applied Sciences, 13(11), 6355.
government. Conference on e-Business, e-Services and e-Society (pp. 279–291). Yeşilçınar, S. (2023). Personalized learning through gamification: A ChatGPT approach
Tiwari, C. K., Bhat, M. A., Khan, S. T., Subramaniam, R., & Khan, M. A. I. (2023). What to English Language learning. In Transforming the language teaching experience in the
drives students toward ChatGPT? An investigation of the factors influencing age of AI (pp. 44–64). IGI Global.
adoption and usage of ChatGPT. Interactive Technology and Smart Education. https:// Yilmaz, H., Maxutov, S., Baitekov, A., & Balta, N. (2023). Student attitudes towards chat
doi.org/10.1108/ITSE-04-2023-0061 GPT: A technology acceptance model survey. International Educational Review, 1(1),
Tlili, A., Shehata, B., Adarkwah, M. A., Bozkurt, A., Hickey, D. T., Huang, R., & 57–83.
Agyemang, B. (2023). What if the devil is my guardian angel: ChatGPT as a case Yu, H., & Guo, Y. (2023). Generative artificial intelligence empowers educational reform:
study of using chatbots in education. Smart Learning Environments, 10(1), 15. https:// Current status, issues, and prospects. Frontiers in Education, 8, Article 1183162.
doi.org/10.1186/s40561-023-00237-x https://doi.org/10.3389/feduc.2023.1183162
Totlis, T., Natsis, K., Filos, D., Ediaroglou, V., Mantzou, N., Duparc, F., & Piagkou, M. Yusup, M. (2023). Analysis of chatbot development for learning and teaching principles
(2023). The potential role of ChatGPT and artificial intelligence in anatomy based on service efficiency: Chatbot development for learning and teaching.
education: A conversation with ChatGPT. Surgical and Radiologic Anatomy, 45(10), International Journal Of Computer Sciences and Mathematics Engineering, 2(1), 45–51.
1321–1329. https://doi.org/10.1007/s00276-023-03229-1 Zhai, X. (2022). ChatGPT user experience: Implications for education. Available at: SSRN
Van Merrienboer, J. J., & Sweller, J. (2005). Cognitive load theory and complex learning: 4312418.
Recent developments and future directions. Educational Psychology Review, 17, Zhang, R., Zou, D., & Cheng, G. (2023). Chatbot-based learning of logical fallacies in EFL
147–177. writing: Perceived effectiveness in improving target knowledge and learner
Vargas-Murillo, A. R., de la Asuncion Pari-Bedoya, I. N. M., & de Jesús Guevara-Soto, F. motivation. Interactive Learning Environments, 1–18. https://doi.org/10.1080/
(2023). Challenges and opportunities of AI-assisted learning: A systematic literature 10494820.2023.2220374
review on the impact of ChatGPT usage in higher education. International Journal of Zhu, C., Sun, M., Luo, J., Li, T., & Wang, M. (2023). How to harness the potential of
Learning, Teaching and Educational Research, 22(7), 122–135. https://doi.org/ ChatGPT in education? Knowledge Management and E-Learning, 15(2), 133–152.
10.26803/ijlter.22.7.7 https://doi.org/10.34105/j.kmel.2023.15.008
Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of
information technology: Toward a unified view. MIS Quarterly, 425–478.
24

câu hỏi

Uploaded by

Copyright:

Available Formats

câu hỏi

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

câu hỏi

Uploaded by

Copyright:

Available Formats

Computers and Education: Artificial Intelligence 7 (2024) 100274

Contents lists available at ScienceDirect

Computers and Education: Artificial Intelligence

A Structural equation modeling analysis of generative AI chatbots adoption

Fig. 1. Phases of the research process.

• HOC1 (“Trust and Privacy Factor”): Its corresponding LOCs are

Fig. 2. Preliminary second order model for AI-chatbot adoption.

Fig. 3. Approaches for measurement models evaluation.

Fig. 4. Updated model for AI-chatbot adoption.

outer weights, illustrating their contribution to forming the construct.

2 Adoption of AI-chatbots 0.70

Fig. 5. Stage 2 runs using the LV scores.

Fig. 6. Redundancy analysis tests for convergent validity.

dictor constructs of each subpart of the structural model is conducted

Fig. 7. Approach for path model evaluation.

R2 0.356 0.579 0.393 0.604 0.571 0.452 0.796

related to these innovative technologies are perceived and the pivotal

Fig. 8. Slope analysis – Interaction plots.

Availability of data and materials Acknowledgements

Fig. A.1. Slope analyses – Interaction plots.

You might also like