[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Next Article in Journal
Research on Factors Influencing Global Carbon Emissions and Forecasting Models
Previous Article in Journal
Multi-Scale Effect of Land Use Landscape on Basin Streamflow Impacts in Loess Hilly and Gully Region of Loess Plateau: Insights from the Sanchuan River Basin, China
Previous Article in Special Issue
Artificial Intelligence Literacy Competencies for Teachers Through Self-Assessment Tools
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 16
Issue 23
10.3390/su162310780
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Can Multimodal Large Language Models Enhance Performance Benefits Among Higher Education Students? An Investigation Based on the Task–Technology Fit Theory and the Artificial Intelligence Device Use Acceptance Model

by
Amany Al-Dokhny
1,*,
Omar Alismaiel
1,
Samia Youssif
2,
Nermeen Nasr
2,
Amr Drwish
1 and
Amira Samir
2
1
Curriculum and Instruction Department, College of Education, King Faisal University, Al-Ahsa, P.O. Box 400, Hofuf 31982, Saudi Arabia
2
Educational Technology Department, College of Specific Education, Ain Shams University, Cairo 11566, Egypt
*
Author to whom correspondence should be addressed.
Sustainability 2024, 16(23), 10780; https://doi.org/10.3390/su162310780
Submission received: 17 October 2024 / Revised: 27 November 2024 / Accepted: 29 November 2024 / Published: 9 December 2024
(This article belongs to the Special Issue Sustainable Digital Education: Innovations in Teaching and Learning)
Browse Figures
Figure 1
<p>Research model.</p> "> Figure 2
<p>Path coefficient results.</p> "> Figure 3
<p>Path (t-value) results.</p> ">
Versions Notes

Abstract

:
The current study highlights the potential of multimodal large language models (MLLMs) to transform higher education by identifying key factors influencing their acceptance and effectiveness. Aligning technology features with educational needs can enhance student engagement and learning outcomes. The study examined the role of MLLMs in enhancing performance benefits among higher education students, using the task–technology fit (T-TF) theory and the artificial intelligence device use acceptance (AIDUA) model. A structured questionnaire was used to assess the perceptions of 550 Saudi university students from various academic disciplines. The data were analyzed via structural equation modeling (SEM) using SmartPLS 3.0. The findings revealed that social influence negatively affected effort expectancy regarding MLLMs and that hedonic motivation was also negatively related to effort expectancy. The findings revealed that social influence and hedonic motivation negatively affected effort expectancy for MLLMs. Effort expectancy was also negatively associated with T-TF in the learning context. In contrast, task and technology characteristics significantly influenced T-TF, which positively impacted both performance benefits and the willingness to accept the use of MLLMs. A strong relationship was found between adoption willingness and improved performance benefits. The findings empower educators to strategically enhance MLLMs adoption strategically, driving transformative learning outcomes.

1. Introduction

Generative artificial intelligence (GAI) has the potential to revolutionize higher education by leveraging artificial intelligence (AI) algorithms to generate educational content, provide personalized learning experiences, and facilitate real-time feedback. This can enhance student engagement, promote critical thinking, and foster creativity in previously unattainable ways. Generative AI applications offer innovative and effective solutions across various educational contexts, including learning, training, assessment, evaluation, and feedback. They enable educators to adapt and optimize teaching methodologies to meet students’ evolving needs in these multiple contexts [1].
Multimodal large language models (MLLMs) are one of the most important applications of GAI. They are complex natural language processing (NLP) models trained on vast datasets. This enables them to perform diverse tasks, such as text generation and comprehension [1]. As this technology relies primarily on NLP techniques, it consists of transformer architectures that carry out a similar process for language processing [2]. They serve as a way for the model to analyze the relationships between words in a sentence [3]. By providing specific prompts, users can guide MLLMs to perform their desired tasks, such as writing different kinds of creative content, translating languages, and answering questions in a particular style [1].
MLLMs are effective in achieving valuable learning outcomes for higher education students [4,5], especially regarding learning methods and strategies [6]. They can generate creative, contextually diverse educational content, improving student engagement and interaction by personalizing their learning experiences [7,8,9].
However, fit factors must be considered for information systems in the context of their use among higher education students [10]. These factors are user characteristics, task characteristics, technology characteristics, and performance benefits. The technology characteristics of the information system used in the learning environment affect performance benefits and, consequently, its educational outcomes [10]. Similarly, the nature of the task characteristics assigned to students affects the fit of these tasks to the nature as well as technology characteristics [11]. Moreover, the future adoption of these information systems in these students’ various learning activities depends on their performance benefits [12,13]. These critical factors are the main driving factors of the task–technology fit (T-TF) theory [14].
The T-TF theory is a core concept in the field of information systems, as it refers to the extent to which technology can support users’ needs and goals efficiently and effectively. It has three basic components: the technology, the task, and the user. T-TF is a multidimensional concept that requires careful analysis of its basic components. Understanding these components is essential for selecting the appropriate technology that meets users’ needs and supports their goals efficiently and effectively [14]. MLLMs possess enormous potential to enhance performance benefits for students interacting with them [15]. This entails positively influencing the three core factors of fit: task, technology, and user. MLLMs can positively impact performance benefits through an accurate understanding of the task context, correctly analyzing the user’s needs and clearly defining the goals [12]. They can also personalize the task to the user by customizing it for each user, considering the user’s characteristics and preferences, which enhances their feelings of engagement and motivation. In addition, they provide dynamic support for the task, which occurs through its interaction with the user while completing the task, and thus offer guidance in real time, which helps the user make the right decisions and avoid mistakes [12,16,17]. Regarding technology factors, such as perceived usefulness and perceived ease of use, MLLMs can enhance the capabilities of technology by performing more complex and precise tasks, such as text and image analysis, language translation, and creative content creation. Furthermore, the integration of different technologies into one model can contribute to creating a smoother and more effective user experience [18]. Regarding the user factor, MLLMs improve the user experience by making it personalized, easy to use, and effective, which enhances users’ satisfaction and loyalty. As a result, it contributes to efficiently and effectively ensuring performance benefits for users. Therefore, the current study seeks to examine the role of T-TF model factors in the context of MLLMs [17,18].
The cognitive appraisal theory (CAT) of emotions, proposed by psychologist Richard Lazarus, is one of the most prominent theories in emotional psychology [19,20]. This theory seeks to understand the cognitive mechanisms that lead to the emergence and experience of emotions [19,20]. Lazarus argues that emotions are not simply direct physiological reactions to events but are the result of an active cognitive evaluation of these events. In essence, the way an individual interprets an event determines the type of emotion they feel and how they respond to it [19,20]. Lazarus suggests that the evaluation process includes two main stages. The first is initial evaluation, which aims to determine the importance of the event and its relevance to the self. This includes assessing whether the event represents a threat or an opportunity and whether it is aligned with one’s goals and values. The second is secondary assessment, which focuses on assessing the resources available to the individual to deal with the event. This includes the assessment of personal competencies, social support, and available coping strategies. Lazarus argues that the results of these two assessments jointly determine the type of emotion an individual feels and how they respond to it behaviorally and physiologically. For example, if a person appraises an event as a major threat and does not have sufficient resources to deal with it, they are likely to feel fear or anxiety. In contrast, if a person evaluates an event as an opportunity and believes they could take advantage of it, they are likely to feel happy or excited [19,20].
Gursoy et al. [21] proposed and validated a theoretical framework known as the AIDUA model. This framework outlines three progressive phases: initial evaluation, secondary evaluation, and resulting outcomes. It identifies six key elements that shape acceptance, namely, social influence, hedonic motivation, anthropomorphism, performance expectancy, effort expectancy, and emotion [19,20]. The model offers valuable insights into the psychological aspects that affect individuals’ readiness to embrace and engage with intelligent technologies [21].
MLLMs are models capable of understanding and generating text, images, and sounds in a human-like manner. These models make interacting with smart devices more natural and easier, reducing effort on the part of the user. In addition, the social influence factor plays a crucial role in this context. If there is a positive culture around AI and its applications, individuals will be more willing to try to engage with such technology [19,20,22]. Hedonic motivation also plays an important role in the acceptance of these models. If MLLMs can provide enjoyable and rewarding experiences for users, they will be encouraged to continue using them [19,20,23]. Finally, the willingness to accept the use of the technology plays an important role in the acceptance of these models. Individuals who have a high level of self-confidence and ability to deal with technology are more willing to try new technologies, such as large language models [21].
Given the preceding analysis, the current study posits that MLLMs have emerged as indispensable tools within the educational landscape of universities. These models possess transformative potential to enhance teaching and learning methodologies, thereby ensuring the sustainability and advancement of this crucial sector. The sustainability of MLLMs in the educational landscape lies in their ability to adapt and evolve with the ever-changing demands of higher education. By integrating diverse data types and providing comprehensive, context-aware solutions, these models enable universities to address complex challenges such as personalized learning, efficient resource allocation, and enhanced accessibility. Their transformative potential in improving teaching and learning methodologies ensures the long-term viability of educational systems by fostering innovation, inclusivity, and adaptability. This capacity to continuously improve and respond to dynamic educational needs underpins its role in sustaining and advancing the higher education sector.
Despite the growing significance of MLLMs, empirical and descriptive research exploring their impact on higher education students remains relatively scarce. To address this knowledge gap, this study aims to investigate the potential of maximizing performance benefits for higher education students through the strategic utilization of MLLMs. This is accomplished by examining the influence of task characteristics, technology characteristics, and user characteristics (collectively known as T-TF factors) within the framework of the AIDUA model (derived from CAT) as well as by examining the effects of social influence, hedonic motivation (collectively known as effort expectancy factors), and the willingness to accept the use of the technology. Thus, this study seeks to identify the optimal conditions for maximizing performance benefits for students when employing MLLMs.

2. Theoretical Framework and Research Hypotheses

2.1. A Comprehensive Review of MLLMs in Higher Education

The world has witnessed a massive technological revolution in the field of AI in recent years, manifested in the emergence of MLLMs with advanced capabilities to carry out remarkably complex tasks. These models are the result of tremendous progress in the field of NLP, as they provide extensive possibilities for creating diverse content by utilizing available digital content (video, images, graphics, texts, etc.) [24].
This technology has engendered a qualitative shift in the field of education, leading to transformation in current educational practices. This transformation is evident in the enhancement of personal and interactive learning, facilitated by the provision of rich and diverse educational environments that meet the needs of each student [25]. Furthermore, it offers a wide range of formative and summative assessment methods and tools, which provide ongoing feedback to students and teachers alike. This feedback contributes to improving the learning process and adapting it to the needs of each student [24,25].
The effectiveness of using MLLMs efficiently depends on an important aspect: prompt engineering [26]. This makes it the main driver in terms of benefiting from MLLMs’ potential. Prompt engineering concerns directing and shaping the response tone for a system based on language generation [27]. Examples of this include generating creative texts in different formats and answering user questions based on multiple creative methods (argumentative, comic, informational, or another specific tone) [26]. In essence, prompt engineering refers to how a system’s response is guided by the relevant context to obtain more accurate outputs relevant to the context.
Several scholarly works have analyzed the function and importance of MLLMs within the educational domain and the burgeoning interest in their applications. A thorough examination of the literature, as illustrated in Table 1, reveals various examples reinforcing MLLMs’ paramount importance in the educational sphere.
Recent studies highlight the transformative role of MLLMs in higher education, particularly in enhancing personalized learning and cognitive skill development. Experimental and descriptive research shows that MLLMs are effective personalized tutors that can improve comprehension, retention, and engagement with course materials. Qualitative studies further reveal that MLLMs promote critical thinking and collaborative learning, helping students develop advanced problem-solving skills. These findings suggest that integrating MLLMs into higher education can significantly enhance academic performance and better prepare students for real-world challenges.

2.2. The T-TF Theory and the AIDUA Model

The T-TF theory posits that the alignment between a technology’s attributes and the tasks it supports directly influences its adoption, use, and effectiveness. A strong T-TF leads to enhanced user satisfaction, increased technology acceptance, and improved performance outcomes [14]. Goodhue and Thompson argue that the T-TF model is crucial for maximizing individual performance when using information technology, as it positively correlates with user efficiency, error reduction, and decision-making [14,16]. The T-TF model can be assessed across dimensions such as task complexity, technology capabilities, and user interface design. In Goodhue and Thompson’s study, the T-TF model was measured using eight key dimensions to evaluate the alignment between technology and user tasks [14].
Various studies have scrutinized the role and significance of the T-TF model in the field of education, in addition to the increasing enthusiasm regarding its utilization. A comprehensive review of the existing literature, as demonstrated in Table 2, reveals numerous instances that underscore the critical importance of the T-TF model in the educational context.
Although many studies have explored MLLMs and their implications in education, only a few of these have addressed the impact of basic fit factors (task, technology, and user) on developing the concept of performance benefits using these models. The current study focuses on filling this research gap by examining the combined effect of the triad of fit factors (task, technology, user, and T-TF) using MLLMs throughout, thereby maximizing the specific performance benefits for it. This is one of the few studies linking the basic fit factors for the T-TF model, which provides a theoretical framework for understanding how these factors influence the adoption of new technologies.
While existing research has primarily focused on the performance and capabilities of MLLMs in processing and generating integrated outputs from diverse data types, there is limited investigation into how specific task requirements and user characteristics interact with the technological attributes of these models. Understanding this relationship is crucial, as it can inform the design and implementation of multimodal systems tailored to specific user needs and tasks, thereby enhancing usability and overall performance. Furthermore, a comprehensive analysis of T-TF in the context of MLLMs could guide future developments, ensuring that these technologies are not only powerful but also practical and user-friendly across various applications. Addressing this gap will contribute to a more nuanced understanding of how to optimize the integration of tasks, technologies, and users in the evolving landscape of multimodal communication.
The interconnectedness of tasks, technology, and users in the T-TF framework is vital for the successful implementation of MLLMs. A comprehensive examination of how these determinants interrelate can result in the advancement of more effective and accessible systems, ultimately augmenting the overall efficacy and applicability of MLLM technologies across diverse domains. Addressing this correlation will ensure that MLLMs are not only robust in their functionalities but also customized to the contexts in which they are employed, facilitating more effective multimodal communication.
In the AIDUA model, effort expectancy—influenced by social influence and hedonic motivation—is a critical determinant of user acceptance and integration of AI technologies. These factors interact to shape users’ perceptions and intentions, thereby influencing the acceptance and integration of AI technologies [38]. CAT elucidates how these factors impact emotional responses and decision-making in technology adoption [19,20].
Social influence plays a pivotal role in shaping individuals’ perceptions and acceptance of AI technologies, reflecting the impact of others’ opinions regarding the use of new tools. Studies highlight that social influence strongly drives technology adoption, particularly when recommendations and usage by peers are widespread [39]. In the context of e-learning, social influence is especially significant, serving as a more robust determinant of AI acceptance compared to other sectors, emphasizing its situational importance [40,41]. This aligns with the UTAUT, which posits that social influence directly affects both behavioral intentions and usage behavior [42].
Hedonic motivation, defined as the enjoyment and satisfaction derived from using a technology, is another critical factor in AI device acceptance. Research confirms that hedonic motivation strongly predicts user intentions to adopt AI, particularly when the primary goal is entertainment [43]. In e-learning, hedonic motivation notably influences learners’ attitudes toward AI tools, increasing the likelihood of adoption for technologies that enhance enjoyment and engagement [44,45]. This is consistent with CAT, which asserts that emotional responses to a technology are shaped by perceived enjoyment and benefits [20].
Effort expectancy, or the perceived ease of use of a technology, is also essential for technology acceptance. Users generally favor tools that require minimal effort. However, research indicates that as users gain experience, the influence of effort expectancy on acceptance diminishes [46]. Initial perceptions of ease of use are critical during early adoption phases but tend to lose significance as familiarity with the technology grows. CAT supports this perspective, suggesting that users evaluate the effort required to use a technology, influencing their emotional reactions and acceptance behaviors [47].
Building on this analysis, this study proposes a theoretical model integrating factors from the AIDUA model, rooted in CAT (effort expectancy, social influence, hedonic motivation, and willingness to accept the use of the technology), with those derived from T-TF theory (task characteristics, technology characteristics, T-TF, and performance benefits). Further details about the model and its development are outlined in the subsequent hypotheses section.

2.3. MLLMs’ Relationships with the T-TF Theory and the AIDUA Model

According to CAT and the AIDUA model, the effort expectancy factor refers to the effort required by an individual to achieve a specific goal [19,20]. This factor is also affected by several indicators, including previous experiences, skills, capabilities, and available resources. In addition, these indicators are influenced by the social influence factor, which refers to the influences exerted by the social environment and others on an individual’s motivations and actions. Several indicators affect the social influence factor as well, with social pressures, social comparisons, and social support being the most important [19,20]. Thus, the relationship between the two factors is clear: the social influence factor significantly negatively affects the effort expectancy factor. If social influence is high and expectations are high regarding the ease of using MLLMs, this may lead to an increase in the individual’s expectation to exert little effort when using MLLMs, and vice versa.
Several studies have explored the relationship between social influence and effort expectancy in the context of MLLMs. Social influence negatively impacts effort expectancy [48,49,50]. Thus, when individuals perceive social influence or encouragement, they may find AI systems easier to use; consequently, effort expectancy decreases. Some studies have found that social influence indirectly affects effort expectancy by enhancing the initial factors in AI systems, and this, in turn, can make the technology seem easier to use [48,51,52,53,54]. Therefore, regarding social influence and effort expectancy in the context of MLLMs, the following hypothesis was formulated:
Hypothesis 1.
Social influence and effort expectancy are negatively related for MLLMs.
According to CAT and the AIDUA model, hedonic motivation refers to the enjoyment or pleasure derived from using an AI system. This factor is critical as it directly impacts a user’s willingness to adopt and engage with AI technology. Research indicates that hedonic motivation is positively associated with performance expectancy; the more enjoyable the user finds an AI system, the higher their expectations of its performance [51,55]. A thorough review of studies addressing hedonic motivation and effort expectancy within the frameworks of CAT and the AIDUA model highlights their significance. For example, one study found that hedonic motivation and effort expectancy were pivotal in eliciting positive emotions essential for the acceptance of autonomous vehicles [47]. Another study demonstrated that both factors significantly influenced e-learning adoption among university students in Qatar and the United States [45]. Additionally, hedonic motivation emerged as a crucial determinant in consumer technology usage [56]. Further analysis of the UTAUT and UTAUT2 models, focusing on factors such as hedonic motivation and effort expectancy, revealed strong correlations between these constructs and behavioral intention to accept and use technology in academic and digital library contexts. A meta-analysis reinforced the importance of these relationships in predicting user acceptance [57]. Based on these findings, the following hypothesis regarding hedonic motivation and effort expectancy in the context of MLLMs (Massive Language Learning Models) was proposed:
Hypothesis 2.
Hedonic motivation and effort expectancy are negatively related for MLLMs.
Related studies derived from the T-TF theory and the AIDUA model predict an inverse relationship between the effort expectancy factor and the T-TF factor. The more integrated the technology is with the task—that is, the more the technology becomes an integral part of accomplishing the task—the less effort the user is expected to expend. Thus, users tend to expect less effort when a technology is easy to use and integrated with the nature of the work they do, making them more likely to accept and adopt it. Two studies in the same context confirmed that T-TF and effort expectancy significantly impact employee intention to adopt the technology and the job [58,59]. Another study found T-TF and effort expectancy to be excellent predictors of adoption behavior toward high prefabrication level technologies [60]. Moreover, T-TF and effort expectancy were shown to positively affect consumers’ behavioral intention to use wearable health care devices, accounting for 68.0% of its variance [61,62]. T-TF and effort expectancy were determined to be direct predictors of the behavioral intention to continue using MOOCs in higher education institutions [63]. Therefore, regarding hedonic motivation and effort expectancy in the context of MLLMs, the following hypothesis was formulated:
Hypothesis 3.
Effort expectancy and T-TF are negatively related for MLLMs.
Task characteristics are fundamentally related to T-TF, influencing how well a given technology supports the tasks it is intended to assist. Proper alignment between task requirements and technology capabilities is crucial for achieving optimal performance outcomes [64,65]. In educational settings, task characteristics have been shown to significantly influence T-TF, which subsequently affects students’ satisfaction and academic performance [66]. In a study focused on employee performance in mobile banking and other settings, task characteristics had a direct and significant impact on T-TF, which, in turn, positively affected individual performance [67]. In the context of metaverse services, task characteristics such as presence, usefulness, and gamification were found to positively affect T-TF [68]. This alignment led to a higher intention among users to continue using metaverse services, demonstrating that effective task–technology alignment enhances user engagement. These results are also consistent with those of several other studies [59,68,69,70,71,72]. Therefore, regarding task characteristics and T-TF in the context of MLLMs, the following hypothesis was formulated:
Hypothesis 4.
Task characteristics and T-TF are positively related for MLLMs.
Findings from related studies highlight the significant impact of technology characteristics on enhancing T-TF in higher education learning environments. For instance, one study investigated the relationship between technology characteristics and T-TF in the context of technostress among university students in online learning environments. It revealed that features such as ease of use and usefulness reduced technostress by improving T-TF, which, in turn, led to greater satisfaction and enhanced academic performance [73]. Another study examined the influence of technology characteristics on T-TF in the adoption of e-learning systems. It found that perceived ease of use and perceived usefulness were critical features that positively impacted T-TF. This alignment improved student satisfaction and academic performance, emphasizing the importance of technology characteristics in fostering effective e-learning environments [33]. In general, the successful integration of technology into higher education depends on aligning technology characteristics with educational tasks. This alignment enhances T-TF, which drives improvements in student satisfaction, performance, and overall success in online and blended learning settings. As educational institutions increasingly adopt digital tools, prioritizing T-TF will be crucial for achieving sustainable and meaningful educational outcomes [64,66].
Based on these insights, the following hypothesis regarding technology characteristics and T-TF in the context of MLLMs was proposed:
Hypothesis 5.
Technology characteristics and T-TF are positively related for MLLMs.
Numerous studies have empirically supported the relationship between T-TF and performance benefits. Some key findings are listed below:
  • Improved performance benefits: when technology effectively supports tasks, users can accomplish more in less time;
  • Enhanced efficiency: T-TF can lead to streamlined workflows and reduced errors;
  • Increased accuracy: technology that is well suited to the task can help minimize mistakes and improve data quality;
  • Higher job satisfaction: when technology is user-friendly and facilitates task completion, employees are more likely to be satisfied with their work;
  • Greater technology adoption: a strong T-TF can encourage users to adopt and utilize technology effectively [14,33,74].
Therefore, regarding T-TF and the performance benefits of MLLMs, the following hypothesis was formulated:
Hypothesis 6.
T-TF and the performance benefits of MLLMs are positively related.
The connection between T-TF and the willingness to accept the use of e-learning systems is well documented in studies that integrate T-TF with models such as the TAM. Research shows that T-TF has a substantial impact on both the initial acceptance and continued use of e-learning systems. When learners perceive that the technology aligns effectively with their tasks, they are more likely to view the system as useful and easy to use, leading to increased satisfaction, improved performance, and sustained adoption of e-learning [75]. Moreover, T-TF not only influences the willingness to accept the use of e-learning systems initially but also plays a vital role in maintaining long-term usage. Studies highlight that T-TF significantly affects perceived usefulness and system satisfaction, both of which are essential for continuous use [33].
Based on these findings, the following hypothesis regarding T-TF and the willingness to accept the use of MLLMs was proposed:
Hypothesis 7.
T-TF and the willingness to accept the use of MLLMs are positively related.
Users’ willingness to accept the use of e-learning systems is a strong predictor of their performance benefits. By understanding the factors influencing this willingness, educators and organizations can create more effective e-learning environments and maximize the potential of technology-enhanced learning [76,77,78]. Moreover, the relationship between willingness to accept the use of e-learning systems and performance benefits is significantly mediated by T-TF and the constructs of the AIDUA model. A strong alignment between the technology and the educational tasks, combined with favorable perceptions of the system’s social influence and hedonic motivation, fosters a greater willingness to adopt e-learning, ultimately enhancing performance outcomes in educational settings [79,80,81]. Therefore, regarding the willingness to accept the use of MLLMs and the performance benefits of MLLMs, the following hypothesis was formulated:
Hypothesis 8.
The willingness to accept the use of MLLMs and the performance benefits of MLLMs are positively related.
Based on the preceding discussion, the current study suggests the following theoretical framework model (see Figure 1 below).
The preceding figure presents a conceptual model illustrating the proposed relationships between the study’s factors and their corresponding hypotheses. The model visually represents the hypothesized positive and negative influences among these factors, providing a framework for understanding their interconnectedness. This model operates similarly to an algorithm, guiding the analysis of the data and the testing of the proposed relationships as in Table 3 below:

3. Materials and Methods

This study examined the influence of MLLMs on enhancing performance benefits among higher education students, contextualized within the framework of the T-TF theory and the AIDUA model. To fulfill the study objectives, a quantitative descriptive methodology was employed, utilizing a questionnaire in conjunction with partial least squares structural equation modeling (PLS-SEM) analysis.

3.1. Study Sample

Undergraduate and postgraduate students from six Saudi Arabian universities participated in this study during the initial semester of the 2024–2025 academic year. The Scientific Research Ethics Committee at King Faisal University (KFU—REC-2024-OCT-ETHICS2686) approved the research. Table 4 details the demographic characteristics of the study population. The normal distribution of participants’ ages (M = 4.16, SD = 0.89, N = 550) and gender (M = 3.26, SD = 0.67, N = 550) was verified, allowing for subsequent statistical analysis. All participants gave informed consent and had the option to withdraw at any point.

3.2. Study Instrument

A questionnaire was designed to evaluate the performance gains associated with MLLMs among higher education students. Building upon existing research on MLLMs, T-TF, and the AIDUA model, the questionnaire delved into eight factors: social influence, hedonic motivation, effort expectancy, task characteristics, technological features, T-TF, willingness to accept the use of MLLMs, and performance benefits of MLLMs (as shown in Figure 1). The questionnaire items, tailored to the study’s context, comprised demographic questions and 48 Likert-scale items (1 = strongly disagree to 5 = strongly agree). The questionnaire underwent expert review by four instructional education specialists.

3.3. Pilot Study

A pilot study involving 100 male and female higher education students (not part of the main study) was conducted to assess the questionnaire’s statistical reliability. SPSS Statistics version 26 was used to calculate Cronbach’s alpha for each factor (Table 5). All factors exhibited satisfactory (α ≥ 0.7) or excellent (α ≥ 0.9) internal consistency [82].

4. Quantitative Data Analysis and Results

This study collected data using an online questionnaire presented via Google Forms. The questionnaire was available online for four weeks, inviting voluntary participation. Data were collected from 550 participants, which is more than the recommended minimum of 400 for social science surveys. This larger sample size resulted in a lower error rate of 4%, according to Weisberg and Bowen’s guidelines [92]. The collected data were analyzed using SPSS Statistics version 26 and SmartPLS version 3.0. According to Hair et al. [93], PLS-SEM involves two stages. The first entails assessing the measurement model using confirmatory factor analysis (CFA); this includes assessing the validity of the measurement model, which involves calculating construct, convergent, and discriminant validity. The second involves evaluating the structural model to examine the relationships between factors.

4.1. Measurement Model Analysis

To evaluate the measurement model, construct validity was first examined. This involves ensuring that the questionnaire items accurately measure the intended factors [82]. The indicator loadings for all items were calculated (shown in Table 6), and it was found that they were all above the recommended threshold of 0.7, indicating good construct validity. According to Hair et al. [93], in CFA, this is considered acceptable for a good level of loading [92].
To assess internal consistency reliability, composite reliability (CR) and Cronbach’s alpha coefficient (CA) were calculated (shown in Table 7). Both values were above the recommended threshold, indicating good internal consistency [93]. Convergent validity was also examined by calculating the average variance extracted (AVE). All eight factors had values greater than 0.5, suggesting that they adequately represent the underlying concepts [93].
Discriminant validity, which refers to the extent to which a factor in the model is distinct from another factor [93], was assessed by comparing the correlations between constructs to their AVE (shown in Table 8). To ensure discriminant validity, the correlations between different factors must be lower than the square root of their AVE. The results showed that each construct was distinct from the others, confirming discriminant validity.
In addition to the Fornell–Larcker criterion presented in Table 8, Hair et al. [94] suggested using the Heterotrait–Monotrait Ratio of Correlations (HTMT) to rigorously assess discriminant validity. The HTMT is a crucial tool for assessing the discriminant validity of latent variables in PLS-SEM. It measures the extent to which two different latent variables are more correlated with themselves than with each other, indicating the degree of distinctiveness among the items or indicators that constitute each latent variable. HTMT is fundamental to evaluating the quality of PLS-SEM models, ensuring that latent variables represent distinct concepts and contributing to improved model fit and more accurate interpretation of results. By verifying that latent variables are indeed distinct and not mere replicas of one another, HTMT supports the assessment of construct validity, signifying the extent to which the items represent the intended theoretical construct. Moreover, it enhances model quality by identifying any issues related to discriminant validity, thereby allowing for necessary adjustments. While there is no universal consensus on a specific cut-off value for HTMT, many researchers [94,95] suggest that a value below 0.85 indicates an acceptable level of discriminant validity, signifying that the latent variables represent distinct and sufficiently differentiated concepts. Conversely, an HTMT value of 0.85 or higher may indicate a problem with discriminant validity, necessitating a reevaluation of the model’s structure or the items that constitute the latent variables. Accordingly, Table 9 reveals that the highest HTMT value is 0.83 (PBMLLMs-WAUMLLMs), proving that the discriminant validity between the latent variables is good.
Regarding the normality test, the PLS-SEM approach is characterized by greater flexibility regarding normal distribution requirements compared to Covariance-Based SEM (CB-SEM). According to Hair et al. [96], PLS-SEM does not require data to follow a normal distribution, allowing it to handle datasets with significant deviations from normality, such as skewness and kurtosis. This flexibility stems from its reliance on least squares estimation, making it suitable for analyzing data that do not adhere to the strict distributional assumptions required by CB-SEM.
The literature suggests that skewness values between −2 and +2 and kurtosis values below ±10 are generally acceptable when using PLS-SEM. However, even when these thresholds are exceeded, PLS-SEM remains robust in providing accurate estimates due to its nonparametric nature, which does not depend on stringent statistical assumptions. In some studies [97], lower thresholds for kurtosis, such as ±7, have been proposed, but they are not mandatory in the context of PLS-SEM. Its flexibility allows it to accommodate a wide range of data characteristics, including those with substantial deviations from normality. Furthermore, in accordance with the guidelines of [98], calculating kurtosis and skewness before applying PLS-SEM is essential for assessing the nature of the data distribution and determining its deviation from normality. While PLS-SEM is flexible and does not require strict normality, understanding these characteristics helps ensure more accurate interpretation of results. Kurtosis indicates the peakedness or flatness of the data distribution, whereas skewness reflects the symmetry of the data around the mean. Identifying these values helps detect potential outliers or extreme deviations that could impact the model’s stability. This process enhances the quality of analysis and ensures the reliability of the final outcomes. Accordingly, descriptive statistics were used to analyze the characteristics of respondents’ answers. The results include the mean, standard deviation, skewness, and kurtosis.
As shown in Table 10, the calculated skewness values, ranging from −0.48 to 0.47, and kurtosis values, ranging from 2.80 to 3.19, suggest that the distribution of responses for each dimension of the questionnaire is approximately normal. These values fall within the statistically acceptable range according to the criteria established by [96,97,99]. Consequently, the use of PLS-SEM, which assumes a normal distribution, is appropriate for analyzing this data.

4.2. Structural Model Analysis

The measurement model was found to be satisfactory. Using SmartPLS version 3.0, the structural model was evaluated. The results are presented in Figure 2 and Figure 3 and Table 11, which show the path coefficients, t-values, and hypothesis testing results for the eight proposed relationships.

4.3. Hypothesis Testing

For Hypothesis 1, this study found that social influence in the context of learning using MLLMs had a significant negative impact on effort expectancy (β = −0.852, SE = 0.051, t = −17.214, p < 0.01). This suggests that increased positive social influence leads to decreased expected student effort. Similarly, for Hypothesis 2, hedonic motivation and students’ effort expectancy regarding learning using MLLMs were significantly negatively related (β = −0.723, SE = 0.062, t = −12.188, p < 0.01). This indicates that, when students are less motivated to learn using MLLMs, they exert more effort in their learning experience. Hypothesis 3 was also supported, as a significant and negative relationship was found between effort expectancy and T-TF in the context of learning using MLLMS (β = −0.558, SE = 0.094, t = −5.812, p < 0.01). This suggests that a lowered effort expectancy when using MLLMs is linked to increased T-TF. Furthermore, for Hypothesis 4, this study revealed significant positive relationships between task characteristics and T-TF (β = 0.648, t = 5.618, p < 0.01). Similar results were found for technology characteristics and T-TF (β = −0.692, t = 4.952, p < 0.01), T-TF and performance benefits (β = 0.515, t = 5.104, p < 0.01), and T-TF and the willingness to accept the use of MLLMs (β = 0.555, t = 4.594, p < 0.01); thus, Hypotheses 5, 6, and 7, respectively, were accepted. These findings suggest that engaging tasks, user-friendly technology, and positive learning experiences contribute to higher T-TF and improved student outcomes. Finally, regarding Hypothesis 8, this study confirmed a significant positive relationship between the willingness to accept the use of MLLMs and performance benefits (β = 0.568, t = 6.344, p < 0.01), indicating that students’ willingness to accept the use of MLLMs is associated with better academic performance.

5. Discussion and Implications

MLLMs are a revolutionary technology in the field of GAI, opening new horizons in many fields, including education. However, the success of integrating these models into higher education does not depend solely on their technical capabilities. Rather, it requires a comprehensive understanding of the factors affecting this technology’s acceptance by end users, especially higher education students. This study aims to fill the current knowledge gap by exploring the factors influencing the improvement in higher education students ‘performance benefits when they use MLLMs in their educational activities. This study employs a combination of psychological and social theories, such as CAT. Based on the AIDUA model and the T-TF theory, a comprehensive model linking the factors of task characteristics, technology characteristics, and T-TF, as well as social influence, hedonic motivation, effort expectancy, and willingness to accept the use of MLLMs, was developed to predict the expected performance benefits of using MLLMs for educational purposes. This model facilitates a deeper understanding of how students interact with this technology, which enables the design of effective strategies to integrate it into the educational process. The results of this study are expected to contribute to enriching academic knowledge about the use of MLLMs in education, in addition to providing practical recommendations to policymakers, educational institutions, and teachers to facilitate the integration of this technology into Saudi higher education, in line with the country’s Vision 2030.

5.1. Social Influence and Effort Expectancy

Social influence is one of the fundamental drivers of human behavior, including individuals’ interactions with new technologies. Regarding Hypothesis 1, the study results indicate that social influence directly affects users’ effort expectancy. In essence, positive social norm signals from peers or experts, such as their recommendations to adopt these models or stories of their successful experiences, increase users’ expectations of the ease of using these models and achieving positive results. In contrast, negative social cues, such as doubts about the effectiveness of these models or concerns about them, increase users’ expectations that they will be difficult to use and reduce their motivation to try them. The results of several recent relevant studies confirm this close relationship; for example, in one study, students who received positive feedback from their classmates about using MLLMs were more willing to accept and use them and expressed greater expectations regarding their ability to achieve better results [100], suggesting that positive social norms can enhance the tendency toward use because of effort expectancy. Several other studies concur with this result [21,97,101,102].

5.2. Hedonic Motivation and Effort Expectancy

Regarding the result for Hypothesis 2, several recent studies show an inverse positive relationship between hedonic motivation and effort expectancy in the context of learning to use MLLMs. Individuals who see these models as a means of enjoyment and entertainment tend to expect that the learning process will be enjoyable and rewarding. Therefore, they expect that they will not need to put in extra effort while learning. For example, in one study, users using chatbots based on linguistic models expressed an expectation that the effort expended would be less when using the technology in the future [103]. Several relevant studies also indicate that, when individuals engage in enjoyable activities that do not require more effort during use, they become motivated to repeat these activities, which leads to increased internal motivation to learn [21,58,104].

5.3. Effort Expectancy and T-TF

The result for Hypothesis 3 states that the more the T-TF for the educational task, the less the students’ effort expectancy regarding achieving the desired goals. Thus, MLLMs, with their ability to generate high-quality texts and summarize complex information, can significantly reduce the effort expectancy of students and teachers alike, if they are designed and applied in a manner that suits the requirements of specific educational tasks. Supporting this, a study on AI-powered design tools suggests that perceived average effort expectancy is a key predictor of users’ intentions to adopt such tools [54]. Another study focused on tools that support the use of AI tools in social development organizations, concluding that the expectation of high effort expectancy reduces cooperation, which affects T-TF and the effectiveness of the tool in general [105]. These results support the hypothesis that there is an inverse relationship between effort expectancy and T-TF, namely that increasing the technology fit reduces users’ effort expectancy. The relationship between effort expectancy and T-TF, as many recent studies indicate, is an inversely complementary relationship that cannot be separated. The higher the compatibility, the lower the effort expectancy. In other words, if the tasks are simple, MLLMs can be sufficient, and thus, effort expectancy is low. Complex tasks also require advanced MLLMs, and if the models used are not appropriate, effort expectancy increases [58,59,62,63].

5.4. Task Characteristics and T-TF

The result for Hypothesis 4 indicates that task characteristics directly and strongly affect T-TF. Specifically, indicators such as analyzing complex data, identifying the main themes and arguments in a scientific document, providing adaptive assignment practices and individual feedback, and developing solutions for real-world problems will lead to greater T-TF for MLLMs among higher education students.
This result is consistent with a study that determined that, during the COVID-19 pandemic, task characteristics (e.g., the complexity of educational tasks) directly communicated with T-TF [106]. The more complex the task characteristics, the greater the reliance on technology specifically designed to support those tasks. Thus, the technology becomes more compatible with the task, and this compatibility between task characteristics and T-TF is conducive to improving performance benefits among students. Another study found that task characteristics affect T-TF, which, in turn, affects students’ intentions to use social media in education [67]. The study concluded that students who effectively used these tools experienced an improvement in academic performance and satisfaction with the educational process.

5.5. Technology Characteristics and T-TF

The result for Hypothesis 5 indicates that technology characteristics directly and strongly affect T-TF. Providing support for creative writing, context awareness, immediate constructive feedback, transparency in thinking, and methods for entering data and outputting information in multiple ways; meeting users’ diverse needs according to social norms and cultural backgrounds; and offering integration for people with disabilities or users with learning difficulties will lead to more T-TF for MLLMs among higher education students.
These findings align with two previous studies on higher education students and e-learning systems. These studies found that technology characteristics like ease of use and perceived usefulness positively impacted technology–task compatibility, leading to improved academic performance [16,33]. Another study examining technology stress in coeducational university settings revealed that user-friendly and useful technology reduced student stress and increased online learning satisfaction [107]. Similarly, user-friendly and useful MLLMs are more likely to be compatible with students’ learning tasks, enhancing their use, academic satisfaction, and performance. User-friendly interfaces and flexible interaction with MLLMs, like ChatGPT, can simplify information access, problem-solving, and research, reducing student stress and effort [107]. Additionally, MLLMs can personalize education, facilitate academic interaction, and assist with tasks like source searching, summarizing, and quick answers, improving education quality and student efficiency, ultimately maximizing performance benefits [108].

5.6. T-TF and the Performance Benefits of MLLMs

The result for Hypothesis 6 indicates that T-TF positively, directly, and strongly affects performance benefits. Providing various options to specify the style or length of the generated content, tools to edit and review created drafts, feedback about performance outputs, integrations with LMS for effective use, and outputs compatible with other educational platforms will lead to greater performance benefits of MLLMs among higher education students.
This finding is consistent with a study that found that T-TF enhances academic performance, which is an indicator of performance benefits [108]. Therefore, using MLLMs designed to match students’ needs and educational tasks can enhance learning effectiveness and contribute to improved performance. T-TF plays a critical role in improving performance benefits in educational and professional settings, as previous studies have indicated [16,58,59,60,62], and MLLMs can have a similar impact if they are used in ways consistent with academic and professional task needs.

5.7. T-TF and the Willingness to Accept the Use of MLLMs

The result for Hypothesis 7 indicates that T-TF directly and strongly affects the willingness to accept the use of MLLMs. This aligns with the results of a study that combined T-TF and TAM to analyze the use of AI-powered design tools; it found performance expectancy and ease of use to be key factors influencing the willingness to accept the use of these tools [109]. Another study showed that T-TF increases users’ intentions to continue learning and adapting to new technologies, which enhances the willingness to accept future use [76]. This is consistent with the way MLLMs work, which requires compatibility between task characteristics and technology characteristics to promote sustained use. Another study on smart devices showed that T-TF had a direct impact on the willingness to accept the use of these devices and their perceived usefulness [110]. By applying these findings to MLLMs, it can be argued that their alignment with the required tasks enhances users’ willingness to accept MLLMs’ future use more effectively. If these forms are compatible with users’ needs—for instance, automatic typing and instant search would give higher education students the opportunity to be more productive in their learning, leading to MLLMs enhancing the value of their studies—the chances of them being accepted and utilized increase dramatically.

5.8. The Willingness to Accept the Use of MLLMs and Performance Benefits

The result for Hypothesis 8 indicates that the willingness to accept the use of MLLMs positively, directly, and strongly affects the performance benefits of MLLMs, helping improve results in courses. Participating in real and virtual classroom activities, facilitating self-directed learning and exploration, enabling automated regular tasks, developing adaptive plans and practices that improve learning performance and reduce unnecessary effort, and developing proficiency in interacting with and using various AI technologies will lead to greater performance benefits of MLLMs among higher education students.
Recent studies highlight a strong, reciprocal relationship between perceived benefits and willingness to adopt technology [76,77,78]. Higher perceived benefits lead to greater willingness to adopt technology, and vice versa. Additionally, a general willingness to adopt GAI tools influences the realization of their benefits. Users must be prepared to use tools correctly to achieve the desired outcomes [79,80,81]. Research suggests that MLLMs should prioritize intelligence, contextual understanding, accuracy, reliability, bias mitigation, and user privacy. Training quality and data volume significantly impact model performance and user acceptance. Adherence to development standards enhances user acceptance and performance benefits.
T-TF, a framework for understanding technology acceptance and adoption, is particularly useful when combined with the AIDUA model (derived from CAT). The fit between task characteristics (e.g., complexity, interdependence) and technology characteristics (e.g., ease of use, perceived usefulness) significantly enhances technology adoption and performance benefits at individual and organizational levels. This fit positively impacts user satisfaction, willingness to adopt the technology, and the sustainability of MLLMs and their performance benefits. The AIDUA model and CAT factors like social influence and hedonic motivation shape user behavior by influencing perceived ease of use, usefulness, and enjoyment of MLLMs. Effort expectancy and social norms, key factors in GAI acceptance models, influence the willingness to adopt GAI systems. A strong fit between these factors leads to enhanced user performance benefits. As MLLMs become more prevalent in professional and academic settings, their success hinges on optimizing the technology–task fit and addressing cognitive, social, and motivational factors to ensure effective support for complex tasks and sustained performance benefits through T-TF factors.

6. Conclusions, Limitations, and Future Work

In conclusion, the T-TF framework offers a robust foundation for enhancing the performance benefits of MLLMs by aligning task characteristics with technology characteristics to enhance individual performance benefits via the willingness to accept the use of the technology among higher education students. When combined with factors derived from the AIDUA model and CAT, including social influence, hedonic motivation, and effort expectancy, this alignment boosts users’ willingness to accept the use of MLLMs. The high compatibility between T-TF and MLLMs not only improves performance benefit outcomes but also ensures the continuity of use in professional and academic environments. Ultimately, the success of MLLMs in supporting complex, interdependent tasks relies on optimizing this fit while addressing the cognitive, social, and motivational dimensions that influence user behavior.
Several limitations may impact this study’s findings and their broader applicability. The use of convenience sampling focused on Saudi universities limits the generalizability of the results to the wider student population in Arab countries. In addition, data collected via Google Forms may include respondents with limited familiarity or experience with MLLMs, potentially skewing the results. The exclusive focus on the T-TF framework, the AIDUA model, and factors related to CAT may also overlook other significant factors affecting the willingness to accept the use of MLLMs and the performance benefits of MLLMs. Future research should address these limitations by expanding the sample to include a more diverse range of Arab countries and student demographics, with inclusion criteria focused on participants experienced in using MLLMs for educational purposes. Moreover, the research framework could be broadened to consider additional factors and their interactions with both TAM and UTAUT factors.
Moreover, future research on the acceptance and use of MLLMs should integrate a broader range of factors, particularly those related to CAT and self-determination theory (SDT). It would also be valuable to explore how cognitive appraisals, such as perceptions of task relevance and emotional responses to MLLMs, influence acceptance and performance. In addition, SDT’s components—namely, autonomy, competence, and relatedness—should be examined to understand how MLLMs can foster intrinsic motivation and enhance educational engagement. Investigating the role of perceived control over learning, the satisfaction of users’ psychological needs, and the capability of MLLMs to provide personalized support could offer deeper insights into their impact on student performance. By incorporating these cognitive and motivational factors, along with the T-TF framework and the AIDUA model, future studies can develop a more holistic model of how MLLMs facilitate academic achievement and sustained user engagement.

Author Contributions

Conceptualization, A.A.-D., O.A. and A.S.; Data curation, A.A.-D., O.A., S.Y., N.N., A.D. and A.S.; Formal analysis, A.A.-D., O.A., A.D. and A.S.; Funding acquisition, A.A.-D.; Investigation, A.A.-D., O.A., S.Y., N.N., A.D. and A.S.; Methodology, A.A.-D., O.A. and A.D.; Project administration, A.A.-D.; Resources, A.A.-D., O.A., S.Y., N.N., A.D. and A.S.; Software, A.A.-D., O.A., S.Y., N.N., A.D. and A.S.; Supervision, A.A.-D.; Validation, A.A.-D., O.A., S.Y., N.N., A.D. and A.S.; Visualization, A.A.-D.; Writing—original draft, A.A.-D., O.A., S.Y., N.N., A.D. and A.S.; Writing—review & editing, A.A.-D. All authors have read and agreed to the published version of the manuscript.

Funding

The authors acknowledge the Deanship of Scientific Research at King Faisal University for its financial support under the grant number: KFU242591.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Research Ethics Committee at King Faisal University (number: KFU-REC-2024-OCT-ETHICS2686, 2 October 2024).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The data are not available due to confidentiality concerns.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. McTear, M.; Ashurkina, M. Transforming Conversational AI: Exploring the Power of Large Language Models in Interactive Conversational Agents; Springer: Cham, Switzerland, 2024; pp. 1–256. [Google Scholar]
  2. Jho, H.; Ha, M. Towards Effective Argumentation: Design and Implementation of a Generative AI-Based Evaluation and Feedback System. J. Balt. Sci. Educ. 2024, 23, 280–291. [Google Scholar] [CrossRef]
  3. Du, Y.; Watkins, O.; Wang, Z.; Colas, C.; Darrell, T.; Abbeel, P.; Gupta, A.; Andreas, J. Guiding Pretraining in Reinforcement Learning with Large Language Models. In Proceedings of the International Conference on Machine Learning (ICML), Honolulu, HI, USA, 23–29 July 2023; PMLR: New York, NY, USA, 2023; pp. 8657–8677. [Google Scholar]
  4. Huang, H.; Zheng, O.; Wang, D.; Yin, J.; Wang, Z.; Ding, S.; Yin, H.; Xu, C.; Yang, R.; Zheng, Q.; et al. ChatGPT for Shaping the Future of Dentistry: The Potential of Multi-Modal Large Language Model. Int. J. Oral Sci. 2023, 15, 29. [Google Scholar] [CrossRef] [PubMed]
  5. Alqahtani, T.; Badreldin, H.A.; Alrashed, M.; Alshaya, A.I.; Alghamdi, S.S.; bin Saleh, K.; Alowais, S.A.; Alshaya, O.A.; Rahman, I.; Al Yami, M.S.; et al. The Emergent Role of Artificial Intelligence, Natural Language Processing, and Large Language Models in Higher Education and Research. Res. Soc. Adm. Pharm. 2023, 19, 1236–1242. [Google Scholar] [CrossRef] [PubMed]
  6. Agostini, D.; Picasso, F. Large Language Models for Sustainable Assessment and Feedback in Higher Education: Towards a Pedagogical and Technological Framework. In Proceedings of the First International Workshop on High-Performance Artificial Intelligence Systems in Education Co-Located with 22nd International Conference of the Italian Association for Artificial Intelligence (AIxIA 2023), Roma, Italy, 6 November 2023; CEUR Workshop Proceedings: Aachen, Germany, 2024. [Google Scholar]
  7. Kasneci, E.; Seßler, K.; Küchemann, S.; Bannert, M.; Dementieva, D.; Fischer, F.; Kasneci, G. ChatGPT for Good? On Opportunities and Challenges of Large Language Models for Education. Learn. Individ. Differ. 2023, 103, 102274. [Google Scholar] [CrossRef]
  8. Bewersdorff, A.; Hartmann, C.; Hornberger, M.; Seßler, K.; Bannert, M.; Kasneci, E.; Kasneci, G.; Zhai, X.; Nerdel, C. Taking the Next Step with Generative Artificial Intelligence: The Transformative Role of Multimodal Large Language Models in Science Education. arXiv 2024, arXiv:2401.00832. [Google Scholar]
  9. Perkins, M. Academic Integrity Considerations of AI Large Language Models in the Post-Pandemic Era: ChatGPT and Beyond. J. Univ. Teach. Learn. Pract. 2023, 20. [Google Scholar] [CrossRef]
  10. Chen, J.; Zhuo, Z.; Lin, J. Does ChatGPT Play a Double-Edged Sword Role in the Field of Higher Education? An In-Depth Exploration of the Factors Affecting Student Performance. Sustainability 2023, 15, 16928. [Google Scholar] [CrossRef]
  11. Alyoussef, I.Y. Massive Open Online Course (MOOCs) Acceptance: The Role of Task-Technology Fit (TTF) for Higher Education Sustainability. Sustainability 2021, 13, 7374. [Google Scholar] [CrossRef]
  12. Al-Maatouk, Q.; Othman, M.S.; Aldraiweesh, A.; Alturki, U.; Al-Rahmi, W.M.; Aljeraiwi, A.A. Task-Technology Fit and Technology Acceptance Model Application to Structure and Evaluate the Adoption of Social Media in Academia. IEEE Access 2020, 8, 78427–78440. [Google Scholar] [CrossRef]
  13. Alamri, M.M.; Almaiah, M.A.; Al-Rahmi, W.M. The Role of Compatibility and Task-Technology Fit (TTF): On Social Networking Applications (SNAs) Usage as Sustainability in Higher Education. IEEE Access 2020, 8, 161668–161681. [Google Scholar] [CrossRef]
  14. Goodhue, D.L.; Thompson, R.L. Task-Technology Fit and Individual Performance. MIS Q. Manag. Inf. Syst. 1995, 19, 213–233. [Google Scholar] [CrossRef]
  15. Al-Maroof, R.S.; Alhumaid, K.; Alshaafi, A.; Akour, I.; Bettayeb, A.; Alfaisal, R.; Salloum, S.A. A Comparative Analysis of ChatGPT and Google in Educational Settings: Understanding the Influence of Mediators on Learning Platform Adoption. In Artificial Intelligence in Education: The Power and Dangers of ChatGPT in the Classroom; Al-Marzouqi, A., Salloum, S.A., Al-Saidat, M., Aburayya, A., Gupta, B., Eds.; Studies in Big Data; Springer: Cham, Switzerland, 2024; Volume 144, pp. 431–448. [Google Scholar]
  16. Alyoussef, I.Y. Acceptance of E-Learning in Higher Education: The Role of Task-Technology Fit with the Information Systems Success Model. Heliyon 2023, 9, e13751. [Google Scholar] [CrossRef] [PubMed]
  17. Al-Rahmi, W.M.; Al-Adwan, A.S.; Al-Maatouk, Q.; Othman, M.S.; Alsaud, A.R.; Almogren, A.S.; Al-Rahmi, A.M. Integrating Communication and Task–Technology Fit Theories: The Adoption of Digital Media in Learning. Sustainability 2023, 15, 8144. [Google Scholar] [CrossRef]
  18. Zhou, Y.; Muresanu, A.I.; Han, Z.; Paster, K.; Pitis, S.; Chan, H.; Ba, J. Large Language Models Are Human-Level Prompt Engineers. arXiv 2022, arXiv:2211.01910. [Google Scholar]
  19. Lazarus, R.S. Cognition and Motivation in Emotion. Am. Psychol. 1991, 46, 352. [Google Scholar] [CrossRef]
  20. Lazarus, R.S. Progress on a Cognitive-Motivational-Relational Theory of Emotion. Am. Psychol. 1991, 46, 819. [Google Scholar] [CrossRef]
  21. Gursoy, D.; Chi, O.H.; Lu, L.; Nunkoo, R. Consumers Acceptance of Artificially Intelligent (AI) Device Use in Service Delivery. Int. J. Inf. Manag. 2019, 49, 157–169. [Google Scholar] [CrossRef]
  22. Almogren, A.S.; Al-Rahmi, W.M.; Dahri, N.A. Exploring Factors Influencing the Acceptance of ChatGPT in Higher Education: A Smart Education Perspective. Heliyon 2024, 10, e14315. [Google Scholar] [CrossRef]
  23. Sitar-Tăut, D.A. Mobile Learning Acceptance in Social Distancing during the COVID-19 Outbreak: The Mediation Effect of Hedonic Motivation. Hum. Behav. Emerg. Technol. 2021, 3, 366–378. [Google Scholar] [CrossRef]
  24. Baidoo-Anu, D.; Ansah, L.O. Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning. J. AI 2023, 7, 52–62. [Google Scholar] [CrossRef]
  25. Michel-Villarreal, R.; Vilalta-Perdomo, E.; Salinas-Navarro, D.E.; Thierry-Aguilera, R.; Gerardou, F.S. Challenges and Opportunities of Generative AI for Higher Education as Explained by ChatGPT. Educ. Sci. 2023, 13, 856. [Google Scholar] [CrossRef]
  26. Marvin, G.; Hellen, N.; Jjingo, D.; Nakatumba-Nabende, J. Prompt Engineering in Large Language Models. In International Conference on Data Intelligence and Cognitive Informatics 2023; Springer Nature Singapore: Singapore, 2023; pp. 387–402. [Google Scholar]
  27. Ivanashko, O.; Kozak, A.; Knysh, T.; Honchar, K. The Role of Artificial Intelligence in Shaping the Future of Education: Opportunities and Challenges. Futur. Educ. 2024, 4, 126–146. [Google Scholar]
  28. Cooper, G. Examining Science Education in ChatGPT: An Exploratory Study of Generative Artificial Intelligence. J. Sci. Educ. Technol. 2023, 32, 444–452. [Google Scholar] [CrossRef]
  29. Menon, D.; Shilpa, K. “Chatting with ChatGPT”: Analyzing the factors influencing users’ intention to use the Open AI’s ChatGPT using the UTAUT model. Heliyon 2023, 9, e20962. [Google Scholar] [CrossRef]
  30. Wang, L.; Ma, C.; Feng, X.; Zhang, Z.; Yang, H.; Zhang, J.; Chen, Z.; Tang, J.; Chen, X.; Lin, Y.; et al. A Survey on Large Language Model Based Autonomous Agents. Front. Comput. Sci. 2024, 18, 186345. [Google Scholar] [CrossRef]
  31. Li, H.; Wang, S.X.; Shang, F.; Niu, K.; Song, R. Applications of Large Language Models in Cloud Computing: An Empirical Study Using Real-World Data. Int. J. Innov. Res. Comput. Sci. Technol. 2024, 12, 59–69. [Google Scholar] [CrossRef]
  32. Jin, K.; Wang, C.Y.; Pham, H.V.; Hemmati, H. Can ChatGPT Support Developers? An Empirical Evaluation of Large Language Models for Code Generation. In Proceedings of the 2024 IEEE/ACM 21st International Conference on Mining Software Repositories (MSR), Lisbon, Portugal, 15–16 April 2024; IEEE: Piscataway, NJ, USA, 2024; pp. 167–171. [Google Scholar]
  33. Alyoussef, I.Y. E-Learning Acceptance: The Role of Task–Technology Fit as Sustainability in Higher Education. Sustainability 2021, 13, 6450. [Google Scholar] [CrossRef]
  34. Muchenje, G.; Seppänen, M. Unpacking task-technology fit to explore the business value of big data analytics. Int. J. Inf. Manag. 2023, 69, 102619. [Google Scholar] [CrossRef]
  35. Shin, A.; Song, H.D. The Effects of Basic Psychological Needs, Task–Technology Fit, and Student Engagement on MOOC Learners’ Continuance Intention to Use. Hum. Behav. Emerg. Technol. 2022, 1, 6444509. [Google Scholar] [CrossRef]
  36. Navarro, M.M.; Prasetyo, Y.T.; Young, M.N.; Nadlifatin, R.; Redi, A.A.N.P. The Perceived Satisfaction in Utilizing Learning Management System among Engineering Students during the COVID-19 Pandemic: Integrating Task Technology Fit and Extended Technology Acceptance Model. Sustainability 2021, 13, 10669. [Google Scholar] [CrossRef]
  37. Abdekhoda, M.; Dehnad, A.; Zarei, J. Factors Influencing Adoption of E-Learning in Healthcare: Integration of UTAUT and TTF Model. BMC Med. Inform. Decis. Mak. 2022, 22, 327. [Google Scholar] [CrossRef]
  38. Ma, X.; Huo, Y. Are Users Willing to Embrace ChatGPT? Exploring the Factors on the Acceptance of Chatbots from the Perspective of AIDUA Framework. Technol. Soc. 2023, 75, 102362. [Google Scholar] [CrossRef]
  39. Baabdullah, A.; Alalwan, A.; Rana, N.; Kizgin, H.; Patil, P. Consumer Use of Mobile Banking (m-Banking) in Saudi Arabia: Towards an Integrated Model. Int. J. Inf. Manag. 2019, 44, 38–52. [Google Scholar] [CrossRef]
  40. Sequeiros, H.; Oliveira, T.; Thomas, M. The Impact of IoT Smart Home Services on Psychological Well-Being. Inf. Syst. Front. 2021, 24, 1009–1026. [Google Scholar] [CrossRef]
  41. Kurdi, B.; Alshurideh, M.; Salloum, S. Investigating a Theoretical Framework for E-Learning Technology Acceptance. Int. J. Electr. Comput. Eng. 2020, 10, 6484–6496. [Google Scholar] [CrossRef]
  42. Raza, S.A.; Qazi, W.; Khan, K.A.; Salam, J. Social Isolation and Acceptance of the Learning Management System (LMS) in the Time of COVID-19 Pandemic: An Expansion of the UTAUT Model. J. Educ. Comput. Res. 2021, 59, 183–208. [Google Scholar] [CrossRef]
  43. Venkatesh, V. Determinants of Perceived Ease of Use: Integrating Control, Intrinsic Motivation, and Emotion into the Technology Acceptance Model. Inf. Syst. Res. 2000, 11, 342–365. [Google Scholar] [CrossRef]
  44. Davis, F.; Bagozzi, R.P.; Warshaw, P.R. Extrinsic and Intrinsic Motivation to Use Computers in the Workplace. J. Appl. Soc. Psychol. 1992, 22, 1111–1132. [Google Scholar] [CrossRef]
  45. Strzelecki, A. To Use or Not to Use ChatGPT in Higher Education? A Study of Students’ Acceptance and Use of Technology. Interact. Learn. Environ. 2023, 32, 5142–5155. [Google Scholar] [CrossRef]
  46. Wang, W.; Qiao, H.; Li, X.; Wang, J. User Willingness to Use Generative Artificial Intelligence Based on AIDUA Framework. J. Libr. Inf. Sci. Agric. 2024, 36, 36. [Google Scholar]
  47. Hengxuan, O.; Gürsoy, D.; Christina, G. Tourists’ Attitudes Toward the Use of Artificially Intelligent (AI) Devices in Tourism Service Delivery: Moderating Role of Service Value Seeking. J. Travel Res. 2020, 61, 170–185. [Google Scholar]
  48. Ribeiro, M.; Gürsoy, D.; Chi, O.H. Customer Acceptance of Autonomous Vehicles in Travel and Tourism. J. Travel Res. 2021, 61, 620–636. [Google Scholar] [CrossRef]
  49. Tran, A.Q.; Nguyen, L.H.; Nguyen, H.S.A.; Nguyen, C.T.; Vu, L.G.; Zhang, M.; Vu, T.M.T.; Nguyen, S.H.; Tran, B.X.; Latkin, C.A.; et al. Determinants of Intention to Use Artificial Intelligence-Based Diagnosis Support System among Prospective Physicians. Front. Public Health 2021, 9, 755644. [Google Scholar] [CrossRef] [PubMed]
  50. Schepart, A.; Burton, A.; Durkin, L.; Fuller, A.; Charap, E.; Bhambri, R.; Ahmad, F.S. Artificial Intelligence–Enabled Tools in Cardiovascular Medicine: A Survey of Current Use, Perceptions, and Challenges. Cardiovasc. Digit. Health J. 2023, 4, 101–110. [Google Scholar] [CrossRef] [PubMed]
  51. Sharma, S.; Islam, N.; Singh, G.; Dhir, A. Why Do Retail Customers Adopt Artificial Intelligence (AI) Based Autonomous Decision-Making Systems? IEEE Trans. Eng. Manag. 2022, 71, 1846–1861. [Google Scholar] [CrossRef]
  52. Lu, V.N.; Wirtz, J.; Kunz, W.H.; Paluch, S.; Gruber, T.; Martins, A.; Patterson, P.G. Service Robots, Customers, and Service Employees: What Can We Learn from the Academic Literature and Where Are the Gaps? J. Serv. Theory Pract. 2020, 30, 361–391. [Google Scholar] [CrossRef]
  53. Roy, P.; Ramaprasad, B.S.; Chakraborty, M.; Prabhu, N.; Rao, S. Customer Acceptance of Use of Artificial Intelligence in Hospitality Services: An Indian Hospitality Sector Perspective. Glob. Bus. Rev. 2024, 25, 832–851. [Google Scholar] [CrossRef]
  54. Emon, M.M.H.; Hassan, F.; Nahid, M.H.; Rattanawiboonsom, V. Predicting Adoption Intention of Artificial Intelligence. AIUB J. Sci. Eng. (AJSE) 2023, 22, 189–199. [Google Scholar] [CrossRef]
  55. Libunao, M.A.D. Adoption of Artificial Intelligence in the Games and Amusements Board: A Stepwise Multiple Linear Regression Analysis. Int. J. Bus. Technol. 2023, 5, 30–48. [Google Scholar]
  56. Al-Momani, A.M.; Mahmoud, M.A.; Ahmad, M.S. A Review of Factors Influencing Customer Acceptance of Internet of Things Services. Int. J. Inf. Syst. Serv. Sect. 2019, 11, 54–67. [Google Scholar] [CrossRef]
  57. Al-Azawei, A.; Alowayr, A. Predicting the Intention to Use and Hedonic Motivation for Mobile Learning: A Comparative Study in Two Middle Eastern Countries. Technol. Soc. 2020, 62, 101325. [Google Scholar] [CrossRef]
  58. Ali, I.; Warraich, N. Use and Acceptance of Technology with Academic and Digital Libraries Context: A Meta-Analysis of UTAUT Model and Future Direction. J. Librariansh. Inf. Sci. 2023, 57, 158–177. [Google Scholar] [CrossRef]
  59. Ilyoubi, B.; Yamin, M. The Impact of Task Technology Fit on Employee Job Performance. Mark. Manag. Innov. 2019, 9, 43–56. [Google Scholar]
  60. Rai, R.; Selnes, F. Conceptualizing Task-Technology Fit and the Effect on Adoption: A Case Study of a Digital Textbook Service. Inf. Manag. 2019, 56, 103–161. [Google Scholar] [CrossRef]
  61. Cai, J.; Li, Z.; Dou, Y.; Li, T. Investigating Adoption of High Prefabrication Level Technologies for Enterprises: An Integrated Model. Build. Res. Inf. 2023, 51, 648–666. [Google Scholar] [CrossRef]
  62. Wang, H.; Tao, D.; Yu, N.; Qu, X. Understanding Consumer Acceptance of Healthcare Wearable Devices: An Integrated Model of UTAUT and TTF. Int. J. Med. Inform. 2020, 139, 104–156. [Google Scholar] [CrossRef]
  63. Alkhwaldi, A.; Alobidyeen, B.; Abdulmuhsin, A.; Al-Okaily, M. Investigating the Antecedents of HRIS Adoption in Public Sector Organizations: Integration of UTAUT and TTF. Int. J. Organ. Anal. 2022, 31, 3251–3274. [Google Scholar] [CrossRef]
  64. Zaremohzzabieh, Z.; Roslan, S.; Mohamad, Z.; Ismail, I.; Jalil, H.; Ahrari, S. Influencing Factors in MOOCs Adoption in Higher Education: A Meta-Analytic Path Analysis. Sustainability 2022, 14, 8268. [Google Scholar] [CrossRef]
  65. Wanti, L.P.; Insan, H.F.M.; Prasetya, N.W.A. End User Satisfaction for Location Health Service Application with Analysis of Task Technology Fit. Lontar Komput. J. Ilm. Teknol. Inf. 2020, 11, 76. [Google Scholar] [CrossRef]
  66. Howard, M.; Rose, J. Refining and Extending Task-Technology Fit Theory: Creation of Two Task-Technology Fit Scales and Empirical Clarification of the Construct. Inf. Manag. 2019, 56, 103–134. [Google Scholar] [CrossRef]
  67. Al-Rahmi, A.; Shamsuddin, A.; Alismaiel, O. Task-Technology Fit Model: The Factors Affecting Students’ Academic Performance in Higher Education. Univers. J. Educ. Res. 2020, 8, 6831–6843. [Google Scholar] [CrossRef]
  68. Kang, S. The Effects of Metaverse Service Characteristics on Continuance Use Intention: A Task-Technology Fit Perspective. J. Korean Soc. Cult. Converg. 2022, 44, 87–98. [Google Scholar] [CrossRef]
  69. Mishra, U. Individual Performance and Mobile Banking: Role of Task Technology Fit Model. J. Knowl. Innov. 2023, 9, 18–27. [Google Scholar] [CrossRef]
  70. Park, C. Exploring a New Determinant of Task Technology Fit: Content Characteristics. J. Int. Technol. Inf. Manag. 2019, 27, 100–118. [Google Scholar] [CrossRef]
  71. Valaei, N.; Nikhashemi, S.; Bressolles, G.; Jin, H. A(n) (a)symmetric perspective towards task-technology-performance fit in mobile app industry. J. Enterp. Inf. Manag. 2019, 32, 887–912. [Google Scholar] [CrossRef]
  72. Kurniawati, D.; Aryati, A.; Masyhuri. Task Technology Fit and Corporate Entrepreneurship in Islamic Banking: Post-Merger Perspective. Int. J. Res. Bus. Soc. Sci. 2023, 12, 2910. [Google Scholar] [CrossRef]
  73. Sheng, H.; Wang, J.; Xu, Z. Factors Influencing Users’ Intention to Adopt Intelligent Customer Service: A Task-Technology Fit Perspective. In Proceedings of the 2022 International Conference on Computer Engineering and Artificial Intelligence (ICCEAI), Sanya, China, 16–18 December 2022; IEEE: New York, NY, USA, 2022; pp. 668–670. [Google Scholar]
  74. Lin, C.-T.; Huang, C.; Tseng, T.-C. Impact of Task-Technology Fit on Improving the Performance of Meituan Application. Sci. Soc. Res. 2022, 4, 134–138. [Google Scholar] [CrossRef]
  75. Wang, W.; Sari, M. Examining the Effect of the Task-Technology Fit of Game Mechanisms on Learning Outcomes in Online Gamification Platforms. J. Educ. Comput. Res. 2023, 61, 32–59. [Google Scholar] [CrossRef]
  76. Lo Presti, A.; De Rosa, A.; Viceconte, E. I Want to Learn More! Integrating Technology Acceptance and Task–Technology Fit Models for Predicting Behavioural and Future Learning Intentions. J. Workplace Learn. 2021, 33, 591–605. [Google Scholar] [CrossRef]
  77. Rakic, S.; Tasic, N.; Marjanovic, U.; Softic, S.; Lüftenegger, E.; Turcin, I. Student Performance on an E-Learning Platform: Mixed Method Approach. Int. J. Emerg. Technol. Learn. 2020, 15, 187–203. [Google Scholar] [CrossRef]
  78. Wongwatkit, C.; Panjaburee, P.; Srisawasdi, N.; Seprum, P. Moderating Effects of Gender Differences on the Relationships Between Perceived Learning Support, Intention to Use, and Learning Performance in a Personalized E-Learning. J. Comput. Educ. 2020, 7, 229–255. [Google Scholar] [CrossRef]
  79. Alkhawaja, M.I.; Halim, M.S.A.; Abuma Al-Rahmi, A.M.; Shamsuddin, A.; Wahab, E.; Al-Rahmi, W.M.; Alturki, U.; Aldraiweesh, A.; Almutairy, S. Integrating the Role of UTAUT and TTF Model to Evaluate Social Media Use for Teaching and Learning in Higher Education. Front. Public Health 2022, 10, 905968. [Google Scholar]
  80. Abumandil, M.S.S.; Al-Adwan, A.S. System Quality and Student’s Acceptance of the E-Learning System: The Serial Mediation of Perceived Usefulness and Intention to Use. Contemp. Educ. Technol. 2022, 14, ep350. [Google Scholar]
  81. Sahoo, O. An Empirical Investigation on Mobile Payment Adoption in India. Doctoral Dissertation, BITS Pilani, Pilani, Rajasthan, 2022. [Google Scholar]
  82. Hair, J.F.; Black, W.C.; Babin, B.J.; Anderson, R.E. Multivariate Data Analysis, 8th ed.; Pearson: Upper Saddle River, NJ, USA, 2019. [Google Scholar]
  83. Alanazi, A.; Frey, B.; Niileksela, C.; Lee, S.; Nong, A.; Alharbi, F. The Role of Task Value and Technology Satisfaction in Student Performance in Graduate-Level Online Courses. TechTrends 2020, 64, 922–930. [Google Scholar] [CrossRef]
  84. Sebastián, M.; Guede, J.; Antonovica, A. Application and Extension of the UTAUT2 Model for Determining Behavioral Intention Factors in Use of the Artificial Intelligence Virtual Assistants. Front. Psychol. 2022, 13, 993935. [Google Scholar]
  85. Lin, H.; Chi, O.; Gursoy, D. Antecedents of Customers’ Acceptance of Artificially Intelligent Robotic Device Use in Hospitality Services. J. Hosp. Mark. Manag. 2020, 29, 530–549. [Google Scholar] [CrossRef]
  86. Eldokhny, A.A.; Drwish, A.M. Effectiveness of Augmented Reality in Online Distance Learning at the Time of the COVID-19 Pandemic. Int. J. Emerg. Technol. Learn. 2021, 16, 198. [Google Scholar] [CrossRef]
  87. Al-Abdullatif, A.M.; Al-Dokhny, A.A.; Drwish, A.M. Implementing the Bashayer Chatbot in Saudi Higher Education: Measuring the Influence on Students’ Motivation and Learning Strategies. Front. Psychol. 2023, 14, 1129070. [Google Scholar] [CrossRef]
  88. Al-Dokhny, A.; Drwish, A.; Alyoussef, I.; Al-Abdullatif, A. Students’ Intentions to Use Distance Education Platforms: An Investigation into Expanding the Technology Acceptance Model Through Social Cognitive Theory. Electronics 2021, 10, 2992. [Google Scholar] [CrossRef]
  89. Al Abdullatif, A.; Al-Dokhny, A.; Drwish, A. Critical Factors Influencing Pre-Service Teachers’ Use of the Internet of Things (IoT) in Classrooms. Int. J. Interact. Mob. Technol. 2022, 16, 85. [Google Scholar] [CrossRef]
  90. Al-Dokhny, A.A.; Bukhamseen, A.M.; Drwish, A.M. Influence of Assistive Technology Applications on Dyslexic Students: The Case of Saudi Arabia During the COVID-19 Pandemic. Educ. Inf. Technol. 2022, 27, 12213–12249. [Google Scholar] [CrossRef]
  91. Drwish, A.M.; Al-Dokhny, A.A.; Al-Abdullatif, A.M.; Aladsani, H.K. A Sustainable Quality Model for Mobile Learning in Post-Pandemic Higher Education: A Structural Equation Modeling-Based Investigation. Sustainability 2023, 15, 7420. [Google Scholar] [CrossRef]
  92. Weisberg, H.F.; Krosnick, J.A.; Bowen, B.D. An Introduction to Survey Research and Data Analysis; Sage Publications: Thousand Oaks, CA, USA, 1977. [Google Scholar]
  93. Hair, J.F.; Risher, J.J.; Sarstedt, M.; Ringle, C.M. When to use and how to report the results of PLS-SEM. Eur. Bus. Rev. 2019, 31, 2–24. [Google Scholar] [CrossRef]
  94. Hair, J.F.; Hult, G.T.M.; Ringle, C.; Sarstedt, M. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM); SAGE Publications: Thousand Oaks, CA, USA, 2016. [Google Scholar]
  95. Hu, L.; Bentler, P.M. Fit Indices in Covariance Structure Modeling: Sensitivity to Underparameterized Model Misspecification. Psychol. Methods 1998, 3, 424–453. [Google Scholar] [CrossRef]
  96. Hair, J.F., Jr.; Hult, G.T.M.; Ringle, C.M.; Sarstedt, M.; Danks, N.P.; Ray, S. Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R: A Workbook; Springer Nature: The Hague, The Netherlands, 2021. [Google Scholar]
  97. Byrne, B.M. Structural Equation Modeling with Mplus: Basic Concepts, Applications, and Programming; Routledge: London, UK, 2013. [Google Scholar]
  98. Chin, W.W. The Partial Least Squares Approach to Structural Equation Modeling. Mod. Methods Bus. Res. 1998, 295, 295–336. [Google Scholar]
  99. Lei, M.; Lomax, R.G. The Effect of Varying Degrees of Non-Normality in Structural Equation Modeling. Struct. Equ. Model. 2005, 12, 1–27. [Google Scholar] [CrossRef]
  100. Padiyath, A.; Hou, X.; Pang, A.; Viramontes Vargas, D.; Gu, X.; Nelson-Fromm, T.; Wu, Z.; Guzdial, M.; Ericson, B. Insights from Social Shaping Theory: The Appropriation of Large Language Models in an Undergraduate Programming Course. In Proceedings of the 2024 ACM Conference on International Computing Education Research, Dunedin, New Zealand, 7–10 August 2024; ACM: New York, NY, USA, 2024; pp. 114–130. [Google Scholar]
  101. An, X.; Chai, C.S.; Li, Y.; Zhou, Y.; Yang, B. Modeling Students’ Perceptions of Artificial Intelligence Assisted Language Learning. Comput. Assist. Lang. Learn. 2023, 1–22. [Google Scholar] [CrossRef]
  102. Rizkalla, N.; Tannady, H.; Bernando, R. Analysis of the Influence of Performance Expectancy, Effort Expectancy, Social Influence, and Attitude Toward Behavior on Intention to Adopt live.on. Multidiscip. Rev. 2023, 6, 2023spe017. [Google Scholar] [CrossRef]
  103. Camilleri, M.A. Factors Affecting Performance Expectancy and Intentions to Use ChatGPT: Using SmartPLS to Advance an Information Technology Acceptance Framework. Technol. Forecast. Soc. Change 2024, 201, 123247. [Google Scholar] [CrossRef]
  104. Tamilmani, K.; Rana, N.; Prakasam, N.; Dwivedi, Y. The Battle of Brain vs. Heart: A Literature Review and Meta-Analysis of “Hedonic Motivation” Use in UTAUT2. Int. J. Inf. Manag. 2019, 46, 222–235. [Google Scholar] [CrossRef]
  105. Jain, R.; Garg, N.; Khera, S.N. Adoption of AI-Enabled Tools in Social Development Organizations in India: An Extension of UTAUT Model. Front. Psychol. 2022, 13, 893691. [Google Scholar] [CrossRef] [PubMed]
  106. Butt, S.; Mahmood, A.; Saleem, S.; Rashid, T.; Ikram, A. Students’ Performance in Online Learning Environment: The Role of Task Technology Fit and Actual Usage of System During COVID-19. Front. Psychol. 2021, 12, 759227. [Google Scholar] [CrossRef] [PubMed]
  107. Khan, A.K.; Quratulain, S. An Examination of Antecedents and Consequences of Technostress Among University Students: Task-Technology Fit Perspective. In Proceedings of the 36th Bled eConference, BLED 2023, Bled, Slovenia, 25–28 June 2023; University of Maribor Press: Maribor, Slovenia, 2023; pp. 827–834. [Google Scholar]
  108. Zhang, X.; Jiang, H.; Qiao, Z.; Li, P. Students’ Response to ChatGPT: An Adaptive Technology-To-Performance Model. J. Comput. Inf. Syst. 2024, 1–18. [Google Scholar] [CrossRef]
  109. Chuyen, N.T.H. An Empirical Analysis of Predictors of AI-Powered Design Tool Adoption. TEM J. 2023, 12, 1482. [Google Scholar] [CrossRef]
  110. Al-Emran, M. Evaluating the Use of Smartwatches for Learning Purposes through the Integration of the Technology Acceptance Model and Task-Technology Fit. Int. J. Hum.-Comput. Interact. 2021, 37, 1874–1882. [Google Scholar] [CrossRef]
Sustainability 16 10780 g001
Figure 1. Research model.
Figure 1. Research model.
Sustainability 16 10780 g001
Sustainability 16 10780 g002
Figure 2. Path coefficient results.
Figure 2. Path coefficient results.
Sustainability 16 10780 g002
Sustainability 16 10780 g003
Figure 3. Path (t-value) results.
Figure 3. Path (t-value) results.
Sustainability 16 10780 g003
Table 1. Recent empirical and qualitative studies on integrating MLLM technology to promote teaching and learning.
Table 1. Recent empirical and qualitative studies on integrating MLLM technology to promote teaching and learning.
AuthorsPurpose of the StudyApproach/Sample and ContextFindings
[25], 2023, United Kingdom, cited by 316.The study used ethnographic methods, including interviews and textual analysis, to understand perspectives about ChatGPT in higher education.An ethnographic methodology was employed to investigate the application of ChatGPT as MLLMs in higher education.The research highlighted the importance of developing robust policies, guidelines, and frameworks to ensure the responsible incorporation of ChatGPT into higher education.
[24], 2023, Canda, cited by 2036.The research examined both the advantages and challenges of using ChatGPT in higher education, focusing on how GAI can improve teaching and learning methodologiesAn exploratory methodology was employed to investigate ChatGPT as an MLLM.The study highlighted ChatGPT’s potential benefits in education, such as personalization, interactive content, and feedback, but also identified risks, such as misinformation, bias, and privacy issues.
[28], 2023, Australia, cited by 750.The study focused on three primary research objectives: analyzing ChatGPT’s responses to science education inquiries, investigating strategies for integrating ChatGPT into science instruction, and assessing ChatGPT’s utility as a research tool.An exploratory methodology was employed to investigate the application of science education in ChatGPT as an MLLM.The study concluded that ChatGPT can be a helpful tool for educators. However, it is important for teachers to critically evaluate and adapt the generated content to suit their specific needs and students. ChatGPT should be used as a supplement, not a replacement, for teachers’ expertise.
[22], 2024, Kingdom of Saudi Arabia (KSA), cited by 20.The study aimed to understand why users accept ChatGPT as a learning tool in higher education. The technology acceptance model (TAM) and other relevant factors were used.A quantitative approach was employed to investigate the adoption of ChatGPT for smart education among 458 university students.The study found that ease of use and usefulness are key factors influencing how users feel about ChatGPT in smart education. The quality of feedback, assessments, and adherence to subject norms also positively influence users’ intentions to use ChatGPT. In addition, a strong relationship was found between how users feel about ChatGPT and how much they use it.
[29], 2023, India, cited by 116.The study investigated user acceptance of and interaction with ChatGPT through the lens of the unified theory of acceptance and use of technology (UTAUT) model.Qualitative interpretive research methods were employed to investigate the factors influencing the intentions of 32 Indian ChatGPT users to adopt OpenAI’s ChatGPT, using the UTAUT model.The study found that the four main factors from the UTAUT model, along with perceived interactivity and privacy concerns, significantly influence how users engage with ChatGPT. Age and experience can also affect how these factors impact ChatGPT usage.
[30], 2024, China, cited by 718.This paper focused on augmenting LLMs with human values and preferences.The methodology employed involved fine-tuning, prompting, and mechanism engineering to enhance agent capabilities.A survey was conducted to establish taxonomies for LLM-based autonomous agents across various domains. Furthermore, the study identified challenges and explored potential future directions for the development and application of LLM-based autonomous agents.
[31], 2024, United States of America, cited by 34.This research explored the practical applications of LLMs in cloud computing environments.Bayesian inference was employed for resource allocation in cloud computing environments, while Markov decision processes were utilized to enhance predictive accuracy and decision-making efficiency.LLMs demonstrated enhancements in central processing unit (CPU) utilization, memory usage, network latency, and storage performance. Visual aids were employed to illustrate key findings, validating the integration of AI within cloud services.
[32], 2024, Portugal, cited by 9.This study analyzed developer interactions with ChatGPT for code generation purposes. It evaluated the usefulness of ChatGPT-generated code for developers and explored the practical utilization of ChatGPT in real-world coding practices.Automated data-cleaning techniques were employed to filter out irrelevant prompts from conversations. Manual labeling of conversation rounds was conducted to determine the frequency of code snippet usage.The findings revealed a limited adoption of LLM-generated code in production environments. Future research is necessary to further refine LLMs for practical software development applications.
Table 2. Recent T-TF model studies in the field of interest.
Table 2. Recent T-TF model studies in the field of interest.
AuthorsPurpose of the StudyApproach/Sample and ContextFindings
[33], 2021, KSA, cited by 66.This study explored the impact of T-TF on student satisfaction and performance in e-learning settings. Using a sample of 432 university students from public institutions, the findings revealed that a strong T-TF significantly improves academic outcomes and satisfaction with e-learning platforms, supporting the sustainability of educational practices.The sample comprised 432 university students from public institutions.Strong T-TF improves academic performance and satisfaction with e-learning platforms, ensuring sustainability in education processes.
[34], 2023, Finland, cited by 28.This study investigated the alignment between task characteristics and the capabilities of big data analytics (BDA) to assess their impact on business value. The research framework was grounded in the T-TF theory.Case studies of successful BDA implementations can illuminate key success factors. Surveys and interviews with BDA practitioners can uncover perceived benefits and challenges. Quantitative analysis of BDA data can identify patterns and trends.The primary finding of this study is a 2 × 2 matrix framework that elucidates the relationship between task reconfigurability and BDA editability in determining task-BDA fit. Future research should investigate how this relationship evolves over time.
[35], 2022, South Korea, cited by 4.This study examined the factors influencing massive open online course (MOOC) learners’ continued use of educational technology. It explored the roles of basic psychological needs, T-TF, and student engagement in shaping learners’ intentions.A sample of 201 Korean MOOC learners was studied. T-TF and student engagement were assessed using adapted measures. Structural equation modeling was employed to evaluate the proposed model. A mediation significance test with phantom variables was conducted.The research found that although basic psychological needs shape student engagement, they do not have a direct effect on continuance intention. In contrast, T-TF directly influences both student engagement and continuance intention. Additionally, student engagement acts as a mediator between basic psychological needs, T-TF, and continuance intention. Despite the scalability and flexibility of massive open online courses (MOOCs), they still face challenges with high dropout rates.
[36], 2021, Philippines, cited by 78.This study investigated the factors influencing the satisfaction of engineering students using learning management systems (LMS) during the COVID-19 pandemic. By integrating the T-TF and TAM frameworks, it aimed to evaluate student satisfaction with LMS. In addition, it explored specific factors affecting perceived satisfaction among engineering students in the Philippines.A total of 1011 engineering students participated in the online survey. Structural equation modeling (SEM) was used to evaluate the factors influencing perceived student satisfaction. The study tested hypotheses based on the T-TF and TAM frameworks.The study revealed that T-TF positively influences learners’ intentions to use an LMS. Perceived usefulness and ease of use were identified as significant predictors of behavioral intention. Additionally, social presence and the content quality of the LMS were found to be positively correlated with perceived satisfaction.
[37], 2022, Iran, cited by 16.The study focused on uncovering the factors that influenced e-learning adoption in healthcare during the COVID-19 pandemic. By combining the UTAUT and T-TF models, it aimed to understand faculty intentions toward adopting e-learning. Moreover, it analyzed how technology and task characteristics affected e-learning adoption in the healthcare sector.The study utilized a descriptive-analytical research design involving 143 faculty members from Iran. By integrating the Unified Theory UTAUT and T-TF models, it aimed to explain the adoption of e-learning among Iranian faculty members.The study revealed a positive correlation between technology characteristics and task characteristics. T-TF was found to be significantly correlated with the UTAUT model. Moreover, all hypothesized paths within the model were significant and aligned with the expected direction.
Table 3. The relationship between proposed factors and their corresponding hypotheses.
Table 3. The relationship between proposed factors and their corresponding hypotheses.
FactorDescriptionHypothesis LinkEffect Direction
Social InfluenceThe more individuals’ expectations about the ease of using MLLMs increase due to their surrounding social influence, the less effort they anticipate needing when using them in the future.H1− →* Effort ExpectancyNegative (−)
Hedonic MotivationThe more individuals’ expectations about the hedonic benefits of using MLLMs increase due to their surrounding social influence, the less effort they anticipate needing when using them in the future.H2− → Effort ExpectancyNegative (−)
Effort ExpectancyThe higher the individuals’ perceived effort required to use MLLMs, the lower the T-TF.H3− → T-TFNegative (−)
Task CharacteristicsTask characteristics positively influence the T-TF.H4 → T-TFPositive (+)
Technology CharacteristicsTechnology characteristics positively influence the T-TF.H5 → T-TFPositive (+)
T-TFT-TF positively impacts performance benefits for MLLMs.H6 → Performance Benefits for MLLMs.Positive (+)
T-TFT-TF positively influences willingness to accept the use of MLLMs.H7 → Willingness to accept the use of MLLMs.Positive (+)
Willingness to Accept the Use of MLLMsWillingness to accept the use of MLLMs positively influences performance benefits for MLLMs.H8 → Willingness to accept the use of MLLMsPositive (+)
* The sign of → means that the construct in the “factor column” affects the construct in the “Hypothesis Link column”.
Table 4. The study’s sampled demographic data through descriptive statistics (n = 550).
Table 4. The study’s sampled demographic data through descriptive statistics (n = 550).
ItemNumber and PercentageMeanStandard Deviation
GenderMale230 (41.8%)3.260.67
Female320 (58.2%)
Age≤20100 (18.2%)4.160.89
21:25200 (36.4%)
26:30200 (36.4%)
>3050 (9.1%)
FacultyEducation300 (54.6%)2.742.61
Arts50 (9.1%)
Engineering80 (14.6%)
Nursing70 (12.7%)
Other50 (9.1%)
Academic MajorScientific197 (35.8%)2.431.35
Literary353 (64.2%)
StageUndergraduate250 (45.5%)2.311.44
Postgraduate300 (54.6%)
Table 5. The instrument factors’ reliability in the study.
Table 5. The instrument factors’ reliability in the study.
FactorsReliability
Number of ItemsSourceReliability
(Cronbach’s Alpha)
Social Influence (SI)6 items[19,20,21]0.775
Hedonic Motivation (HM)4 items[19,20,21,43]0.875
Effort Expectancy (EE)5 items[19,20,21]0.882
Task Characteristics (TC)5 items[14,34,35]0.935
Technology Characteristics (TeC)10 items[14,83]0.989
T-TF7 items[11,12,14]0.975
Willingness to Accept the Use of MLLMs (WAUMLLMs)5 items[21,84,85,86,87]0.983
Performance Benefits for MLLMs (PBMLLMs)6 items[73,74,78,79,80,81,82,83,84,85,86,87,88,89,90,91]0.996
Table 6. The loading of items and cross-loadings.
Table 6. The loading of items and cross-loadings.
FactorsItemsSIHMEETCTeCT-TFWAUMLLMsPBMLLMs
Social Influence (SI)SI10.8560.6120.5790.6060.4480.3660.4520.672
SI20.801 0.5220.5610.5030.5910.5850.6320.575
SI30.896 0.6540.5200.6130.4250.5940.6890.664
SI40.825 0.4400.5260.5510.4730.4900.5700.649
SI50.809 0.4120.4240.4500.3980.4870.5220.476
SI60.788 0.5320.5320.5860.3630.5690.5920.576
Hedonic Motivation (HM)HM10.4820.73603360.4560.4340.5410.5860.449
HM20.4640.811 0.5120.6530.6310.6320.7360.620
HM30.4590.897 0.5760.6760.5380.4560.7360.538
HM40.4980.765 0.4760.5670.3390.3960.7360.333
Effort Expectancy (EE)EE10.581 0.4180.8760.688 0.391 0.486 0.581 0.678
EE20.681 0.3530.883 0.622 0.424 0.328 0.614 0.525
EE30.462 0.3810.891 0.566 0.492 0.5660.662 0.562
EE40.3950.5640.885 0.5910.3990.3350.5970.305
EE50.569 0.6190.871 0.460 0.678 0.345 0.578 0.660
Task Characteristics (TC)TC10.6120.4180.6140.7770.4830.4370.4540.411
TC20.5390.3330.3400.892 0.5330.3930.5340.603
TC30.5680.3610.3770.836 0.3610.4600.5010.499
TC40.5670.5600.4420.829 0.5940.5140.5200.590
TC50.6940.6100.5950.886 0.6110.6890.6340.690
Technology Characteristics (TeC)TeC10.3230.5180.3190.3450.9760.4370.4150.417
TeC20.5300.4440.3500.4530.808 0.4290.4730.484
TeC30.3410.5610.3910.5340.886 0.4470.3910.479
TeC40.5980.6520.5800.3540.955 0.3770.5600.446
TeC50.3930.4120.4100.4350.809 0.4370.5860.483
TeC60.4910.4190.6180.4660.788 0.4370.5040.457
TeC70.3320.6660.4930.5550.9560.4370.5810.333
TeC80.6210.3980.4640.3210.801 0.4270.3610.636
TeC90.5060.5510.5900.4980.798 0.5000.5600.546
TeC100.6090.3400.4650.6120.895 0.5200.4250.335
T-TFT-TF10.624 0.6480.5180.4910.3770.9360.6840.324
T-TF20.592 0.3910.5550.5610.4800.888 0.5540.319
T-TF30.4990.4280.6580.6860.6790.873 0.5520.486
T-TF40.478 0.5730.3400.624 0.5200.923 0.4430.576
T-TF50.3210.3900.4120.312 0.4830.799 0.5930.560
T-TF60.3980.5630.3320.5970.3800.898 0.4500.598
T-TF70.4120.6480.4180.37800.3830.8580.5330.393
Willingness to Accept the Use of MLLMs (WAUMLLMs)WAUMLLMs10.4210.6160.3670.688 0.4320.4040.8870.516
WAUMLLMs20.5060.5140.4770.390 0.4370.5040.897 0.317
WAUMLLMs30.4090.3130.4440.423 0.3390.3040.985 0.428
WAUMLLMs40.5190.4780.448 0.4370.5040.5180.982 0.422
WAUMLLMs50.3430.4660.508 0.6770.4840.3180.889 0.310
Performance Benefits for MLLMs (PBMLLMs)PBMLLMs10.5230.6570.3160.4770.6370.5270.4110.982
PBMLLMs20.470 0.4030.3400.424 0.3200.323 0.5030.905
PBMLLMs30.3980.4630.5320.6070.3300.498 0.5320.876
PBMLLMs40.4180.6230.5420.5270.5400.608 0.4500.795
PBMLLMs50.3020.5030.5300.5010.3700.432 0.6540.869
PBMLLMs60.5060.4920.4920.4970.3200.491 0.5500.983
Table 7. Reliability and convergent validity analysis.
Table 7. Reliability and convergent validity analysis.
FactorsItemsFactor LoadingCACRAVER2
Social Influence (SI)SI10.8560.7940.8960.698
SI20.801
SI30.896
SI40.825
SI50.809
SI60.788
Hedonic Motivation (HM)HM10.7360.8250.9340.784
HM20.811
HM30.897
HM40.765
Effort Expectancy (EE)EE10.8760.9720.8800.869
EE20.883
EE30.891
EE40.885
EE50.871
Task Characteristics (TC)TC10.7770.9950.9560.852
TC20.892
TC30.836
TC40.829
TC50.886
Technology Characteristics (TeC)TeC10.9760.8970.9270.731
TeC20.808
TeC30.886
TeC40.955
TeC50.809
TeC60.788
TeC70.976
TeC80.808
TeC90.886
TeC100.955
T-TFT-TF10.9360.8590.8990.7690.294
T-TF20.888
T-TF30.873
T-TF40.923
T-TF50.799
T-TF60.898
T-TF70.858
Willingness to Accept the Use of MLLMs (WAUMLLMs)WAUMLLMs10.8870.9820.9530.8810.487
WAUMLLMs20.897
WAUMLLMs30.985
WAUMLLMs40.982
WAUMLLMs50.889
Performance Benefits for MLLMs (PBMLLMs)PBMLLMs10.9820.8470.8740.7430.560
PBMLLMs20.905
PBMLLMs30.876
PBMLLMs40.795
PBMLLMs50.869
PBMLLMs60.983
Table 8. Discriminant validity.
Table 8. Discriminant validity.
FactorsSIHMEETCTeCT-TFWAUMLLMsPBMLLMs
Social Influence (SI)0.853
Hedonic Motivation (HM)0.4270.783
Effort Expectancy (EE)0.3520.3990.721
Task Characteristics (TC)0.2840.2500.3280.811
Technology Characteristics (TeC)0.3150.3710.2940.4700.752
T-TF0.4570.4920.4170.3910.5830.790
Willingness to Accept the Use of MLLMs (WAUMLLMs)0.6290.6540.5880.5240.6130.6830.824
Performance Benefits for MLLMs (PBMLLMs)0.5820.5260.4720.4150.4980.5510.6740.762
Table 9. Additional validity discriminant measurement results based on HTMT.
Table 9. Additional validity discriminant measurement results based on HTMT.
FactorsSIHMEETCTeCT-TFWAUMLLMs
Social Influence (SI)
Hedonic Motivation (HM)0.65
Effort Expectancy (EE)0.580.79
Task Characteristics (TC)0.720.690.66
Technology Characteristics (TeC)0.490.560.620.54
T-TF0.610.710.770.630.70
Willingness to Accept the Use of MLLMs (WAUMLLMs)0.680.750.800.780.670.74
Performance Benefits for MLLMs (PBMLLMs)0.730.820.810.820.710.760.83
Table 10. Descriptive statistic for samples’ answers.
Table 10. Descriptive statistic for samples’ answers.
FactorsItemsMeanStd. Dev.SkewnessKurtosis
Social Influence (SI)SI13.430.78−0.123.02
SI23.030.540.452.87
SI33.111.110.233.19
SI43.031.000.10 3.11
SI53.640.55−0.34 3.18
SI63.310.78 −0.343.16
Hedonic Motivation (HM)HM13.511.41 −0.443.04
HM23.910.740.37 3.17
HM33.250.640.10 2.84
HM43.410.990.21 2.88
Effort Expectancy (EE)EE13.761.49−0.482.82
EE23.230.740.47 2.93
EE33.081.170.332.96
EE43.291.26−0.292.91
EE53.160.74−0.323.13
Task Characteristics (TC)TC13.931.23−0.322.94
TC23.810.87−0.202.91
TC33.631.130.023.02
TC43.871.13−0.072.86
TC53.801.04−0.213.12
Technology Characteristics (TeC)TeC13.190.590.112.83
TeC23.891.34−0.363.19
TeC33.540.82−0.213.11
TeC43.810.69−0.132.88
TeC53.900.54−0.042.80
TeC63.321.090.293.13
TeC73.111.18−0.303.08
TeC83.230.520.013.09
TeC93.431.010.093.11
TeC103.820.73−0.452.83
T-TFT-TF13.861.150.11 2.94
T-TF23.010.67−0.332.85
T-TF33.511.19−0.433.15
T-TF43.420.890.45 3.05
T-TF53.221.440.47 2.93
T-TF63.120.640.31 2.83
T-TF73.340.84−0.202.92
Willingness to Accept the Use of MLLMs (WAUMLLMs)WAUMLLMs13.940.61−0.402.93
WAUMLLMs23.321.420.18 3.09
WAUMLLMs33.521.38−0.063.06
WAUMLLMs43.700.76−0.383.15
WAUMLLMs53.361.16−0.002.99
Performance Benefits for MLLMs (PBMLLMs)PBMLLMs13.971.32−0.47 2.85
PBMLLMs23.961.06 0.41 3.09
PBMLLMs33.251.03 −0.243.10
PBMLLMs43.500.74 0.16 3.02
PBMLLMs53.300.59 −0.193.11
PBMLLMs63.431.400.02 3.00
Table 11. Hypothesis testing.
Table 11. Hypothesis testing.
HIndependent VariablePathDependent VariablePath Coefficient (Β)Standard Error (SE)t-ValueDecision
H1Social Influence (SI)* Sustainability 16 10780 i001Effort Expectancy (EE)−0.8520.051−17.214Supported
H2Hedonic Motivation (HM)Sustainability 16 10780 i001Effort Expectancy (EE)−0.7230.062−12.188Supported
H3Effort Expectancy (EE)Sustainability 16 10780 i001T-TF−0.5580.094−5.812Supported
H4Task Characteristics (TC)Sustainability 16 10780 i001T-TF0.6480.0715.618Supported
H5Technology Characteristics (TeC)Sustainability 16 10780 i001T-TF0.6920.0834.952Supported
H6T-TFSustainability 16 10780 i001Performance Benefits for MLLMs (PBMLLMs)0.5150.1055.104Supported
H7T-TFSustainability 16 10780 i001Willingness to Accept the Use of MLLMs (WAUMLLMs)0.5550.1264.594Supported
H8Willingness to Accept the Use of MLLMs (WAUMLLMs)Sustainability 16 10780 i001Performance Benefits for MLLMs (PBMLLMs)
ss		0.5680.1006.344Supported
* The sign of Sustainability 16 10780 i001 means that the construct in the “Independent Variable column” affects the construct in the “Dependent Variable column”.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Al-Dokhny, A.; Alismaiel, O.; Youssif, S.; Nasr, N.; Drwish, A.; Samir, A. Can Multimodal Large Language Models Enhance Performance Benefits Among Higher Education Students? An Investigation Based on the Task–Technology Fit Theory and the Artificial Intelligence Device Use Acceptance Model. Sustainability 2024, 16, 10780. https://doi.org/10.3390/su162310780

AMA Style

Al-Dokhny A, Alismaiel O, Youssif S, Nasr N, Drwish A, Samir A. Can Multimodal Large Language Models Enhance Performance Benefits Among Higher Education Students? An Investigation Based on the Task–Technology Fit Theory and the Artificial Intelligence Device Use Acceptance Model. Sustainability. 2024; 16(23):10780. https://doi.org/10.3390/su162310780

Chicago/Turabian Style

Al-Dokhny, Amany, Omar Alismaiel, Samia Youssif, Nermeen Nasr, Amr Drwish, and Amira Samir. 2024. "Can Multimodal Large Language Models Enhance Performance Benefits Among Higher Education Students? An Investigation Based on the Task–Technology Fit Theory and the Artificial Intelligence Device Use Acceptance Model" Sustainability 16, no. 23: 10780. https://doi.org/10.3390/su162310780

APA Style

Al-Dokhny, A., Alismaiel, O., Youssif, S., Nasr, N., Drwish, A., & Samir, A. (2024). Can Multimodal Large Language Models Enhance Performance Benefits Among Higher Education Students? An Investigation Based on the Task–Technology Fit Theory and the Artificial Intelligence Device Use Acceptance Model. Sustainability, 16(23), 10780. https://doi.org/10.3390/su162310780

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop