Business Finance - Management ASSIGNMENT (APA, NO PLAGARISM, GREAT WORK, ON TIME)

IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024 1341

ChatGPT for Educational Purposes: Investigating the Impact of Knowledge Management Factors on

Student Satisfaction and Continuous Usage Thi Thuy An Ngo , Thanh Tu Tran , Gia Khuong An , and Phuong Thy Nguyen

Abstract—The growing prevalence of advanced generative ar- tificial intelligence chatbots, such as ChatGPT, in the educational sector has raised considerable interest in understanding their im- pact on student knowledge and exploring effective and sustainable implementation strategies. This research investigates the influence of knowledge management factors on the continuous usage of ChatGPT for educational purposes while concurrently evaluating student satisfaction with its use in learning. Using a quantitative approach, a structured questionnaire was administered to 513 Viet- namese university students via Google Forms for data collection. The partial least squares structural equation modeling statistical technique was employed to examine the relationships between iden- tified factors and evaluate the research model. The results provided strong support for several hypotheses, revealing significant positive effects of expectation confirmation on perceived usefulness and satisfaction, as well as perceived usefulness on user satisfaction and continuous usage of ChatGPT. These findings suggest that when students recognize the usefulness of ChatGPT for their learnings, they experience higher satisfaction and are more likely to continue using it. In addition, knowledge acquisition significantly impacts both satisfaction and continuous usage of ChatGPT, while knowl- edge sharing and application influence satisfaction exclusively. This indicates that students prioritize knowledge acquisition over sharing and applying knowledge through ChatGPT. The study has theoretical and practical implications for ChatGPT developers, educators, and future research. Theoretically, it contributes to understanding satisfaction and continuous usage in educational settings, utilizing the expectation confirmation model and integrat- ing knowledge management factors. Practically, it provides insights into comprehension and suggestions for enhancing user satisfaction and continuous usage of ChatGPT in education.

Index Terms—ChatGPT, continuous usage (CUS), generative artificial intelligence (AI), knowledge management (KM), satisfac- tion.

I. INTRODUCTION

ARTIFICIAL intelligence (AI) plays an increasingly promi- nent role in the field of education [1]. Among AI-based

tools, chatbots are being evaluated as a valuable educational technology to assist learning [2], [3]. According to a study by

Manuscript received 31 August 2023; revised 22 December 2023 and 16 March 2024; accepted 23 March 2024. Date of publication 1 April 2024; date of current version 11 April 2024. (Corresponding author: Thi Thuy An Ngo.)

The authors are with the Department of Business, FPT University, 94100 Can Tho, Vietnam (e-mail: [email protected]; [email protected]; khuon- [email protected]; [email protected]).

Digital Object Identifier 10.1109/TLT.2024.3383773

Wang et al. [4], chatbots hold great potential in transforming the methods by which individuals acquire knowledge and seek information. The research also revealed the utilization of chat- bots within the educational sector, highlighting the consider- able prospects for enhancing the learning process and learning outcomes through their implementation. The effects of various types of chatbots on learning outcomes and processes have been found to differ [5]. Therefore, it is crucial to carefully select and implement chatbots that can bring the most advantages in a specific educational environment. Among the generative AI chatbots available today, ChatGPT stands out as an advanced ap- plication developed by OpenAI that generates text in response to user input, as mentioned by Halaweh [6]. Since its introduction, this cutting-edge tool has gained immense popularity, garnering a substantial user base across the globe. In education, ChatGPT offers students and educators a range of new possibilities, such as tailored feedback, enhanced accessibility, interactive conver- sations, lesson preparation, and more [7].

There is a growing interest in finding innovative and sus- tainable approaches for delivering and managing knowledge. ChatGPT, through its ability to engage learners in natural and meaningful conversations on various subjects, has emerged as a promising tool that can enhance learner satisfaction. According to Liu et al. [8], ChatGPT is frequently utilized in the educational sector for question and answer testing; it serves as a tool for learners to study, cross-check, and validate answers across a variety of academic disciplines, including physics, mathematics, and chemistry, as well as conceptual subjects, such as phi- losophy and religion. Moreover, users can pose open-ended and analytical queries to explore the capabilities of ChatGPT [8]. As a result, ChatGPT can help students develop critical thinking and problem-solving skills. It can facilitate adaptive learning, provide personalized feedback, support research and data analysis, offer automated administrative services, and aid in developing innovative assessments. It can answer questions, assist with doubts, prepare for exams, provide writing assistance, and offer personalized tutoring. It can also interact with students and solve queries rapidly, enhancing the learning experiences which affect their outcomes. According to An et al. [9], educa- tors use ChatGPT’s function to provide customized academic learning, suggestions, and guidance to each student. Its access to an extensive database enables it to assess assignments, correct errors, and provide feedback that accurately targets specific

1939-1382 © 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

1342 IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024

mistakes. Notably, more and more students use ChatGPT as an assistant whose support is available without any time or location constraints, setting it apart from conventional educational envi- ronments. In addition, ChatGPT’s abilities include analyzing students’ past performances, writing styles, and preferences across previous courses [9]. Also, in the study by An et al. [9], the authors proposed creating a teaching mode using ChatGPT as an assistant because it is believed that ChatGPT has benefits and is a great tool to become an assistant; it can innovate and facilitate better teaching and overcome the disadvantages of traditional education. Despite the potential benefits, Bahrini et al. [10] urged caution and emphasized the importance of being mindful of potential threats and implementing necessary precautions when using chatbots in the educational context. These threats include the risk of overreliance on technology and dependence, the potential for disseminating inaccurate or biased information, as well as ethical concerns related to issues, such as plagiarism, privacy, and misuse of the technology, reduced human inter- action and motivation, and security risks. Besides, the current concern also is that effective knowledge management (KM) goes hand in hand with the continuous usage (CUS) of ChatGPT for educational purposes while still satisfying users’ information needs. In a study on generative AI chatbot, Al-Sharafi et al. [11] highlighted the close relationship of KM factors with CUS. This study also revealed that the satisfaction factor of users should also be developed in tandem with CUS as they are closely related. Nguyen and Gregar [12] also expressed similar interest in KM to innovation in higher education using chatbot. However, current research on the utilization of ChatGPT in educational settings, specifically regarding the impact of KM factors on user satisfaction and CUS, is limited. This knowledge gap has motivated this study to investigate the effects of KM factors on user satisfaction with the CUS of ChatGPT for educational purposes.

The primary objective of this research is to address and bridge the existing gaps in the literature through the application of the expectation confirmation model (ECM) and the consideration of KM factors. This study aims to substantiate the impact of expectation confirmation (EC) and perceived usefulness (PU) on student satisfaction, as well as their CUS of ChatGPT for educational purposes. Moreover, the research seeks to identify the impact of crucial KM elements, such as knowledge acquisi- tion (KA), knowledge sharing (KS), and knowledge application (KAP), on both user satisfaction and the CUS of ChatGPT.

II. LITERATURE REVIEW

A. Expectation Confirmation Model

The ECM is a well-established and widely used framework in Information Systems research, aiming to comprehend the factors impacting users’ continued use of technology [13]. Previous studies in AI chatbots have confirmed the positive association between EC and PU [3]. In addition, the positive effect of PU on user satisfaction has been substantiated [14]. Moreover, satisfaction has been identified as the primary driver of users’ intention to continue using technology [15]. Accordingly, the

model proposes that user satisfaction is the result of the con- firmation of their expectations and the PU of the technology. In other words, users are more likely to continue using the technology if it meets their expectations and provides them with valuable benefits.

1) Expectation Confirmation: EC refers to an individual’s perception that the actual outcome aligns with their initial ex- pectations [16]. When users’ preuse expectations are fulfilled during their actual usage, it signifies that their expectations have been confirmed [17]. PU, on the other hand, serves as a measure of how users perceive the value and utility of a product or service, playing a crucial role in determining whether users will adopt and continue to use it. Many studies have shown a positive relationship between EC and PU in learning. For example, research by Youjae [18] found a significant positive impact of EC on the PU of an online learning system. In the context of AI chatbots, a study conducted by Nguyen et al. [3] concluded that EC has a positive effect on users’ PU. Therefore, based on these findings, the following hypothesis was proposed.

H1: EC significantly and positively influences the PU of ChatGPT for learning purposes.

According to the ECM theory, customer satisfaction (STS) is determined by the disparity between a customer’s expecta- tions and their perceptions of the received product or service [16]. Thus, prior expectations and the post-adoption perception and confirmation of performance can explain satisfaction [12]. Several studies, such as those by Youjae [18] and Li et al. [19], have discovered a significant positive effect of EC on satisfaction across various products and services. In addition, Davis [20] established a significant relationship between AI chatbot usage, EC, and satisfaction, indicating that users whose expectations were met tend to be more satisfied compared with those whose expectations were not fulfilled. Notably, in the research conducted by Al-Sharafi et al. [11], it was discovered that EC has a positive impact on satisfaction during the utiliza- tion of AI chatbots for educational purposes. This implies that students experience satisfaction when their expectations are met while using ChatGPT for educational objectives. Therefore, the following hypothesis was proposed.

H2: EC significantly and positively influences the STS of using ChatGPT for learning purposes.

2) Perceived Usefulness: PU refers to an individual’s belief about a specific technology’s ability to enhance their perfor- mance or job productivity [21]. This factor exerts a positive influence on customer satisfaction [22], indicating that higher PU leads to greater satisfaction [23]. Users are more likely to use a system if they perceive it as user-friendly and effective in solving their problems, resulting in satisfaction and CUS [24]. In the context of education, students have expressed that chatbots have been beneficial in clarifying conceptual content in their subjects [25]. This suggests that when users perceive the usefulness of utilizing a chatbot, they are more likely to experience satisfaction. ChatGPT, known for generating text based on user input [7], can be a valuable resource in higher education, assisting in improved writing through text generation,

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

NGO et al.: CHATGPT FOR EDUCATIONAL PURPOSES 1343

fact summarization, and creating outlines, thereby saving time and enhancing work quality. Consequently, if ChatGPT can meet students’ needs, they are more likely to be satisfied with its usage. Therefore, based on these observations, the following hypothesis was proposed.

H3: PU significantly and positively influences the STS of using ChatGPT for learning purposes.

According to Silva et al. [26], the intention of individuals to use information systems is positively influenced by the PU, which is a crucial factor for the CUS of technology. PU has also been found to impact attitudes and intentions toward using chatbots, as noted by Pillai and Sivathanu [27]. Moreover, pre- vious studies by Rese et al. [28] and Joyner [29] emphasized the significant influence of PU on chatbot service reuse. In the context of ChatGPT, PU is expected to play a role in the CUS of this technology in education. If students perceive ChatGPT as valuable for educational purposes such as generating creative content, translating languages, or answering questions, they are more likely to continue using it over time. Several studies by Sok [30] and Lee [31] have provided evidence of ChatGPT’s significant impact on education. Furthermore, ChatGPT has been examined across various disciplines, particularly in recent research in the fields of medical education by Sallam [32], Sallam et al. [33], and Cooper [34], as well as science education by Folkes [35]. As anticipated, when students experience the benefits of ChatGPT that are relevant to their learning objectives, they are more likely to continue using it. Hence, the following hypothesis was proposed.

H4: PU significantly and positively influences the CUS of ChatGPT for learning purposes.

3) Satisfaction (STS): Satisfaction (STS) is the evaluation that a product or service has either met or exceeded expectations, as described by Eren [16]. It is also associated with a positive emotional state that arises from a favorable comparison between one’s expectations and the actual performance of the product, as stated by Hsu and Lin [36]. Recent research conducted by Rieke and Martins [37] has demonstrated that user satisfaction significantly influences the intention to use chatbots in the fu- ture. Similarly, Hassanian et al. [38] examined the relationships between chatbot characteristics, user intention, and satisfac- tion, revealing a strong interconnection among these factors. ChatGPT presents exciting opportunities for both students and educators, including the provision of personalized feedback, interactive conversations, enhanced accessibility, streamlined lesson preparation, efficient assessment processes, and the adop- tion of innovative teaching methodologies [7]. Therefore, it is anticipated that if ChatGPT effectively fulfills students’ needs for learning purposes, they will be motivated to continue using it in the future. Hence, the authors hypothesize that students’ satisfaction with ChatGPT will increase their intention to use this tool in educational activities in the future.

H5: STS significantly and positively influences the CUS of ChatGPT for learning purposes.

4) Continuous Usage: The concept of CUS arises from mon- itoring and evaluating how users interact with software programs over time. It represents a user’s persistent engagement with a specific application, measured by their ongoing use and overall positive perception [39]. Wollny et al. [40] found that a user- friendly interface, a quick response time, and a high language competence can enhance user satisfaction, thereby leading to CUS. Similarly, Gatzioufa and Saprikis [41] noted that PU and enjoyment significantly affect users’ intention to continue using chatbots. In this study, CUS of ChatGPT for educational purposes was assessed based on ECM theory and KM factors to examine how these factors, PU, and satisfaction affect the CUS of ChatGPT.

B. Knowledge Management

KM refers to the process of KA, which can be understood as the creation, collection, storage, distribution, and application of knowledge [42]. However, according to Gao et al. [43], the main idea of KM, regardless of the different definitions and descriptions, is to assist individuals in enhancing their learning efficiency and integrating various information sources to boost their competitive edge. KM also equips individuals with the necessary tools and methods to overcome the excessive infor- mation they face and to improve their learning effectiveness and competitive advantage [43]. According to Mbaya [44], the prominence of KM as a management instrument and a novel area of research has surged in recent years. In the context of education, Ramaditya et al. [45] and Kumar [46] studied the impact of KM on higher education. According to Kumar [46], KM is a vital component of higher education, as it enables institutions to efficiently handle and disseminate knowledge to improve teaching, learning, and research results. Al-Sharafi et al. [11] studied the impact of KM factors including KS, KA, and KAP to determine the continued long-term use of chatbots. Based on that, this study applies KM factors, such as KA, KS, and KAP, to study their impact on university students when they use ChatGPT for learning purposes.

1) Knowledge Acquisition: KA involves the process of gain- ing new knowledge and understanding, which can be achieved through various methods, such as reading, attending lectures, or engaging in discussions [47], [48]. The acquisition of knowledge is essential for both KM and the learning process [49], and it can have a positive impact on users’ satisfaction when using Chat- GPT for educational purposes. Accordingly, effective learning and deepening understanding of a topic through interactions with ChatGPT can lead to a sense of satisfaction among users.

ChatGPT is a valuable tool for acquiring knowledge, as it can access and process real-world information through platforms, such as Google Search. This ensures that its responses remain consistent with search results [50], [51], providing students with accurate and up-to-date information. Moreover, as a generative AI chatbot, ChatGPT can respond to questions using various formats, including text or code [35], [37], making it a versatile learning tool across a wide range of subjects. Moreover, KA helps users feel more engaged and motivated in the learning process, which further enhances their satisfaction with ChatGPT.

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

1344 IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024

Generative AI chatbots, such as ChatGPT, offer learners in- stant feedback, personalized learning experiences, and access to vast knowledge resources [52], [53]. These features enhance the learning process and facilitate KA. As learners acquire knowl- edge through AI chatbots, their engagement with the technology increases, promoting its CUS [54], [55]. Similarly, Martin and Bolliger [56] found that as users acquire new knowledge and insights, they are more likely to remain motivated to continue using the technology as a learning tool, leading to a greater level of satisfaction with the overall learning experience. Furthermore, the CUS of technology in education is influenced by various factors, including “ease of use,” “PU,” and “attitude” toward the technology [57], [58]. In conclusion, KA plays a role in shaping these factors and influencing the CUS of generative AI chatbots for learning purposes. Accordingly, the following hypotheses were proposed.

H6: KA significantly and positively influences the STS of using ChatGPT for learning purposes.

H7: KA significantly and positively influences the CUS of ChatGPT for learning purposes.

2) Knowledge Sharing: KS involves the exchange of infor- mation, ideas, or expertise among individuals or groups [59]. ChatGPT can facilitate users’ comprehension and learning about various topics by sharing insights [60]. When users receive helpful and informative responses from ChatGPT, it enhances their learning experience and contributes to their satisfaction. User satisfaction is influenced by factors, such as information quality, response speed, and overall user experience [61]. In addition, positive user experiences shared with others can create a virtuous cycle of learning, satisfaction, and advocacy [62]. Therefore, KS plays a crucial role in promoting user satisfaction with ChatGPT for learning purposes. Based on the literature, the following hypothesis was proposed.

H8: KS significantly and positively influences the STS of using ChatGPT for learning purposes.

Furthermore, KS contributes to the CUS of ChatGPT for learning. It fosters a community of learners who can benefit from the capabilities of generative AI chatbots, such as ChatGPT, over the long term by sharing their knowledge and insights [63]. KS ensures that the information provided by ChatGPT remains accurate, relevant, and up to date [64]. Through interactions and KS, users can improve the quality of feedback and ensure the reliability of the information provided [65]. In addition, KS pro- motes ethical and responsible use by sharing best practices and guidelines, addressing privacy and security concerns associated with generative AI chatbots, such as ChatGPT [66]. Finally, KS fosters inclusiveness and diversity by making the information provided by ChatGPT relevant and accessible to users from diverse backgrounds with different learning needs [67], [68]. Based on the literature, the following hypothesis was proposed.

H9: KS significantly and positively influences the CUS of ChatGPT for learning purposes.

3) Knowledge Application: KAP refers to the practical uti- lization of acquired knowledge and skills in real-world scenarios

Fig. 1. Research framework.

to solve problems, make decisions, and achieve goals [69]. This aspect of learning significantly influences user satisfaction with generative AI chatbots. When users apply the knowledge gained from interacting with generative AI chatbots, such as ChatGPT, to practical situations, they can recognize its relevance and value, leading to a sense of accomplishment and satisfaction [70]. This process enhances users’ confidence and expertise in specific domains, further elevating their satisfaction with ChatGPT. In addition, KAP helps identify knowledge gaps, motivating users to seek more information and engage further with ChatGPT, resulting in a more fulfilling learning experience [71], [72]. Moreover, KAP fosters collaboration and peer-to-peer learning as users share their experiences and insights with others [73], [74], creating a sense of community and a supportive learning environment that enhances satisfaction with ChatGPT. There- fore, the following hypothesis was proposed.

H10: KAP significantly and positively influences the STS of using ChatGPT for learning purposes.

The application of knowledge acquired through interactions with ChatGPT is vital for promoting the CUS of this technology for learning purposes. It enables users to recognize the value and relevance of the information provided, motivating them to continue utilizing ChatGPT as a learning tool [75]. Furthermore, KAP facilitates the transfer of learning across different contexts, enabling users to develop a broader knowledge base that can be applied to various challenges [76], [77]. It also assists in identify- ing areas for improvement, encouraging developers to enhance ChatGPT’s capabilities and ensure its long-term relevance [10]. Finally, KAP fosters collaboration and peer-to-peer learning, establishing a nurturing learning environment that promotes the CUS of ChatGPT [55]. Therefore, the following hypothesis was proposed.

H11: KAP significantly and positively influences the CUS of Chat- GPT for learning purposes.

The proposed framework for this study is demonstrated in Fig. 1.

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

NGO et al.: CHATGPT FOR EDUCATIONAL PURPOSES 1345

Fig. 2. PLS-SEM results.

III. METHODOLOGY

A. Participants

The study involved students from various universities in Viet- nam who employed ChatGPT for educational purposes. In total, 513 valid responses were collected to assess the research frame- work and test the hypotheses. Among the participants, 60.04% (n = 308) were male, while 39.96% (n = 205) were female. The participants represented various academic disciplines, with Information Technology being the most prevalent (35.09%, n = 180), followed by Linguistics (28.07%, n = 144), and Business and Management (5.85%, n = 30).

All the participants were undergraduate university students who had utilized ChatGPT for educational purposes for at least two months. Regarding the purposes of using ChatGPT in learning, students utilized it for tasks, such as searching and synthesizing information, translation, idea generation, writing, coding, and more. Among the participants, 25.93% (n = 133) occasionally used ChatGPT for educational purposes, while 60.23% (n = 309) frequently used it, and 13.74% (n = 71) reported very frequent usage.

B. Questionnaire Design

The research employed a quantitative approach, utilizing a structured questionnaire as the main data collection tool. The questionnaire design was based on the theoretical framework proposed by Najeeb et al. [78] and extended from the model introduced by Al-Sharafi et al. [11]. It consisted of three main sections. The first section aimed to gather demographic informa- tion, including participants’ names, genders, and fields of study, which helped understanding their background. The second sec- tion covered general information about KM and ChatGPT. This part included a definition of KM, an overview of ChatGPT, and questions about its usage, such as the purposes, duration, and frequency of use. The third section consisted of 28 statements representing three constructs of KM factors, including KA, KS, and KAP, and four constructs of ECM, including EC, PU,

TABLE I CONSTRUCTS’ ITEMS

satisfaction (STS), and CUS (see Table I). Participants were requested to rate their responses on a five-point Likert scale, ranging from 1—“Strongly Disagree” to 5—“Strongly Agree.” By structuring the questionnaire in this manner, the study aimed to comprehensively explore the participants’ understanding and usage of ChatGPT for their educational purposes and their per- spectives on various constructs related to KM and the ECM. This

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

1346 IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024

approach allowed for a thorough and systematic investigation of the research objectives.

C. Data Collection

In this study, a nonprobability convenience sampling method was used to collect data due to its advantages in terms of convenience, time efficiency, and cost effectiveness. Before con- ducting the main data collection, a pilot phase was implemented to test the research instruments and 50 responses were gathered to validate the questionnaire. The questionnaire was distributed via email using Google Forms to conduct an online survey. To ensure the reliability of the questionnaire, Cronbach’s alpha, a statistical measure commonly used in research to determine internal consistency and agreement among items within a scale or questionnaire, was employed. The Cronbach’s alpha coef- ficients obtained ranged from 0.784 to 0.904 with 50 replies, indicating a high level of reliability. After testing Cronbach’s alpha, a reliable and appropriate set of questions was determined for the research purpose and context, which was then used for the empirical study.

According to Hair et al. [79], a quantitative research study should have a sample size of at least five times the total number of observed factors. In this research, there were 28 observed variables, leading to a minimum required sample size of 28 × 5 = 140. The survey successfully gathered over 600 responses from May 15, 2023 to July 15, 2023. Out of these, 513 responses were deemed valid and utilized for subsequent analysis.

D. Data Analysis

The study employed two software tools to analyze the data. IBM SPSS version 26 was used to perform descriptive statistics and check the data quality. SmartPLS 3.0 software was used to conduct partial least squares structural equation modeling (PLS- SEM) and test the hypotheses. This method was appropriate for examining complex relationships, enabling a comprehensive investigation of the impact of KM on user STS. The study applied confirmatory composite analysis (CCA) to validate measure- ment models in the context of PLS-SEM. It involves confirming the adequacy and reliability of the chosen indicators for latent constructs [80]. Since PLS-SEM relies on total variance rather than covariance, it lacks an equivalent measure for goodness of fit. In the context of CCA in PLS-SEM, the assessment includes evaluating the reliability, convergent validity, and discriminant validity of the measurement models. Subsequently, the focus shifts to appraising the predictive performance of the structural model, which is measured by f2, R2, and Q2.

IV. RESULTS

A. Measurement Model Assessment

To assess the reliability and accuracy of the measurement instruments employed in this study, Table II presents essen- tial findings concerning internal consistency and convergent validity. The results showed that Cronbach’s alpha values for all the indicators were within an acceptable to good range, ranging from 0.748 to 0.850, indicating a satisfactory level of internal consistency for most of the investigated constructs [81].

TABLE II CONVERGENT VALIDITY

In addition, all the indicators displayed high loadings, ranging from 0.726 to 0.836, signifying a strong relationship with the underlying latent variable. Nonetheless, it was observed that KAP4 exhibited a slightly weaker loading (0.670) compared to the remaining KAP indicators, suggesting a relatively less pronounced association with the latent variable. Moreover, all the constructs had average variance extracted (AVE) values above the recommended threshold of 0.5, demonstrating strong convergent validity. These results implied that the indicators used to measure each construct were adequately related to the intended representations [82].

This study employed a questionnaire-based data collection method, which may introduce common method bias. To mitigate this potential bias, the study utilized the variance inflation factor (VIF) approach. In addition, the VIF was employed as a metric to evaluate the impact of multicollinearity on regression coeffi- cients. Typically, a VIF value exceeding 10 is indicative of high multicollinearity, and values above 5 may raise concerns [83]. However, the VIF values in the presented table range from 1.263 to 2.056, well below the critical threshold of 5. This suggests that all the variables, including EC, PU, satisfaction (STS), CUS, KA, KS, and KAP, demonstrated minimal correlation issues. The acceptable VIF values imply that the regression coefficients remain unbiased despite the presence of multicollinearity, thus confirming the good discriminant validity of the items. As a result, these indicators can be confidently employed for robustly measuring the respective constructs within the study.

To assess the normality of the data, the study computed the skewness and kurtosis values for each variable. Skewness indicates how symmetrically the data are distributed around the mean, and kurtosis indicates how peaked or flat the data are, compared to a normal distribution [84]. According to Hair et al. [84], these values should be between−2 and 2 for normality. The

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

NGO et al.: CHATGPT FOR EDUCATIONAL PURPOSES 1347

TABLE III HTMT RATIOS (DISCRIMINANT VALIDITY)

TABLE IV FORNELL–LARCKER CRITERION

skewness values ranged from −1.169 to −0.290, showing that the data had a moderate negative skew, with most participants having high scores on the items. The kurtosis values ranged from −0.526 to 1.380, showing that the data were either platykurtic (flatter than normal) or leptokurtic (more peaked than normal). However, both the skewness and kurtosis values were within the −2 to +2 range, suggesting that the data were approximately normal.

The discriminant validity assessment results are presented in Tables III and IV, where the heterotrait–monotrait (HTMT) ratios and the Fornell–Larcker criterion were employed, re- spectively. Table III shows the discriminant validity among the constructs, with most HTMT ratios falling below 0.9, indicating their distinctiveness. Although a few indicators, such as EC- PU (0.916), EC-STS (0.908), PU-STS (0.918), and PU-CUS (0.909), showed marginal values, the results of the Fornell– Larcker criterion were within acceptable ranges, ensuring that there were no significant problems concerning discriminant validity. Table IV displays the outcomes of the Fornell–Larcker criterion, which gauged discriminant validity through the com- parison of the square root of the AVE and the correlations among different constructs. The aim was to verify that the square root of AVE for each construct exceeded its correlations with other constructs. The findings indicated the fulfillment of this criterion for all the constructs, as the square root of AVE for each construct surpassed its correlations with other constructs. Therefore, this confirmed that the measures used in the study were discriminately valid.

B. Structural Model Assessment

The results presented in Table V provide strong support for the majority of hypotheses tested in this study, except for H9 and H11, which did not receive support from the data analysis. Specifically, H1 received strong empirical support, showing a significant positive relationship between EC and PU (H1: β = 0.719, t = 21.774, and p < 0.001). This suggested that when individuals’ expectations regarding the efficacy of ChatGPT were met, they perceived this tool as highly beneficial for their

TABLE V STRUCTURAL MODELING RESULTS

educational purposes. Similarly, H2 was supported, indicating that EC positively predicted student STS (H2: β = 0.515, t = 12.152, and p < 0.001). The findings underscored the crucial role of aligning users’ expectations with ChatGPT’s actual performance in fostering greater student STS with its utility. In addition, the analysis of the data revealed a strong positive correlation between PU and STS (H3: β = 0.192, t = 4.878, and p < 0.001). This result indicated that when students perceive ChatGPT as a valuable and practical tool for their educational pursuits, it positively impacts their overall STS with the sys- tem. Besides, the examination of CUS indicated a significant influence of PU on the CUS (H4: β = 0.255, t = 5.213, and p < 0.001). In addition, STS was found to have a moderate and positive impact on CUS (H5: β = 0.315, t = 6.082, and p < 0.001). Moreover, hypotheses H6 to H11 explored the positive relationships between various KM factors and both STS and CUS of ChatGPT. Specifically, STS was predicted by KA (H6: β = 0.087, t = 2.072, and p < 0.05), KS (H8: β = 0.077, t = 2.398, and p < 0.05), and KAP (H10: β = 0.088, t = 2.027, and p < 0.05). These results suggested that when students actively engaged in acquiring, sharing, and applying knowledge using ChatGPT, they experienced higher levels of STS with this tool. Furthermore, CUS was found to be predicted by knowledge KA (H7:β= 0.184, t= 3.863, and p< 0.001). However, the analysis revealed that KS (H9: β = 0.057, t = 1.429, and p = 0.153) and KAP (H11: β = 0.076, t = 1.54, and p = 0.124) demonstrated no significant association with the CUS of ChatGPT.

In the regression analysis, the coefficients (β) provided in- sights into how each predictor uniquely influences the dependent variables. EC exhibits a robust and statistically significant impact on both PU and STS, with coefficients of 0.719 and 0.515, respectively. PU affects both STS and CUS, with coefficients of 0.192 and 0.255, respectively, demonstrating a slightly strong effect on CUS. STS predicts CUS independently, with a value of 0.315. KA exerts a moderate impact on both STS and CUS, with coefficients of 0.087 and 0.184, respectively, and a relatively higher influence on CUS. Meanwhile, KS and KAP contribute minor effects to STS and CUS, with coefficients of 0.057 and 0.088, respectively.

Furthermore, to assess the meaningfulness of the effects of predictor variables on the outcome variable, f2 values proposed by Cohen [85] were utilized. The results in Table V indicated that EC had a significant and robust effect on PU (f2 = 1.068) and a moderate effect on STS (f2 = 0.348). In contrast, the effect of PU on STS was found to be relatively small (f2 = 0.052),

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

1348 IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024

and its impact on CUS was of small magnitude (f2 = 0.066). In addition, the effect size of STS on CUS was also small (f2

= 0.094). Regarding KA, KS, and KAP, all exhibited relatively small effects on both STS and CUS, with f2 values ranging from 0.005 to 0.034.

The R2 value is a metric indicating the extent to which the dependent variable changes in response to variations in the in- dependent variable. Ranging from 0 to 1, higher values indicate a more accurate fit and increased explanatory power. The R2 values obtained for PU, STS, and CUS were 0.517, 0.709, and 0.586, respectively. These values signified that a significant proportion of the variance in PU, STS, and CUS can be attributed to the predictor variables integrated into the analysis, amounting to ap- proximately 51.7%, 70.9%, and 58.6%, respectively. Moreover, Q2 is a critical metric used to assess the predictive relevance and external validity of a PLS-SEM model (Fig. 2). It provides valu- able insights into the model’s capacity to predict endogenous constructs beyond the sample used for model estimation [85]. The obtained Q2 values for PU, STS, and CUS were reported as 0.514, 0.688, and 0.545, respectively. These Q2 values indicated that the respective models had moderate to substantial predictive relevance, as they accounted for a considerable proportion of the variance in the endogenous constructs, beyond what could be explained by the model’s exogenous constructs [84], [85].

V. DISCUSSION

The main goal of this study is to examine the influence of different KM factors as well as the roles of EC and PU on student satisfaction (STS) and CUS of ChatGPT for educational purposes. The results of PLS-SEM analysis provided strong empirical support for most of the hypotheses proposed in the research model.

To investigate how students’ EC of ChatGPT performance influenced their PU and satisfaction (STS) with ChatGPT, this study utilized the ECM. The results confirmed the significant positive effects of EC on both PU and satisfaction (STS) aligning with prior studies that applied the ECM to assess user satisfaction and CUS across various information systems [86], [87]. This suggests that when ChatGPT meets or surpasses students’ expec- tations before usage in their learning, it increases satisfaction and fosters CUS. Moreover, the findings indicated positive effects of PU on both satisfaction (STS) and CUS, signifying that students who perceived ChatGPT as useful for their tasks were more satisfied with the tool and expressed a greater intention to continue using it in the future. In a recent study, Ngo [88] found that students acknowledged the usefulness of ChatGPT in their learning, leading to satisfaction with the tool and its adoption intention. The integration of ChatGPT into their learning created positive emotions and attitudes among users, enhancing their satisfaction and loyalty to the tool. These results are consistent with previous studies by Pozón-López et al. [89] and Wang et al. [90], which demonstrated a positive relationship between PU, satisfaction, and CUS in diverse contexts. In prior research, Sohail et al. [91] combined the ECM with uncertainty reduction theory, revealing a positive impact of performance confirmation on satisfaction in an e-commerce context. Similarly, Dhiman and

Jamwal [92] integrated the ECM with task-technology fit theory and demonstrated that user EC, PU, and satisfaction significantly predicted intentions to continue using a chatbot. These studies underscore the versatility of the ECM in comprehending user satisfaction and CUS with AI chatbots across various contexts. This study contributes to the literature by integrating the ECM with other theories to examine user satisfaction with generative AI chatbots in general and ChatGPT in particular in diverse domains.

Regarding the impact of KM factors on satisfaction (STS) and CUS, the findings revealed that KA had a significant positive effect on both satisfaction (STS) and CUS. This implies that as students actively acquired new knowledge or skills using ChatGPT, their trust and satisfaction with the system increased, leading to a greater intention to continue using it. ChatGPT emerged as an effective and engaging tool for facilitating KA among students, enhancing their satisfaction, and promoting CUS. The personalized and interactive feedback, suggestions, and examples provided by ChatGPT contributed to a positive learning experience, stimulating curiosity, interest, and moti- vation [7], [10]. These factors can increase students’ PU, en- joyment, and value of ChatGPT, thereby fostering trust and a commitment to its future use. This aligns with previous studies that identified a positive relationship between KA, satisfaction, and CUS in various e-learning contexts [34], [55]. However, the results also showed that KS and KAP had a significant effect on satisfaction (STS) but not on CUS. This suggests that while students found satisfaction in sharing or applying their knowledge using ChatGPT, it did not impact their intention to continue using it in long term. In addition, there may be a lack of awareness or understanding among students regarding how to effectively utilize ChatGPT for KS or KAP. Possible explanations for this finding include the perception of Chat- GPT as a tool for individual learning rather than collaborative learning, or the absence of features or incentives, such as social interaction, recognition, rewards, or feedback, to encourage KS and KAP. This result is also in line with some recent studies. For example, Borsci et al. [93] identified KA as a key determinant of satisfaction and reuse intention of an AI-enabled CRM chatbot for e-commerce education. They also suggested that chatbot developers and educators should provide rich and diverse content and feedback to enhance KA for users. Likewise, Matar and Raudeliūnienė [94] reported that KA significantly predicted user satisfaction and loyalty of a conversational agent for language learning, recommending the provision of various and relevant contents and quizzes to support KA for learners. However, these studies did not explore the effects of KS and KAP on satisfaction and CUS of chatbots or conversational agents. Therefore, this study contributes to the literature by investigating these effects specifically within the domain of a generative AI chatbot, i.e., ChatGPT, and provides insights into the reasons behind the observed insignificant results.

Furthermore, the findings revealed that satisfaction (STS) had a moderate positive impact on CUS, indicating that when stu- dents derived satisfaction from their interactions with ChatGPT, they were more likely to continue using it in the future. This implies that ChatGPT effectively met or exceeded students’

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

NGO et al.: CHATGPT FOR EDUCATIONAL PURPOSES 1349

expectations, providing them with tangible value and benefits in terms of learning. It also signifies that ChatGPT succeeded in cultivating a loyal customer base among students, potentially leading to recommendations to others or expanded use for diverse purposes. The result further suggests the potential for ChatGPT to leverage its loyal user base for cross-selling or up-selling other products or services that complement its func- tionality. This finding aligns with prior research that observed a positive correlation between satisfaction and CUS in various information systems [95], [96]. It underscores the importance for ChatGPT developers and educators to continually monitor and assess user satisfaction, implementing necessary improvements such as updates, enhancements, or customization options for ChatGPT, and providing guidance for students using the tool for educational purposes.

These findings contribute to the field of education by provid- ing valuable insights into the optimal utilization of ChatGPT in educational settings. Based on the results of this study, it is imperative for both ChatGPT developers and educators to estab- lish clear and attainable expectations for students regarding the chatbot’s capabilities and limitations. Providing timely feedback and assistance can enhance user satisfaction and loyalty. Empha- sizing the chatbot’s value and benefits for students’ educational objectives, supported by evidence and testimonials illustrating its effectiveness and functionality, will further enhance user sat- isfaction and sustained usage. In addition, incorporating features and activities that facilitate KA, such as offering diverse relevant content, feedback, quizzes, or games, can contribute to increased user satisfaction with the system and promote sustained usage. These recommendations aim to improve the user experience and ensure the successful integration of ChatGPT in academic environments. Furthermore, cybersecurity risks must be metic- ulously assessed and actively scrutinized during the endeavor to integrate AI chatbots sustainably into society, as emphasized in the study by Arpaci [97]. By mitigating these risks, the goal is to ensure the enduring and secure utilization of AI chatbots, thereby fostering social sustainability.

VI. LIMITATIONS AND RECOMMENDATIONS

This study acknowledges limitations that should be consid- ered. First, the use of a convenience sample from universities in Vietnam may restrict the generalizability of the findings to other populations or contexts. To enhance external validity, future research should replicate the study with larger and more diverse samples from various countries or regions. Second, relying on self-reported data from a questionnaire may introduce biases like social desirability or acquiescence. To address this, future research should employ multiple methods or data sources, such as interviews, observations, or system logs, to validate and triangulate the results. Third, the study focused on ChatGPT as a specific generative AI chatbot for learning, which may not fully represent other types or features of chatbots available in the market. For a comprehensive understanding, future research should compare different chatbots or conversational agents in terms of their impact on user satisfaction and CUS in the context of education.

Based on the findings of this study, ChatGPT developers and educators should establish realistic and transparent expectations for students regarding the ChatGPT’s capabilities and limita- tions. Providing timely feedback and support can also enhance user satisfaction and retention. Emphasizing the usefulness and benefits of ChatGPT for students’ learning goals and outcomes, along with presenting evidence and feedback to demonstrate its effectiveness and functionality, will further enhance user satis- faction and CUS. Moreover, incorporating features and activities that facilitate KM factors, such as offering diverse and relevant content, feedback, quizzes, or games, can contribute to greater user satisfaction with the system and promote CUS. These recommendations might help optimize the user experience and ensure the successful integration of ChatGPT in educational settings. When formulating future education policies to address the dynamic integration of generative AI tools such as ChatGPT into educational settings, several key recommendations need consideration. First, policymakers should prioritize the estab- lishment of clear guidelines and standards for the ethical use of AI technologies in educational settings, ensuring transparency and accountability. Second, fostering partnerships between ed- ucational institutions and AI developers could facilitate the cus- tomization of AI tools to better suit educational objectives and student needs. Moreover, investing in research and development initiatives aimed at advancing AI literacy among educators and students alike is crucial for promoting responsible and effec- tive integration of generative AI technologies into curricula. By adopting a proactive approach to policy formulation that encompasses these recommendations, educational stakeholders can navigate the complexities of integrating AI into education while maximizing its potential benefits.

VII. CONCLUSION

This study aims to evaluate the potential of ChatGPT for CUS in educational settings. Specifically, the research validates the ECM theory by examining the roles of EC and PU on satisfaction (STS) and CUS. In addition, the research explores the influence of various KM factors, including KA, KS, and KAP on student satisfaction (STS) and CUS of ChatGPT for learning. The study employed a quantitative approach, using a questionnaire to collect data from 513 students who have utilized ChatGPT for educational purposes. The data were analyzed using PLS-SEM to test the hypotheses and examine the relationships between constructs.

The results strongly support most hypotheses, except for H9 and H11. Specifically, the findings validate the significant positive impact of EC on both PU and satisfaction (STS), as well as the positive effects of PU on both satisfaction (STS) and CUS. Moreover, KA exhibits a significant positive effect on both satisfaction (STS) and CUS, whereas KS and KAP influence only satisfaction (STS). Furthermore, the study reveals that satisfaction (STS) has a moderately positive impact on CUS. These findings demonstrate how the students’ expectations and perceptions shape their satisfaction and loyalty. The result suggests that students with higher levels of EC, PU, and KA are more likely to be satisfied with ChatGPT use in their learning and

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

1350 IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024

continue to use it in the future. However, applying the knowledge they acquire and sharing their own situations and problems are important for making students satisfied with the use of ChatGPT in their learning, though not sufficient for fostering commitment to loyalty or intention for CUS.

The study holds both theoretical and practical implications. Theoretically, by using the ECM as a theoretical framework and considering KM factors as antecedents, this study contributes to the recent understanding of satisfaction and CUS of Chat- GPT in educational contexts. Furthermore, the study provides empirical evidence supporting the validity and reliability of the constructs and measures used in the research model. From a practical perspective, this study provides valuable insights and recommendations for ChatGPT developers and educators to enhance both satisfaction and CUS of ChatGPT. By addressing elements such as EC, PU, and KM factors, along with providing high-quality information and services, developers and educators can better tailor their approach to meet students’ needs and preferences. Encouraging users to share and apply knowledge gained from ChatGPT in their specific contexts can further increase satisfaction and loyalty, promoting the development and enhancement of the tool. The study also identifies some limitations and challenges of ChatGPT, such as the lack of impact of KS and KAP on CUS and provides directions for future research and development.

ACKNOWLEDGMENT

The authors would like to express sincere thanks to all the respondents who participated in the survey. Their invaluable contributions and willingness to share their insights were in- strumental in the successful completion of this research project.

REFERENCES

[1] S. Roos, Chatbots in Education: A Passing Trend or a Valuable Pedagog- ical Tool?” Uppsala Univ., Uppsala, Sweden, Jun. 2018. [Online]. Avail- able: http://www.diva-portal.org/smash/record.jsf?pid=diva2:1223692

[2] F. Clarizia, F. Colace, M. Lombardi, F. Pascale, and D. Santaniello, “Chatbot: An education support system for student,” in Cyberspace Safety and Security, A. Castiglione, F. Pop, M. Ficco, and F. Palmieri, Eds., Cham, Switzerland: Springer, 2018, pp. 291–302.

[3] D. M. Nguyen, Y.-T. H. Chiu, and H. D. Le, “Determinants of continuance intention towards banks’ Chatbot services in Vietnam: A necessity for sustainable development,” Sustainability, vol. 13, no. 14, May 2021, Art. no. 7625, doi: 10.3390/su13147625.

[4] J. Wang, G.-H. Hwang, and C.-Y. Chang, “Directions of the 100 most cited chatbot-related human behavior research: A review of academic publica- tions,” Comput. Educ.: Artif. Intell., vol. 2, Jan. 2021, Art. no. 100023, doi: 10.1016/j.caeai.2021.100023.

[5] S. Ullmann and M. Schoop, “Potentials of chatbot technologies for higher education: A systematic review,” in Proc. U.K. Acad. Inf. Syst. Conf. Proc., Jun. 2022, Accessed: Oct. 26, 2022. [Online]. Available: https://aisel.aisnet.org/ukais2022/11/

[6] M. Halaweh, “ChatGPT in education: Strategies for responsible imple- mentation,” Contemporary Educ. Technol., vol. 15, no. 2, Apr. 2023, Art. no. ep421, doi: 10.30935/cedtech/13036.

[7] M. M. Rahman and Y. Watanobe, “ChatGPT for Education and Research: Opportunities, threats, and strategies,” Appl. Sci., vol. 13, no. 9, May 2023, Art. no. 5783, doi: 10.3390/app13095783.

[8] Y. Liu et al., “Summary of ChatGPT-related research and perspective towards the future of large language models,” Meta-Radiol., vol. 1, no. 2, Sep. 2023, Art. no. 100017, doi: 10.1016/j.metrad.2023.100017.

[9] Y. An, W. Ouyang, and F. Zhu, “ChatGPT in higher educa- tion: Design teaching model involving ChatGPT,” Lecture Notes Educ. Psychol. Public Media, vol. 24, pp. 47–56, Nov. 2023, doi: 10.54254/2753-7048/24/20230560.

[10] A. Bahrini et al., “ChatGPT: Applications, opportunities, and threats,” in Proc. Syst. Inf. Eng. Des. Symp., Charlottesville, VA, USA, 2023, pp. 274– 279, doi: 10.1109/SIEDS58326.2023.10137850.

[11] M. A. Al-Sharafi, M. Al-Emran, M. Iranmanesh, N. Al-Qaysi, N. A. Iahad, and I. Arpaci, “Understanding the impact of knowledge management fac- tors on the sustainable use of AI-based chatbots for educational purposes using a hybrid SEM-ANN approach,” Interact. Learn. Environ., vol. 31, no. 10, pp. 1–20, May 2022, doi: 10.1080/10494820.2022.2075014.

[12] T. Nguyen and A. Gregar, “Impacts of knowledge management on innovation in higher education institutions: An empirical evidence from Vietnam,” Econ. Sociol., vol. 11, pp. 301–320, Sep. 2018, doi: 10.14254/2071-789X.2018/11-3/18.

[13] A. Bhattacherjee, “Understanding information systems continuance: An expectation-confirmation model,” MIS Quart., vol. 25, no. 3, pp. 351–370, Sep. 2001, doi: 10.2307/3250921.

[14] M. Ashfaq, J. Yun, S. Yu, and S. M. C. Loureiro, “I, Chatbot: Model- ing the determinants of users’ satisfaction and continuance intention of AI-powered service agents,” Telematics Informat., vol. 54, Nov. 2020, Art. no. 101473, doi: 10.1016/j.tele.2020.101473.

[15] R. L. Oliver, “A cognitive model of the antecedents and consequences of satisfaction decisions,” J. Marketing Res., vol. 17, no. 4, pp. 460–469, Nov. 1980, doi: 10.1177/002224378001700405.

[16] B. A. Eren, “Determinants of customer satisfaction in chatbot use: Ev- idence from a banking application in Turkey,” Int. J. Bank Marketing, vol. 39, no. 2, pp. 294–311, Jan. 2021, doi: 10.1108/IJBM-02-2020-0056.

[17] V. Zeithaml, L. Berry, and A. Parasuraman, “The behavioral conse- quences of service quality,” J. Marketing, vol. 60, pp. 31–46, Apr. 1996, doi: 10.2307/1251929.

[18] Y. Youjae, “A critical review of consumer satisfaction,” Working Paper, 1989. [Online]. Available: http://deepblue.lib.umich.edu/handle/2027.42/ 36290

[19] L. Li, K. Y. Lee, E. Emokpae, and S.-B. Yang, “What makes you con- tinuously use chatbot services? Evidence from Chinese online travel agencies,” Electron Markets, vol. 31, no. 3, pp. 575–599, Sep. 2021, doi: 10.1007/s12525-020-00454-z.

[20] F. D. Davis, “Perceived usefulness, Perceived ease of use, and user accep- tance of information technology,” MIS Quart., vol. 13, no. 3, pp. 319–340, Sep. 1989, doi: 10.2307/249008.

[21] V. Bhatt and D. Nagar, “An empirical study to evaluate factors affecting customer satisfaction on the adoption of Mobile Banking,” Turkish J. Com- put. Math. Educ., vol. 12, pp. 5332–5353, Apr. 2021, doi: 10.17762/tur- comat.v12i10.5338.

[22] M. Olivia and N. Marchyta, “The influence of perceived ease of use and perceived usefulness on E-wallet continuance intention: Intervening role of customer satisfaction,” J. Teknik Ind., vol. 24, pp. 13–22, May 2022, doi: 10.9744/jti.24.1.13-22.

[23] T. Pei Kian, “Factors that affect user satisfaction of using E-commerce chatbot: A study on generation Z,” Int. J. Bus. Technol. Manage., vol. 5, pp. 292–303, Mar. 2023, doi: 10.55057/ijbtm.2023.5.1.23.

[24] M. Saiz-Manzanares, R. Sánchez, L. Martín Antón, I. González-Díez, and L. Almeida, “Perceived satisfaction of university students with the use of chatbots as a tool for self-regulated learning,” Heliyon, vol. 9, Jan. 2023, Art. no. e12843, doi: 10.1016/j.heliyon.2023.e12843.

[25] M. Goli, A. K. Sahu, S. Bag, and P. Dhamija, “Users’ Acceptance of artifi- cial intelligence-based chatbots: An empirical study,” Int. J. Technol. Hum. Interact., vol. 19, no. 1, pp. 1–18, Jan. 2023, doi: 10.4018/IJTHI.318481.

[26] F. A. Silva, A. S. Shojaei, and B. Barbosa, “Chatbot-based services: A study on customers’ reuse intention,” J. Theor. Appl. Electron. Commerce Res., vol. 18, no. 1, pp. 457–474, Mar. 2023, doi: 10.3390/jtaer18010024.

[27] R. Pillai and B. Sivathanu, “Adoption of AI-based chatbots for hospitality and tourism,” Int. J. Contemporary Hospitality Manage., vol. 32, no. 10, pp. 3199–3226, Sep. 2020, doi: 10.1108/IJCHM-04-2020-0259.

[28] A. Rese, L. Ganster, and D. Baier, “Chatbots in retailers’ cus- tomer communication: How to measure their acceptance?,” J. Re- tailing Consum. Serv., vol. 56, Sep. 2020, Art. no. 102176, doi: 10.1016/j.jretconser.2020.102176.

[29] D. A. Joyner, “ChatGPT in Education: Partner or pariah?,” XRDS, vol. 29, no. 3, pp. 48–51, Apr. 2023, doi: 10.1145/3589651.

[30] S. Sok, Opinion: Benefits and Risks of ChatGPT in Educa- tion, Cambodianess, Phnom Penh, Cambodia, Mar. 2023. [Online]. Available: https://cambodianess.com/article/opinion-benefits-and-risks- of-chatgpt-in-education

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

NGO et al.: CHATGPT FOR EDUCATIONAL PURPOSES 1351

[31] H. Lee, “The rise of ChatGPT: Exploring its potential in medical educa- tion,” Anat. Sci. Educ., vol. 1, pp 1–6, 2023, doi: 10.1002/ase.2270.

[32] M. Sallam, “ChatGPT Utility in Healthcare Education, research, and prac- tice: Systematic Review on the promising perspectives and valid concerns,” Healthcare, vol. 11, no. 6, Mar. 2023, Art. no. 887, doi: 10.3390/health- care11060887.

[33] M. Sallam, N. Salim, M. Barakat, and A. Al-Tammemi, “ChatGPT ap- plications in medical, dental, pharmacy, and public health education: A descriptive study highlighting the advantages and limitations,” Narra J, vol. 3, no. 1, p. e103, May 2023, doi: 10.52225/narra.v3i1.103.

[34] G. Cooper, “Examining science education in ChatGPT: An exploratory study of generative artificial intelligence,” J. Sci. Educ. Technol., vol. 32, no. 3, pp. 444–452, Mar. 2023, doi: 10.1007/s10956-023-10039-y.

[35] V. S. Folkes, “Consumer reactions to product failure: An attributional approach,” J. Consum. Res., vol. 10, no. 4, Mar. 1984, pp. 398–409.

[36] C. Hsu and J. C. Lin, “Understanding the user satisfaction and loyalty of customer service chatbots,” J. Retailing Consum. Serv., vol. 71, Mar. 2023, Art. no. 103211, doi: 10.1016/j.jretconser.2022.103211.

[37] T. Rieke and H. Martins, “The relationship between motives for us- ing a Chatbot and satisfaction with Chatbot characteristics: An ex- ploratory study,” SHS Web Conf., vol. 160, Mar. 2023, Art. no. 01007, doi: 10.1051/shsconf/202316001007.

[38] Z. Hassanian, M. Id, H. I. Ahanchian, and K. Hossein, “The pro- cess of knowledge acquiring in nursing education: Grounded the- ory,” Res. Develop. Med. Educ., vol. 7, pp. 68–76, Dec. 2018, doi: 10.15171/rdme.2018.015.

[39] Z. Li et al., “Students’ online learning adaptability and their continuous us- age intention across different disciplines,” Humanities Soc. Sci. Commun., vol. 10, no. 1, Nov. 2023, Art. no. 838, doi: 10.1057/s41599-023-02376-5.

[40] S. Wollny, J. Schneider, D. Di Mitri, J. Weidlich, M. Rittberger, and H. Drachsler, “Are we there yet?—A systematic literature re- view on chatbots in education,” Front. Artif. Intell., vol. 4, Jul. 2021, Art. no. 654924.

[41] P. Gatzioufa and V. Saprikis, “A literature review on users’ behavioral intention toward chatbots’ adoption,” Appl. Comput. Informat., Jul. 2022, doi: 10.1108/ACI-01-2022-0021.

[42] J. A. W. Cortada James, The Knowledge Management Yearbook 2000- 2001. London, U.K.: Routledge, Jul. 2000.

[43] T. Gao, Y. Chai, and Y. Liu, “A review of knowledge management about theoretical conception and designing approaches,” Int. J. Crowd Sci., vol. 2, no. 1, pp. 42–51, Apr. 2018, doi: 10.1108/IJCS-08-2017-0023.

[44] P. C. Mbaya, “Unavoidable practices for effective business correspondence teaching,” Int. J. Sci. Res. Manage., vol. 11, no. 7, pp. 2816–2827, 2023, doi: 10.18535/ijsrm/v11i07.el02.

[45] M. Ramaditya, S. Syamsari, H. Hadirawati, and A. Hanifah, “The effect of digital learning, innovative behavior and knowledge management on pri- vate higher education performance,” AL-ISHLAH: J. Pendidikan, vol. 15, no. 4, pp. 349–372, 2023, doi: 10.35445/alishlah.v15i4.3773.

[46] A. A. Kumar, “Knowledge management in Indian higher education— Issues and challenges,” Prabandhan: Indian J. Manage., vol. 16, no. 6, pp. 60–67, Jun. 2023, doi: 10.17010/pijom/2023/v16i6/172864.

[47] N. Ngoc Thang and P. Anh Tuan, “Knowledge acquisition, knowledge management strategy and innovation: An empirical study of Vietnamese firms,” Cogent Bus. Manage., vol. 7, no. 1, Jun. 2020, Art. no. 1786314, doi: 10.1080/23311975.2020.1786314.

[48] M. Cukurova, J. Bennett, and I. Abrahams, “Students’ knowledge acquisi- tion and ability to apply knowledge into different science contexts in two different independent learning settings,” Res. Sci. Technol. Educ., vol. 36, no. 1, pp. 17–34, Jun. 2018, doi: 10.1080/02635143.2017.1336709.

[49] S. Elbanna and L. Armstrong, “Exploring the integration of ChatGPT in education: Adapting for the future,” Manage. Sustain.: Arab Rev., vol. 3, no. 1, pp. 16–29, 2024, doi: 10.1108/MSAR-03-2023-0016.

[50] S. Grassini, “Shaping the future of education: Exploring the potential and consequences of AI and ChatGPT in educational settings,” Educ. Sci., vol. 13, no. 7, Jul. 2023, Art. no. 692, doi: 10.3390/educsci13070692.

[51] T. T. Nguyen, A. D. Le, H. T. Hoang, and T. Nguyen, “NEU- chatbot: Chatbot for admission of National Economics University,” Comput. Educ.: Artif. Intell., vol. 2, Oct. 2021, Art. no. 100036, doi: 10.1016/j.caeai.2021.100036.

[52] A. Radford and K. Narasimhan, “Improving language understanding by generative pre-training,” 2018. [Online]. Available: https://www. semanticscholar.org/paper/Improving-Language-Understanding-by-Gen erative-Radford-Narasimhan/cd18800a0fe0b668a1cc19f2ec95b5003d0a 5035

[53] S. Roller et al., “Recipes for building an open-domain chatbot,” in Proc. 16th Conf. Eur. Ch. Assoc. Comput. Linguistics, pp. 300–325, doi: 10.18653/v1/2021.eacl-main.24.

[54] Y. H. Yeo et al., “Assessing the performance of ChatGPT in answering questions regarding cirrhosis and hepatocellular carci- noma,” Clin. Mol. Hepatol., vol. 29, no. 3, pp. 721–732, Jul. 2023, doi: 10.3350/cmh.2023.0089.

[55] C.-C. Lin, A. Y. Q. Huang, and S. J. H. Yang, “A review of AI-driven conversational chatbots implementation methodologies and challenges (1999–2022),” Sustainability, vol. 15, no. 5, Mar. 2023, Art. no. 4012, doi: 10.3390/su15054012.

[56] F. Martin and D. U. Bolliger, “Engagement matters: Student percep- tions on the importance of engagement strategies in the online learning environment,” Online Learn., vol. 22, no. 1, pp. 205–222, Mar. 2018, doi: 10.24059/olj.v22i1.1092.

[57] E. R. Mollick and L. Mollick, New Modes of Learning Enabled by AI Chat- bots: Three Methods and Assignments. New York, NY, USA: Rochester, Dec. 2022, doi: 10.2139/ssrn.4300783.

[58] C. W. Okonkwo and A. Ade-Ibijola, “Chatbots applications in education: A systematic review,” Comput. Educ.: Artif. Intell., vol. 2, Sep. 2021, Art. no. 100033, doi: 10.1016/j.caeai.2021.100033.

[59] Y. B. Rajabalee and M. I. Santally, “Learner satisfaction, engagement and performances in an online module: Implications for institutional e-learning policy,” Educ. Inf. Technol., vol. 26, no. 3, pp. 2623–2656, Nov. 2021, doi: 10.1007/s10639-020-10375-1.

[60] N. Burbules, G. Fan, and P. Repp, “Five trends of education and technology in a sustainable future,” Geography Sustain., vol. 1, pp. 93–97, May 2020, doi: 10.1016/j.geosus.2020.05.001.

[61] T. Trust, “Why do we need technology in education?,” J. Digit. Learn. Teacher Educ., vol. 34, no. 2, pp. 54–55, Apr. 2018, doi: 10.1080/21532974.2018.1442073.

[62] J. Degn-Andersen, “Strengthening an organizational knowledge-sharing culture,” in Advances in Knowledge Acquisition, Transfer, and Manage- ment, D. Tessier Ed., Hershey, PA, USA: IGI Global, 2021, pp. 237–258.

[63] I. Arpaci, M. Al-Emran, and M. A. Al-Sharafi, “The impact of knowl- edge management practices on the acceptance of Massive Open On- line Courses (MOOCs) by engineering students: A cross-cultural com- parison,” Telematics Informat., vol. 54, Nov. 2020, Art. no. 101468, doi: 10.1016/j.tele.2020.101468.

[64] L. R. Kalankesh, Z. Nasiry, R. Fein, and S. Damanabi, “Factors influ- encing user satisfaction with information systems: A systematic review,” Galen Med. J., vol. 9, Jun. 2020, Art. no. e1686, doi: 10.31661/gmj.v9i0. 1686.

[65] E. Cao, Y. Duan, J. Jiang, H. Peng, and W. Hu, “Exploring the pos- itive user experience possibilities based on product emotion theory: A beverage unmanned retail terminal case,” Front. Psychol., vol. 13, 2022, Art. no. 889664.

[66] A. Anand, P. Centobelli, and R. Cerchione, “Why should I share knowledge with others? A review-based framework on events leading to knowledge hiding,” J. Org. Change Manage., vol. 33, no. 2, pp. 379–399, Jul. 2020, doi: 10.1108/JOCM-06-2019-0174.

[67] M. Al-Emran, A. A. AlQudah, G. A. Abbasi, M. A. Al-Sharafi, and M. Iranmanesh, “Determinants of using AI-based chatbots for knowledge sharing: Evidence from PLS-SEM and fuzzy sets (fsQCA),” IEEE Trans. Eng. Manage., vol. 71, pp. 4985–4999, 2024, doi: 10.1109/TEM.2023. 3237789.

[68] J. Zhang, Y. Ma, and B. Lyu, “Relationships between user knowledge sharing in virtual community with community loyalty and satisfac- tion,” Psychol. Res. Behav. Manage., vol. 14, pp. 1509–1523, Sep. 2021, doi: 10.2147/PRBM.S331132.

[69] C. Kooli, “Chatbots in education and research: A critical examination of ethical implications and solutions,” Sustainability, vol. 15, no. 7, Mar. 2023, Art. no. 5614, doi: 10.3390/su15075614.

[70] F. Ahmad and G. Widén, “Knowledge sharing and language di- versity in organisations: Influence of code switching and conver- gence,” Eur. J. Int. Manage., vol. 12, no. 4, pp. 351–373, Jul. 2018, doi: 10.1504/EJIM.2018.092839.

[71] S. Y. Sung and J. N. Choi, “Effects of diversity on knowledge sharing and creativity of work teams: Status differential among members as a facilitator,” Hum. Perform., vol. 32, nos. 3/4, pp. 145–164, Jul. 2019, doi: 10.1080/08959285.2019.1639712.

[72] E. Bolisani and C. Brătianu, “The elusive definition of knowledge,” in Emergent Knowledge Strategies. Knowledge Management and Organiza- tional Learning. New York, NY, USA: Springer, 2017, pp. 1–22.

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

1352 IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 17, 2024

[73] T. R. A. Mukhallafi, “Using artificial intelligence for developing English language teaching/learning: An analytical study from university students’ perspective,” Int. J. English Linguistics, vol. 10, no. 6, pp. 40–53, 2020, doi: 10.5539/ijel.v10n6p40.

[74] X. Hu, Y. Tian, K. Nagato, M. Nakao, and A. Liu, “Opportunities and challenges of ChatGPT for design knowledge management,” Procedia CIRP, vol. 119, pp. 21–28, 2023.

[75] M. H. Jarrahi, D. Askay, A. Eshraghi, and P. Smith, “Artificial in- telligence and knowledge management: A partnership between hu- man and AI,” Bus. Horiz., vol. 66, no. 1, pp. 87–99, Jan. 2023, doi: 10.1016/j.bushor.2022.03.002.

[76] Y. Abbas, A. Martinetti, M. Rajabalinejad, F. Schuberth, and L. A. M. van Dongen, “Facilitating digital collaboration through knowledge man- agement: A case study,” Knowl. Manage. Res. Pract., vol. 20, no. 6, pp. 797–813, Feb. 2022, doi: 10.1080/14778238.2022.2029597.

[77] S. Chandra and S. Palvia, “Online education next wave: Peer to peer learning,” J. Inf. Technol. Case Appl. Res., vol. 23, no. 3, pp. 157–172, Sep. 2021, doi: 10.1080/15228053.2021.1980848.

[78] M. M. Najeeb, M. I. Hanif, and A. B. A. Hamid, “The impact of knowledge management (KM) and organizational commitment (OC) on employee job satisfaction (EJS) in banking sector of Pakistan,” Int. J. Manage. Excellence, vol. 11, no. 1, pp. 1476–1491, Jun. 2018, doi: 10.17722/ijme.v11i1.996.

[79] J. F. Hair, R. E. Anderson, and R. L. Tatham, Multivariate Data Analysis With Readings, 2nd ed. New York, NY, USA: Macmillan, 1987.

[80] J. F. Hair, M. C. Howard, and C. Nitzl, “Assessing measurement model quality in PLS-SEM using confirmatory composite analysis,” J. Bus. Res., vol. 109, pp. 101–110, Mar. 2020, doi: 10.1016/j.jbusres.2019.11.069.

[81] K. S. Taber, “The use of Cronbach’s alpha when developing and reporting research instruments in science education,” Res. Sci. Educ., vol. 48, no. 6, pp. 1273–1296, Jun. 2017, doi: 10.1007/s11165-016-9602-2.

[82] J. F. Hair, G. T. M. Hult, C. M. Ringle, M. Sarstedt, N. P. Danks, and S. Ray, “Evaluation of formative measurement models,” in Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R: A Workbook, J. F. Hair, Jr., G. T. M. Hult, C. M. Ringle, M. Sarstedt, N. P. Danks, and S. Ray, Eds., Cham, Switzerland: Springer, Nov. 2021, pp. 91–113, doi: 10.1007/978-3-030-80519-7_5.

[83] G. James, D. Witten, T. Hastie, and R. Tibshirani, An Introduction to Sta- tistical Learning: With Applications in R. New York, NY, USA: Springer, 2013.

[84] J. Hair, G. T. M. Hult, C. Ringle, and M. Sarstedt, A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). Newbury Park, CA, USA: Sage, 2022.

[85] J. Cohen, Statistical Power Analysis For the Behavioral Sciences, 2nd ed. New York, NY, USA: Routledge, 1988.

[86] A. A. Daneji, A. F. M. Ayub, and M. N. Md. Khambari, “The effects of per- ceived usefulness, confirmation and satisfaction on continuance intention in using massive open online course (MOOC),” Knowl. Manage. E-Learn.: Int. J., vol. 11, pp. 201–214, Jun. 2019, doi: 10.34105/j.kmel.2019.11. 010.

[87] M. T. Rajeh et al., “Students’ satisfaction and continued intention toward e-learning: A theory-based study,” Med. Educ. Online, vol. 26, no. 1, Aug. 2021, Art. no. 1961348, doi: 10.1080/10872981.2021.1961348.

[88] T. T. A. Ngo, “The perception by university students of the use of ChatGPT in education,” Int. J. Emerg. Technol. Learn., vol. 18, no. 17, pp. 4–19, 2023, doi: 10.3991/ijet.v18i17.39019.

[89] I. Pozón-López, E. Higueras-Castillo, F. Muñoz-Leiva, and F. J. Liébana- Cabanillas, “Perceived user satisfaction and intention to use massive open online courses (MOOCs),” J. Comput. High Educ., vol. 33, no. 1, pp. 85–120, Jun. 2020, doi: 10.1007/s12528-020-09257-9.

[90] X. Wang et al., “Perceived usefulness predicts second language learners’ continuance intention toward language learning applications: A serial multiple mediation model of integrative motivation and flow,” Educ. Inf. Technol., vol. 27, no. 4, pp. 5033–5049, Jan. 2022, doi: 10.1007/s10639-021-10822-7.

[91] M. Sohail, Z. Mohsin, and S. Khaliq, “User satisfaction with an AI-enabled customer relationship management chatbot,” in Communications in Com- puter and Information Science, C. Stephanidis, M. Antona, and S. Ntoa, Eds., Cham, Switzerland: Springer, 2021, pp. 279–287.

[92] N. Dhiman and M. Jamwal, “Tourists’ post-adoption continuance inten- tions of chatbots: Integrating task–technology fit model and expectation– confirmation theory,” Foresight, vol. 25, no. 2, pp. 209–224, Sep. 2022, doi: 10.1108/FS-10-2021-0207.

[93] S. Borsci et al., “The chatbot usability scale: The design and pi- lot of a usability scale for interaction with AI-based conversational agents,” Pers. Ubiquitous Comput., vol. 26, no. 1, pp. 95–119, Jul. 2021, doi: 10.1007/s00779-021-01582-9.

[94] I. Matar and J. Raudeliuniene, “The role of knowledge acquisition in enhancing knowledge management processes in higher education institu- tions,” in Proc. Int. Sci. Conf. Contemporary Issues Bus., Manage. Econ. Eng., 2021, pp. 1–8, doi: 10.3846/cibmee.2021.646.

[95] I. Arpaci, “Antecedents and consequences of cloud computing adoption in education to achieve knowledge management,” Comput. Hum. Behav., vol. 70, pp. 382–390, May 2017, doi: 10.1016/j.chb.2017.01.024.

[96] Y. Li and J. Wang, “Evaluating the impact of information system quality on continuance intention toward cloud financial information system,” Front. Psychol., vol. 12, 2021, Art. no. 713353.

[97] I. Arpaci, “A multianalytical SEM-ANN approach to investigate the social sustainability of AI chatbots based on cybersecurity and protection moti- vation theory,” IEEE Trans. Eng. Manage., vol. 71, pp. 1714–1725, 2024, doi: 10.1109/TEM.2023.3339578.

Thi Thuy An Ngo received the first master’s degree in aquaculture and the second master’s degree in Economics from Ghent University, Ghent, Belgium, 2011 and 2015.

She is currently a Lecturer in the business field and is the Head of the Soft Skills Department, FPT University, Can Tho, Vietnam. She has participated in scientific research projects in various fields and has authored or coauthored several papers in high-quality international journals. Her research interests include education, innovative marketing, consumer behavior,

business management, social media communication, entrepreneurship, and sus- tainable development.

Thanh Tu Tran is working toward the bachelor’s degree in international business with FPT University, Can Tho, Vietnam.

He has participated in many scientific research projects in many different fields and has authored or coauthored articles in high-quality international journals. His research interests include artificial intel- ligence, human–computer interaction, business man- agement, innovation, sustainable development, en- trepreneurship, and consumer behavior.

Gia Khuong An is working toward the bachelor’s degree in international business with FPT University, Can Tho, Vietnam.

He has participated in many scientific research projects in many different fields and has authored or coauthored articles in high-quality international journals. His research interests include education, artificial intelligence, innovation, business adminis- tration, consumer behavior, sustainable development, and entrepreneurship.

Phuong Thy Nguyen is working toward the bache- lor’s degree in international business with FPT Uni- versity, Can Tho, Vietnam.

She has participated in many scientific research projects in many different fields and has authored or coauthored articles in high-quality international jour- nals. Her research interests include education, arti- ficial intelligence, business administration, consumer behavior, sustainable development, entrepreneurship, and innovation.

Authorized licensed use limited to: American Public University System. Downloaded on May 20,2025 at 21:01:55 UTC from IEEE Xplore. Restrictions apply.

<< /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (Gray Gamma 2.2) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Warning /CompatibilityLevel 1.4 /CompressObjects /Off /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.0000 /ColorConversionStrategy /sRGB /DoThumbnails true /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams true /MaxSubsetPct 100 /Optimize true /OPM 0 /ParseDSCComments false /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo false /PreserveFlatness true /PreserveHalftoneInfo true /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Remove /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true /Algerian /Arial-Black /Arial-BlackItalic /Arial-BoldItalicMT /Arial-BoldMT /Arial-ItalicMT /ArialMT /ArialNarrow /ArialNarrow-Bold /ArialNarrow-BoldItalic /ArialNarrow-Italic /ArialUnicodeMS /BaskOldFace /Batang /Bauhaus93 /BellMT /BellMTBold /BellMTItalic /BerlinSansFB-Bold /BerlinSansFBDemi-Bold /BerlinSansFB-Reg /BernardMT-Condensed /BodoniMTPosterCompressed /BookAntiqua /BookAntiqua-Bold /BookAntiqua-BoldItalic /BookAntiqua-Italic /BookmanOldStyle /BookmanOldStyle-Bold /BookmanOldStyle-BoldItalic /BookmanOldStyle-Italic /BookshelfSymbolSeven /BritannicBold /Broadway /BrushScriptMT /CalifornianFB-Bold /CalifornianFB-Italic /CalifornianFB-Reg /Centaur /Century /CenturyGothic /CenturyGothic-Bold /CenturyGothic-BoldItalic /CenturyGothic-Italic /CenturySchoolbook /CenturySchoolbook-Bold /CenturySchoolbook-BoldItalic /CenturySchoolbook-Italic /Chiller-Regular /ColonnaMT /ComicSansMS /ComicSansMS-Bold /CooperBlack /CourierNewPS-BoldItalicMT /CourierNewPS-BoldMT /CourierNewPS-ItalicMT /CourierNewPSMT /EstrangeloEdessa /FootlightMTLight /FreestyleScript-Regular /Garamond /Garamond-Bold /Garamond-Italic /Georgia /Georgia-Bold /Georgia-BoldItalic /Georgia-Italic /Haettenschweiler /HarlowSolid /Harrington /HighTowerText-Italic /HighTowerText-Reg /Impact /InformalRoman-Regular /Jokerman-Regular /JuiceITC-Regular /KristenITC-Regular /KuenstlerScript-Black /KuenstlerScript-Medium /KuenstlerScript-TwoBold /KunstlerScript /LatinWide /LetterGothicMT /LetterGothicMT-Bold /LetterGothicMT-BoldOblique /LetterGothicMT-Oblique /LucidaBright /LucidaBright-Demi /LucidaBright-DemiItalic /LucidaBright-Italic /LucidaCalligraphy-Italic /LucidaConsole /LucidaFax /LucidaFax-Demi /LucidaFax-DemiItalic /LucidaFax-Italic /LucidaHandwriting-Italic /LucidaSansUnicode /Magneto-Bold /MaturaMTScriptCapitals /MediciScriptLTStd /MicrosoftSansSerif /Mistral /Modern-Regular /MonotypeCorsiva /MS-Mincho /MSReferenceSansSerif /MSReferenceSpecialty /NiagaraEngraved-Reg /NiagaraSolid-Reg /NuptialScript /OldEnglishTextMT /Onyx /PalatinoLinotype-Bold /PalatinoLinotype-BoldItalic /PalatinoLinotype-Italic /PalatinoLinotype-Roman /Parchment-Regular /Playbill /PMingLiU /PoorRichard-Regular /Ravie /ShowcardGothic-Reg /SimSun /SnapITC-Regular /Stencil /SymbolMT /Tahoma /Tahoma-Bold /TempusSansITC /TimesNewRomanMT-ExtraBold /TimesNewRomanMTStd /TimesNewRomanMTStd-Bold /TimesNewRomanMTStd-BoldCond /TimesNewRomanMTStd-BoldIt /TimesNewRomanMTStd-Cond /TimesNewRomanMTStd-CondIt /TimesNewRomanMTStd-Italic /TimesNewRomanPS-BoldItalicMT /TimesNewRomanPS-BoldMT /TimesNewRomanPS-ItalicMT /TimesNewRomanPSMT /Times-Roman /Trebuchet-BoldItalic /TrebuchetMS /TrebuchetMS-Bold /TrebuchetMS-Italic /Verdana /Verdana-Bold /Verdana-BoldItalic /Verdana-Italic /VinerHandITC /Vivaldii /VladimirScript /Webdings /Wingdings2 /Wingdings3 /Wingdings-Regular /ZapfChanceryStd-Demi /ZWAdobeF ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 150 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages false /ColorImageDownsampleType /Bicubic /ColorImageResolution 900 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.00111 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /ColorImageDict << /QFactor 0.40 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages false /GrayImageDownsampleType /Bicubic /GrayImageResolution 1200 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.00083 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /GrayImageDict << /QFactor 0.40 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages false /MonoImageDownsampleType /Bicubic /MonoImageResolution 1600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.00063 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description << /CHS <FEFF4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e55464e1a65876863768467e5770b548c62535370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /CHT <FEFF4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc666e901a554652d965874ef6768467e5770b548c52175370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /DAN <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> /DEU <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> /ESP <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> /FRA <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> /ITA (Utilizzare queste impostazioni per creare documenti Adobe PDF adatti per visualizzare e stampare documenti aziendali in modo affidabile. I documenti PDF creati possono essere aperti con Acrobat e Adobe Reader 5.0 e versioni successive.) /JPN <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> /KOR <FEFFc7740020c124c815c7440020c0acc6a9d558c5ec0020be44c988b2c8c2a40020bb38c11cb97c0020c548c815c801c73cb85c0020bcf4ace00020c778c1c4d558b2940020b3700020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken waarmee zakelijke documenten betrouwbaar kunnen worden weergegeven en afgedrukt. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.) /NOR <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> /PTB <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> /SUO <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> /SVE <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> /ENU (Use these settings to create PDFs that match the "Suggested" settings for PDF Specification 4.0) >> >> setdistillerparams << /HWResolution [600 600] /PageSize [612.000 792.000] >> setpagedevice