Therefore, the authors decided to conduct a research on the topic: “ The Impact of Ho Chi Minh City University Students’ Using Generative AI GenAI for Learning on Self-learning Motivatio
INTRODUCTION
Reason for choosing research topic (Problem Statement)
The rapid advancement of science and technology, coupled with globalization, has significantly transformed education, particularly with the rise of generative artificial intelligence (GenAI) applications in late 2022 and early 2023, which are reshaping teaching and learning methods in universities worldwide To adapt to the evolving knowledge economy, educational institutions must innovate and effectively integrate information technology As highlighted in the Southeast Region of Vietnam's Digital Education Transformation Cooperation Workshop, GenAI not only enhances student learning but also has the potential to revolutionize traditional educational approaches.
Self-learning is a vital component of the educational journey, empowering students to take initiative in their learning by identifying and organizing their study efforts In situations where communication with instructors and peers is limited, self-learning becomes crucial for overcoming obstacles and advancing knowledge This approach not only cultivates essential skills such as research, analysis, and information organization but also plays a significant role in fostering lifelong learning Self-learning motivation, driven by interest, personal achievement, or enjoyment, is key to helping students navigate challenges and enhances their overall progress and personal development, ultimately improving the quality of social education.
Artificial intelligence, particularly Generative AI (GenAI), is transforming learning methods and enhancing self-directed learning for students Traditional teaching methods often left students struggling to find appropriate resources and lacking motivation due to limited interaction and personalization However, GenAI now provides students with access to a wide array of resources, tailored learning experiences, and optimal support through advanced AI technology.
GenAI technology helps students save time and effort in searching for materials, ensuring the reliability and updating of the information they receive.
The research team conducted a study on "The Impact of Ho Chi Minh City University Students' Using Generative AI for Learning on Self-learning Motivation," highlighting the significance of Generative AI in enhancing lifelong learning for students Their findings reveal how the integration of Generative AI tools can boost self-learning motivation among university students, emphasizing the transformative role of technology in education.
This study investigates the impact of Generative AI (GenAI) on students' self-learning motivation in Ho Chi Minh City, a vibrant and culturally diverse educational hub in Vietnam It highlights the changes in self-learning motivation following the integration of GenAI and aims to advance modern educational methods Furthermore, the research addresses the opportunities and challenges that GenAI presents in urban educational settings, particularly in Ho Chi Minh City, where innovation and technology are actively embraced.
Research Objectives
This research investigates the effects of Generative AI (GenAI) on the self-learning motivation of students in Ho Chi Minh City, focusing on three critical psychological factors: perceived autonomy, perceived competence and relatedness, and perceived usefulness alongside intrinsic motivation.
This research aims to identify the key factors that impact student motivation within the GenAI learning environment, with the goal of providing recommendations to optimize educational practices and fully leverage GenAI's potential to enhance self-directed learning.
Research Questions
- Do perceived autonomy, perceived competence, perceived relatedness, perceived usefulness, and intrinsic motivation affect the university students' self learning motivation in Ho Chi Minh City?
- To what extent do these factors influence the self-learning motivation of
Ho Chi Minh City university students in the context of GenAI for learning?
To enhance self-learning motivation among university students in Ho Chi Minh City, it is recommended to integrate Generative AI (GenAI) into learning environments This can be achieved by incorporating personalized learning experiences, fostering interactive and engaging content, and providing timely feedback through AI-driven tools Additionally, training educators to effectively utilize GenAI in their teaching methods will ensure that students receive the support they need to thrive in their self-directed learning journeys By creating a collaborative and adaptive learning atmosphere, institutions can significantly boost student motivation and academic performance.
Research sample and scope
The survey targets students from various universities across Ho Chi Minh city who have utilized GenAI for learning purposes.
Space Scope: The study encompasses both public and private universities across various fields within Ho Chi Minh City.
Time Scope: Data collection occurred between January 25th and February 7th, 2024.
Research Method
This study employs both qualitative and quantitative research methods, with quantitative research as the primary method.
This qualitative research investigates the theoretical foundations of self-learning motivation, Generative AI, and Self-Determination Theory Through the analysis of relevant studies, the authors develop a measurement scale and research model Data is gathered and synthesized from multiple sources, including library resources.
At Ho Chi Minh City University of Economics, research publications by both domestic and foreign authors are utilized alongside relevant databases and reference books Theoretical research plays a crucial role in identifying problems and relationships among research factors, providing foundational theories for the development of the research model Leveraging this theoretical framework, authors adapt and refine existing scales to accurately measure research concepts Group discussions and consultations with supervisors are conducted to tailor the scale and terminology to the specific research environment The refined scale is then employed to create the survey questionnaire.
The study employs a quantitative research approach, beginning with the design of a questionnaire to collect and process data aimed at quantifying the relationship between variables It specifically investigates the effect of using GenAI for learning on students' self-learning motivation in Ho Chi Minh City Data is gathered through surveys, and tools such as Cronbach's Alpha coefficient, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and linear structural equation modeling (SEM) are utilized to ensure the reliability and validity of the research model Additionally, statistical analyses like Independent Sample T-test and One-way ANOVA are performed using SPSS and AMOS software to test the research hypotheses.
THEORETICAL FOUNDATIONS
Generative Artificial Intelligence (GenAI)
The evolution of artificial intelligence (AI) reflects a dynamic and expanding field, initially defined as a program designed to operate in various environments with human-like adaptability and problem-solving skills This foundational goal aimed to create systems capable of navigating and responding to real-world complexities as effectively as humans More recent definitions have built upon this initial perspective, highlighting the continuous advancements in AI technology and its applications.
The evolution of artificial intelligence emphasizes the imitation of intelligent human behavior, proposing a stronger framework for AI development (Kok, J N., 2009) This transition underscores a commitment to replicating human cognitive functions, aiming to create systems that not only execute tasks but also emulate human thought processes and decision-making abilities Additionally, McCarthy's foundational definition of AI as "the science and engineering of making intelligent machines" further solidifies the field's objectives.
The foundational principles established in 1956 highlight the necessity of an interdisciplinary approach to reach ambitious objectives in artificial intelligence (McCarthy, 2007, p.2) This evolution reflects a shift from viewing AI as merely functionally equivalent to human performance to a more sophisticated goal of emulating the complexities of human intelligence.
Generative AI, or Generative Artificial Intelligence, is a revolutionary advancement in the AI field focused on creating unique content across various data types, including text, images, sound, and video Utilizing sophisticated machine learning models, it analyzes large datasets to identify patterns and autonomously generate new content, reducing the need for human intervention This technology has found applications in diverse industries such as art, entertainment, business, and education, with notable contributions from companies like Google and OpenAI, which have developed systems that closely mimic human output Additionally, Generative AI plays a crucial role in breaking down language barriers, aiding non-English speaking students through language editing and translation services Its multifaceted capabilities not only demonstrate its potential to replicate human intelligence and creativity but also enhance access to information and promote global communication and educational opportunities Prominent Generative AI tools include ChatGPT, Gemini, and Claude AI.
Generative AI encompasses a wide range of applications, with OpenAI's DALL-E and ChatGPT as leading examples DALL-E excels in generating realistic images from text descriptions and can create visuals in various artistic styles, including abstract art Meanwhile, ChatGPT, particularly the GPT-3.5 model, has revolutionized natural language processing by producing human-like text that is coherent and fluid, enhancing capabilities in answering questions, essay writing, and conversation These innovations underscore the significant impact of Generative AI on the creative industry, content creation, and human-machine interaction, setting the stage for future advancements in image and text generation.
Theories Related to Self-Learning Motivation
Motivation plays a vital role in psychology, encompassing the factors that compel individuals to take action Researchers across various disciplines have explored this concept, with a significant emphasis on its implications in economics and education (Nguyen Binh Phuong Duy, 2015).
Motivation is defined as the underlying reasons for behavior, encompassing both internal and external energies that drive individuals to take action towards specific tasks (Guay et al., 2010; Pinder, 2008) It is a theoretical construct that explains the initiation, direction, intensity, persistence, and quality of goal-directed behavior (Brophy, 2010) In Vietnam, motivation is described as the source of energy that propels individuals to strive for improvement (Pham Minh Hac, 2013).
In this study, motivation is defined as the reasons or sources of energy that drive individuals to act, work, or achieve goals.
As knowledge expands beyond traditional classroom boundaries, self-learning has become essential Le Khanh Bang (1998) describes self-learning as the independent use of intellectual and psychological abilities to master a specific field Nguyen Canh Toan (2001) emphasizes the importance of critical thinking, observation, and personal qualities such as motivation and perseverance in the self-learning process This form of cognitive activity allows individuals to acquire knowledge and skills independently, whether inside or outside the classroom, without adhering to a strict curriculum (Dang Vu Boat, 2004) Additionally, Dang Thanh Hung (2012) defines self-learning as an autonomous strategy where learners take charge of their educational journey, determining their goals, content, methods, and resources.
According to the research conducted by the author group, self-learning is defined as an integral aspect of the learning process, where individuals proactively pursue and acquire knowledge independently, extending beyond the information provided through formal education, without requiring external guidance.
Self-learning motivation plays a crucial role in achieving success in both education and life Research indicates that learners who are motivated tend to be more engaged, exhibit higher performance levels, take on challenging tasks, and demonstrate resilience in overcoming obstacles.
According to Paris and Turner (1994), motivation serves as the "engine" of learning, propelling students toward their goals It plays a crucial role in shaping what we learn, how we learn, and the timing of our learning experiences (Schunk & Usher, 2012).
Self-learning motivation is defined in this research study as the driving force that compels individuals to begin and maintain their own learning journey independently, without external guidance.
Self-Determination Theory
Self-Determination Theory (SDT), developed by psychologists Edward L Deci and Richard M Ryan, emphasizes the significance of three fundamental psychological needs: autonomy, competence, and relatedness, along with the role of autonomous motivation in learning and development Since the publication of their work in 2000, numerous experimental studies have explored how these needs influence various aspects of human behavior and well-being Research from the Centre for Self-Determination Theory highlights that fulfilling these psychological needs is crucial for healthy human development and psychological health, as evidenced by studies from Ng et al (2012) and others Conversely, obstructing these needs can hinder individuals' ability to adapt to their environments and achieve personal growth.
The three basic psychological needs—autonomy, competence, and relatedness—are interconnected and can support one another, yet they can also function independently, influencing individual behavior in various ways (Ryan & Deci, 2000) When these needs are fulfilled, they foster autonomous motivation, leading to enhanced learning, personal development, and a more fulfilling life (Ryan & Deci, 2000) Conversely, obstructing any of these needs can negatively impact motivation and learning outcomes (Ryan & Deci, 2000).
Ryan & Deci (2000, 2017, 2020) further defined and studied the source of autonomous motivation in humans Autonomous motivation is a key concept in SDT
Self-Determination Theory emphasizes the intrinsic motivation that arises from within individuals, driven by curiosity, the desire to learn, and personal growth rather than external rewards or punishments When the three basic psychological needs are fulfilled, individuals experience higher levels of autonomous motivation, which enhances their interest in work, as highlighted by Amabile's (1996) research Furthermore, Deci & Ryan (2008) demonstrated that autonomous motivation fosters more effective learning and greater creativity.
According to the Centre for Self-Determination Theory (CSDT), various contexts significantly influence basic psychological needs, affecting individual motivation and happiness Self-Determination Theory (SDT) has crucial implications in education, as it can forecast the relationships between the quality of the learning experience and its outcomes Research indicates that fulfilling the three psychological needs of students is essential for cultivating autonomous motivation, which enhances both learning quality and overall student development (Deci, Ryan & Koestner, 1999) Thus, SDT is vital for understanding students' self-learning motivation.
Perceived autonomy refers to the feeling of agency and ownership over one's actions, which is strengthened by engaging in experiences that are interesting and valuable In contrast, autonomy is diminished when individuals feel controlled by external factors, such as rewards or punishments When people experience autonomy, they enjoy a sense of personal integrity in their actions, thoughts, and emotions Conversely, a lack of autonomy can lead to feelings of pressure and internal conflict, as individuals may feel compelled to pursue paths they do not genuinely want.
Research indicates that students with high autonomy in their learning experience greater academic achievement and interest in their studies Additionally, a study found that those who had the freedom to choose their major reported higher satisfaction and increased motivation to engage in their coursework.
In Self-Determination Theory, Perceived Competence highlights an individual's belief in their effectiveness and ability to succeed It encompasses the feeling of being capable of engaging in activities that allow for skill development and expansion A strong sense of competence fosters confidence and motivation, while a lack of it can lead to feelings of ineffectiveness, failure, and helplessness.
Supporting this notion, Ryan and Deci (2002) conducted a study that found that students who perceive themselves as capable learners tend to set higher goals and persist more in achieving them.
Relatedness, a core concept in Self-Determination Theory, emphasizes the essential human need for connection, love, and support from others This psychological need is characterized by feelings of warmth, trust, closeness, and care Conversely, a deficiency in relatedness can lead to negative experiences such as social isolation, ostracism, and loneliness.
Research by Ryan and Deci (2001) indicates that positive relationships between students and their teachers or peers enhance student happiness and motivation to learn Further studies by Wentzel (2003) reveal that students who feel supported in their educational environment are more likely to achieve higher academic success and demonstrate a greater interest in their studies.
Research in Self-Determination Theory (SDT) initially concentrated on intrinsic motivation, which arises from within the individual rather than from external influences such as rewards or punishments According to Deci and Ryan (2000), intrinsic motivation involves participating in activities for their own sake, driven by inherent interest and enjoyment This type of motivation fosters feelings of curiosity, pleasure, and satisfaction, highlighting its significance in personal engagement and fulfillment.
Intrinsic motivation is essential for lifelong learning, contrasting with externally imposed learning methods (Ryan & Deci, 2017) It significantly influences student learning and ongoing personal development To foster holistic and sustainable growth, students must focus on enhancing their intrinsic motivation instead of depending solely on external influences.
Extrinsic motivation involves engaging in activities for reasons beyond personal interest, as outlined by Ryan and Deci (2000) According to Self Determination Theory, various external factors influence extrinsic motivation, but individuals can internalize these influences, gaining autonomy over their actions This internalization process categorizes extrinsic motivation into four types, with one type allowing individuals to recognize the value of an activity while aligning it with their own interests or values (Ryan & Deci, 2017) Consequently, when individuals grasp the significance of their actions, even when driven by external factors, they experience greater comfort and autonomy in their pursuits (Ryan & Deci, 2017).
"Perceived usefulness" is a crucial factor in understanding how external influences affect learners, as highlighted by Deci et al (1994) It refers to a learner's belief that utilizing a learning support tool can effectively address their learning challenges and enhance their educational outcomes This perception significantly motivates learners to take a more proactive approach to their studies A study by Yumei Luo et al (2021) revealed that students who recognized the value of online learning tools exhibited greater engagement and achieved superior learning results, underscoring the vital role of perceived usefulness in fostering proactive learning behaviors.
Perceived usefulness is crucial for students' proactive engagement with learning support tools By focusing on the research and development of effective learning support tools, we can enhance students' learning outcomes significantly.
Research overview
2.4.1 The Impact of ChatGPT on Learning Motivation: A Study Based on Self Determination Theory (Zhou, L & Li, J J., 2023)
Research Objective: This study aims to investigate the impact of using ChatGPT as an auxiliary learning tool on the learning motivation of university students.
Research Method: The study used a questionnaire survey to collect data from 196 university students The linear structural equation model (SEM) and regression analysis were used as data analysis methods.
The research revealed a negative correlation between pressure stress and both inspiration and motivation in learning after utilizing ChatGPT It found that perceived competence significantly positively correlates with learning motivation, whereas the correlation between perceived value and learning motivation was not significant Regression analysis indicated varying levels of influence among these three variables on motivation Additionally, the study concluded that while ChatGPT does impact students' intrinsic motivation, the frequency of use and proficiency among university students remains relatively low, indicating a need for further training.
This study employs Self-Determination Theory to explore how ChatGPT influences students' learning motivation However, the reliance on a scale for intrinsic motivation to measure learning motivation poses challenges in demonstrating the extent to which students' motivation is impacted.
The study utilized the Intrinsic Motivation Scale developed by McAuley et al (1989) and adapted by Yin et al (2021), yet it does not address the three psychological needs outlined in Self-Determination Theory While the research references Self-Determination Theory, it fails to demonstrate the impact of autonomy and relatedness on students' intrinsic motivation, focusing solely on the perceived competence variable.
- Nevertheless, the study still contributes both theoretically and in terms of results for the authors to refer to.
Figure 2.1 Research Model by Zhou, L & Li, J J (2023)
2.4.2 Exploring the role of intrinsic motivation in ChatGPT adoption to support active learning: An extension of the technology acceptance model (Lai, Chung Yee
Research Objective: To investigate the roles of autonomous motivation (from
Self-Determination Theory) and the factors of the Technology Acceptance Model (TAM) in the acceptance of using ChatGPT for self-directed learning among university students in Hong Kong.
The study employed structural equation modeling (SEM) to evaluate an extended Technology Acceptance Model within higher education In July 2023, data was gathered from 473 undergraduate students in Hong Kong using self-report questionnaires The reliability and validity of the collected data were assessed through confirmatory factor analysis (CFA), which was subsequently followed by path analysis to explore the hypotheses outlined in the proposed model.
The research findings highlight that intrinsic motivation is the primary driver for the intention to use ChatGPT in facilitating self-directed learning Additionally, the study identifies perceived usefulness as a significant predictor of the behavioral intention to utilize ChatGPT for educational purposes These insights enhance our understanding of the importance of intrinsic motivation in leveraging ChatGPT to support students' independent learning efforts.
The research utilized an extended Technology Acceptance Model (TAM) alongside Self-Determination Theory to investigate the impact of intrinsic motivation on students' learning motivation The findings highlight the significant effects of both intrinsic motivation and perceived usefulness on the adoption of Generative technologies in educational settings.
AI technology in student-led learning.
The study, while utilizing the Technology Acceptance Model (TAM) rather than Self-Determination Theory (SDT), offers valuable theoretical contributions to the existing literature, despite not being able to demonstrate the impact of the three basic psychological needs.
Figure 2.2 Research Model by Chung Yee Lai, Kwok Yip Cheung, Chee
Source: Chung Yee Lai ei al (2023)
2.4.3 Examining the Impacts of ChatGPT on Student Motivation and Engagement (Munoz & et al., 2023).
This study investigates the influence of ChatGPT on student learning motivation and engagement, focusing on two main objectives: first, to assess the perceptions of both teachers and students regarding ChatGPT's role in enhancing student motivation during the learning process; and second, to explore the correlation between these perceptions to determine if they significantly align in the context of using ChatGPT for educational purposes.
Research Method: Information was gathered via a survey of 350 students and instructors The data were analyzed statistically using ANOVA and post hoc multiple comparison tests.
A recent study revealed that integrating ChatGPT into education significantly enhances student motivation and engagement As a result, it is essential for policymakers to advocate for the inclusion of ChatGPT in the educational framework to boost student learning outcomes.
- The study did not explicitly examine the role of Self-Determination Theory in using ChatGPT to enhance learning motivation.
- However, the study contributes significantly to clarifying how effectively using ChatGPT can increase motivation and learning outcomes.
2.4.4 The mediating effects of needs satisfaction on the relationships between prior knowledge and self-regulated learning through artificial intelligence chatbot (Xia,
Research Objective: This study examines the mediating effect of Self
Determination Theory (SDT) basic psychological needs on the relationship between artificial intelligence (AI) knowledge, students' prior English proficiency, and self regulated learning (SRL) using Generative Al (GenAl).
Research Method: Data were collected from 323 9th-grade students through a questionnaire and a test The students completed a basic AI unit and then learned English with GenAI for 5 days.
The study revealed that students' prior knowledge of English significantly influences their self-regulated learning (SRL) with Generative AI (GenAI), while their familiarity with AI itself does not have the same impact Additionally, the need for autonomy and competence plays a crucial role in mediating the relationship between both English and AI knowledge and SRL, whereas the need for relatedness does not The inherently self-directed nature of SRL demands substantial cognitive effort, and fulfilling the needs for autonomy and competence can enhance engagement among young learners in this learning process Furthermore, the findings indicate that current GenAI technologies (as of 2023) may not provide advantages for students with lower English proficiency.
The research utilized Self-Determination Theory (SDT) as a mediator to explore the self-regulated learning (SRL) behaviors of ninth graders While SRL is defined by its link to autonomous motivation, the findings revealed that fulfilling the three basic psychological needs plays a crucial role in enhancing students' learning capabilities and motivation.
Figure 2.3 Research Model by Qi Xia, Thomas K F Chiu, Ching Sing Chai,
Hypotheses and Research model
The rise of Generative AI (GenAI) is transforming the educational landscape by introducing innovative technologies that enhance student learning Research has explored GenAI's effects on student motivation, inspiration, classroom engagement, and self-directed learning skills Nonetheless, there remains a gap in studies applying Self-Determination Theory to examine how autonomous motivation influences the relationship between GenAI and students' motivation and learning capabilities.
Previous research primarily examines the overall effects of Generative AI (GenAI) on learning motivation without differentiating between intrinsic and extrinsic motivation sources For instance, Zhou and Li (2023) found that GenAI enhances student engagement and interest but focused solely on intrinsic motivation, neglecting the role of internalized extrinsic motivation Similarly, Xia et al (2023) analyzed the impact of AI and English knowledge on self-regulated learning through the lens of the three basic psychological needs but overlooked the intrinsic and extrinsic motivation aspects outlined in Self Determination Theory Additionally, Lai et al (2023) discussed how these two motivation types affect the intention to use ChatGPT for self-directed learning, yet failed to explore how the three basic psychological needs influence these motivational sources.
Self-Determination Theory emphasizes the importance of autonomy, competence, and relatedness in enhancing learning motivation According to Chiu et al (2023), Generative AI can fulfill these psychological needs by enabling students to select their own questions and answers, as highlighted by Shawar & Atwell (2007), and by providing immediate feedback, as noted by Smutny.
& Schreiberova, 2020; Yin et al., 2021), and feel like they are talking to a human-like machine, it can have a positive impact on learner motivation.
Applying Self-Determination Theory to examine the influence of Generative AI on students' self-learning motivation can enhance our understanding of AI's role in education (Chiu et al., 2023) Engaging with technology in learning environments can boost student satisfaction, subsequently increasing their interest and motivation (Heaven, 2020; Kuo et al., 2014) Thus, it is essential to further investigate how Generative AI affects students' self-learning motivation through the lens of Self-Determination Theory, particularly concerning the three fundamental psychological needs and sources of autonomous motivation This research could pave the way for new avenues in the application of Generative AI within educational settings.
(/) Three basic psychological needs and Intrinsic Motivation
Self-Determination Theory posits that fulfilling three essential psychological needs—autonomy, competence, and relatedness—can enhance an individual's intrinsic motivation (Ryan & Deci, 2000) When these needs are satisfied, individuals are naturally motivated to engage in specific behaviors, driven by their sense of autonomy (Ryan & Deci, 2000) This state of autonomous motivation significantly impacts their behavior, emotions, and cognitive processes.
The need for autonomy emphasizes the importance of having the freedom to make personal choices and control one's actions Similarly, the need for competence reflects the desire to feel capable and effective in completing tasks Additionally, the need for relatedness highlights the significance of connecting, interacting, and receiving support from others.
A study by Zhou and Li (2023) highlights the positive influence of student competence on learning motivation, while research by Luo, Lin, and Yang (2021) indicates that fulfilling the three basic psychological needs—autonomy, competence, and relatedness—in online learning environments enhances students' intrinsic motivation and self-regulated learning abilities Furthermore, Deci and Ryan's Self-Determination Theory (1985) suggests that when these psychological needs are satisfied, learners are more likely to engage voluntarily and enthusiastically in their educational activities, leading to greater happiness and satisfaction in their learning journey.
When the three psychological needs outlined in Self Determination Theory are fulfilled, students experience heightened intrinsic motivation while engaging with generative AI Consequently, the authors put forth the following hypotheses.
Hi a: Perceived autonomy has a positive impact (+) on intrinsic motivation when using GenAI.
Hlh: Perceived competence has a positive impact (+) on intrinsic motivation when using Gen AI.
Hlc: Perceived relatedness has a positive impact (+) on intrinsic motivation when using GenAI.
(2) Three basic psychological needs and Perceived Usefulness
Ryan and Deci (2020) emphasize that fulfilling the three basic psychological needs—autonomy, competence, and relatedness—enhances students' intrinsic motivation, which facilitates the internalization of extrinsic motivation and ultimately leads to improved academic achievement When students recognize the value of external motivations, they become more committed and willing to exert effort, thereby boosting their learning motivation Furthermore, perceived usefulness is a key factor in this internalization process (Deci et al., 1994).
Research indicates that students who perceive themselves as autonomous in their e-learning experience find it more beneficial, as noted by Roca and Gagne (2008) Additionally, a study by Yumei Luo, Jinping Lin, and Yi Yang (2021) emphasizes the significant influence of competence on the perceived usefulness of online learning Furthermore, Nikou and Economides (2017) underscore a positive correlation between relatedness and the perceived usefulness of learning activities.
Current scientific research is lacking on how the perceived usefulness of generative AI (GenAI) affects learning motivation Drawing from Self Determination Theory and the concept of perceived usefulness, which relates to the internalization of extrinsic motivation, this study proposes three key hypotheses.
H2a: Perceived Autonomy has a positive (+) impact on Perceived Usefulness with GenAT
H2b: Perceived Competence has a positive (+) impact on Perceived Usefulness with GenAl.
H2c: Perceived Relatedness has a positive (+) impact on Perceived Usefulness with Gen A I.
(3) Intrinsic Motivation, Perceived Usefulness and Self-learning motivation
A study by Lai, Chung Yee et al (2023) emphasizes the significant role of intrinsic motivation in enhancing perceived usefulness According to Self-Determination Theory, the internalization process allows individuals to align their actions with their interests and values (Ryan & Deci, 2017) When individuals experience high intrinsic motivation, they tend to invest greater effort into their tasks, resulting in improved outcomes and a heightened sense of perceived usefulness.
Deci and Ryan's 1985 study revealed that individuals with high intrinsic motivation view their work as more valuable compared to those motivated by external factors lacking autonomy This finding aligns with the research conducted by Lai, Chung Yee, et al., which further supports the importance of intrinsic motivation in enhancing perceived work value.
(2023) on GenAI and the intention to use ChatGPT for learning, perceived usefulness was positively (+) influenced by intrinsic motivation Therefore, this researches propose the hypothesis:
H3: Intrinsic Motivation has a positive (+) impact on Perceived Usefulness.
Self-Determination Theory (Deci & Ryan, 2000) identifies two types of learning motivation: intrinsic motivation, which arises from personal interest, and extrinsic motivation, driven by external factors like rewards or pressure Perceived Usefulness falls under extrinsic motivation, particularly when individuals internalize it When students possess autonomous and internalized motivation, they tend to learn more effectively and achieve superior results, ultimately enhancing the quality of their learning.
Numerous studies highlight the beneficial connection between intrinsic motivation, the perceived usefulness of technology, and students' learning intentions For instance, research by Luo et al (2021) indicates that both intrinsic motivation and the perceived advantages of online learning platforms significantly enhance students' intention to engage in learning activities.
RESEARCH METHODS
Research process
The research process in this study was designed as follows:
Research design
In exploratory factor analysis (EFA), Hair et al (2009) recommend a minimum sample size of 50, with an optimal observation ratio of 5:1 for each variable This implies that each observed variable necessitates at least 5 observations Therefore, for a research model comprising 30 variables, a minimum sample size of 150 is essential to ensure adequate data for effective EFA analysis.
The study aimed to gather a sample size exceeding the minimum requirement to minimize potential losses during the survey process With a minimum target of 150 responses, the survey is designed to collect between 300 and 310 samples to ensure adequate representation, even if initial responses fall short.
Two non-probability sampling methods were applied:
Convenience sampling was employed in this study by utilizing Google Forms to design a questionnaire focused on Self-learning Motivation following the use of Generative AI The authors disseminated the survey through social media platforms, class chat groups, and networks of friends, targeting groups with a significant number of students in Ho Chi Minh City to gather responses effectively.
Snowball Sampling: The authors shared the questionnaire with their seniors, friends in the university environment and asked them to survey and share it with their university friends.
Measurement scales
Informed by qualitative research findings and established measurement scales from prior studies, the research team tailored the scale to fit the specific research context Variables were assessed using a Five-point Likert scale, where respondents indicated their level of agreement on a scale from 1 to 5, with "1" representing "Strongly Disagree," "2" as "Disagree," "3" as "Neutral," "4" as "Agree," and "5" as "Strongly Agree."
The Perceived Autonomy Scale comprises five items, coded from PAI to PA5, with four items (PA2, PA3, PA4, and PA5) derived from the research of Hew and Kadir (2016) Additionally, the authors incorporated item PAI from McAuley et al (1989), believing it to be a fitting inclusion for the scale While the items were adapted to suit the research context, they maintain their original meaning by substituting phrases like "the chatbot" and "this activity."
“GenAI”; learning” with “self-learning” and making other minor changes, as presented in the following table:
Original Items Adjusted Items Translated Items Sources
PAI I did this activity because I wanted to.
I used GenAI for self-learning because I wanted to.
Tôi sử dụng Gen AI đế tự học vì tôi muốn như vậy.
PA2 I feel like 1 can 1 feel like I can Tôi cám thấy minh Hew and
3.3.2 Perceived Competence Scale make a lot of input in deciding how I use the chatbot in learning. make a lot of input in deciding how I use GenAI in self learning. có thể tìm hiểu nhiều nội dung khi tự học có sử dụng GenAI.
PA3 I have a say regarding what input I want to learn with chatbot.
1 have a say regarding what input I want to learn with Gen AI.
Tôi có quyền quyết định nội dung tôi muốn tìm hiểu với GenAĨ.
PA4 I have many opportunities with the chatbot to decide for myself how to learn.
I have many opportunities with GenAI to decide for myself how to self learn.
Tôi cỏ thê quyêt định cách tôi tự học khi sử dụng GenAI.
PA5 I feel a sense of freedom when using the chatbot.
I feel a sense of freedom when using GenAI for self-learning.
Tôi cám thấy tự do trong việc sử dụng GenAI để tự học.
The Perceived Competence Scale includes five items, labeled PCI to PC5, with items PCI, PC2, and PC5 derived from the research of Hew and Kadir (2016), while items PC3 and PC4 are based on the study by McAuley et al (1989) The authors modified these items to align with the specific research context, ensuring the original meaning was preserved by substituting the term "learning."
“self-learning”; “the chatbot” with “GenAI” and making other minor changes, as presented in the following table:
Original Items Adjusted Items Translate Items Sources
PCI I think I am pretty good at learning with the chatbot.
I think I am pretty good at self learning with GenAL
Tôi nghĩ tôi sử dụng tốt GenAI trong việc tự học.
PC2 I am pretty skillful at learning with the chatbot.
I am pretty skillful at self-learning with GenAI.
Tôi khá thành thạo trong việc sử dụng GenAl để tự học.
After working at this activity for a while, I felt pretty competent.
After learning with GenAI for a while, Ỉ felt pretty competent.
Sau một khoảng thời gian sứ dụng GenAl đê tự học thì tôi cám thấy tôi khá giói trong viộc này.
I am satisfied with my performance at this task.
I am satisfied with my GenAI using skills for self learning.
Tôi hài lòng với khá năng sử dụng GenAI để lự học của minh.
PC5 I have been able to learn interesting new knowledge with the chatbot.
I have been able to learn interesting new knowledge with GenAI.
Tôi có khả năng học những kiến thức mới bằng việc sử dụng GenAI.
The Perceived Relatedness Scale, developed by McAuley et al (1989), includes four items labeled PR1 to PR4 The authors modified these items to suit their research context while preserving the original intent by substituting "this person1" with relevant terms.
“GenAT1 and making other minor changes, as presented in the following table:
Original Items Adjusted Items Translate Items Sources
PRl I felt like I could really trust this person.
I felt like 1 could really trust GenAI.
Tôi cám thấy tôi có thê tin tưởng
PR2 I'd like a chance to interact with this person more often.
I'd like a chance to interact with GenAI more often.
Tôi muốn có nhiều cơ hội hơn đê tương tác với GenAI
PR3 I feel close to this person.
Tôi cám thấy gần gũi với Gen AI.
PR4 It is likely that this person and I could become friends if we interacted a lot.
It is likely that GenAI and I could become companions if we interacted a lot.
Tôi cám thấy tôi và GenAI có thê thành những người bạn đồng hành nếu lôi sử dụng nhiêu hơn.
The Intrinsic Motivation Scale, developed from the research of Dysvik and Kuvaas (2008), comprises four items labeled IM1 to IM4 The authors adapted these items to align with the specific research context while preserving their original intent by substituting "ChatGPT" with alternative terms.
“GenAI for self-learning11 and making other minor changes, as presented in the following table:
Original Items Adjusted Items Translate Items Sources
IMl I find using ChalGPT enjoyable.
I find using GenAI for self-learning enjoyable.
Tôi cảm thấy hứng thú với việc sử dụng GenAI đế tự học.
I had fun using GenAI for self learning.
Tôi thấy vui khi sử dụng GenAI đế tự học.
IM3 The actual process of using ChatGPT was pleasant.
The actual process of using GenAl for self-learning was pleasant.
Quá trinh sư dụng GenAI đỗ tự học làm cho tôi rât hài lòng.
IM4 Using ChatGPT to address my academic inquiries is interesting.
Using GenAI to address my academic inquiries is interesting.
Việc sử dụng GenAI đế giái đáp các khúc mắc trong học tập rất thú vị.
The Perceived Usefulness Scale includes five items, labeled PU1 through PU5 Items PU1 and PU4 are derived from the research conducted by McAuley et al (1989), while PU2, PU3, and PU5 originate from the study by Davis et al (1989) The authors modified these items to align with the specific research context, ensuring the original meaning was preserved by substituting "ChatGPT" where necessary.
“GcnAI"; “this activity" with “using Gen AI for self-learning” and making other minor changes, as presented in the following table:
Original Items Adjusted Items Translate Items Sources
PU1 I believe this activity could be of some value to me.
I believe using GenAI for self learning could be of some value to me.
Tôi tin rằng sử dụng GenAl de tự học sẽ mang lại lợi ích cho tôi.
PU2 I find ChatGPT useful for answering academic inquiries.
I find GenAI useful for answering academic inquiries.
Tôi cảm thấy GenAl hữu ích trong việc trâ lời các thắc mắc trong học tập cùa tôi.
PU3 Using ChatGPT addresses my academic inquiries more quickly.
Using GenAI addresses my academic inquiries more quickly.
Sừ dụng GenAI giúp giái đáp các thắc mắc trong học tập của tôi một cách nhanh chóng hơn.
PU4 I would be willing to do this again because it has some value to me.
I would be willing to use GenAI for self-learning because it has some value to me.
Tôi sằn sàng sử dụng GenAI trong việc tự học vì nó có ích đổi với tôi.
PU5 Using ChatGPT to address my academic inquiries would increase my academic performance.
Using Gen AI for self-learning would increase my academic performance.
Sừ dụng GenAI cho việc tự học sẽ giúp tôi nâng cao thành tích học lập.
Following qualitative research, the authors identified a lack of a suitable scale to measure self-learning motivation To address this gap, they adapted the Writing Motivation Scale by substituting "writing in English" with "self-learning" and implementing minor modifications The revised scale includes six items, labeled SLM1 to SLM6, drawing on the framework established by Waller and Papi (2017).
Table 3.6 Self-learning Motivation Scale
Original Items Adjusted Items Translate Items Sources
Tôi hửng thú với việc tự học hơn.
(2017) SLM2 Writing in English is very important to me.
Self-learning is very important to me.
Tự học dần quan trọng hơn đổi với tôi.
SLM3 I always look forward to my ESL writing classes.
I always look forward to my self-learning sessions.
Tôi trông chờ hơn vào các buổi tôi tự học.
SLM4 1 would like to spend lots of lime learning to write in English.
I would like to spend lots of time self-learning.
Tôi muốn dành nhiều thời gian hơn để tự học.
SLM5 1 would like to concentrate on learning to write in English more than any other topic.
I would like to concentrate on learning more than any other topic.
Tôi ưu liên việc tự học hơn các hoạt động khác
Questionnaire Design
SLM6 I actively think about what I have learned in my English writing class.
I actively think about what I have learned.
Tôi chù động hơn trong việc suy nghĩ về những gì tôi học được.
SLM7 1 really try to learn how to write in English.
I really try to learn how to self learn.
Tôi thực sự cố gắng tỉm cách làm thế nào để tự học hiệu quà.
The online survey questionnaire was specifically crafted to gather data for the quantitative analysis phase, targeting students in Ho Chi Minh City It comprised three distinct sections to ensure comprehensive data collection.
To ensure accurate survey results from the target population, two screening questions were implemented: "Which area are you currently studying in?" and "Have you ever utilized GenAI tools such as ChatGPT, Gemini, or Claude AI for learning purposes?" Only those respondents who indicated "Ho Chi Minh City" for the first question and answered "Yes" to the second were permitted to continue with the survey.
The article gathered demographic data, including gender, year of study, and the frequently used GenAI tool among respondents Participants were also given the option to share their email addresses for potential follow-up communication.
The main survey questions examined the influence of GenAI on self-learning motivation, focusing on factors such as Perceived Autonomy, Perceived Competence, Perceived Relatedness, Intrinsic Motivation, Perceived Usefulness, and Self-learning Motivation Respondents rated their agreement with various statements using a 5-point Likert scale, ranging from "1: Strongly disagree" to "5: Strongly agree." The questions were adapted from a qualitative study to ensure relevance and accuracy in measuring students' perceptions and motivations regarding self-directed learning with GenAI.
Data analysis method
After collecting 294 valid responses through Google Forms, the authors rechecked the data and coded it for further analysis and processing using SPSS software.
Descriptive statistics were used to analyze the variables of gender, year of study, and the most commonly used Gen AI tool.
3.5.2 Reliability Test of the Scale using Cronbach’s Alpha
This test was used to assess the reliability of the scale and to eliminate unobserved variables that did not meet the reliability criteria based on the following:
- Testing each group of observed variables for each factor.
- Cronbach's Alpha coefficient value (According to Hoang Trong and Chu Nguyen Mong Ngoc (2008):
• From 0.8 to close to 1: Very good measurement scale.
• From 0.7 to close to 0.8: Good measurement scale.
• From 0.6 and above: Acceptable measurement scale.
To ensure reliability in your scale, if the Cronbach's Alpha coefficient falls below 0.6, it is essential to remove variables that contribute to this low score Focus on eliminating the variable that, when deleted, results in the highest increase in the Cronbach's Alpha value Continue this process until the coefficient reaches the acceptable threshold of 0.6 or higher.
- Eliminate variables with Corrected Item - Total Correlation coefficient less than 0.3.
3.5.3 Evaluating the Scale using Exploratory Factor Analysis (EFA)
Exploratory Factor Analysis (EFA) investigates the relationships among variables across different observation groups to identify those that load onto multiple factors or are mis-factored This process streamlines the observed variables into a more coherent set of meaningful factors that exhibit stronger correlations Key metrics assessed in EFA include the convergence value and the discrimination value, which are essential for evaluating the analysis's effectiveness.
EFA must meet the following conditions:
- Kaiser-Meyer-Olkin (KMO) coefficient is used to consider the appropriateness of the factor 0.5 < KMO < 1.
- Bartlett’s test with Sig < 0.05 (With HO: variables are not correlated with each other in the population, Hl: variables are correlated with each other in the population).
The fixed number of factors method is utilized to determine the number of factors by actively enforcing a specific count In this research, the team has identified six factors to extract, aligning with the focus of their study.
The research team conducted Confirmatory Factor Analysis (CFA) to serve the following purposes:
Measuring unidimensionality is essential for validating models with market data, as noted by Hair et al (2010) A model's fit with observed variables indicates unidimensionality, provided that the errors of these variables are not correlated Key metrics for assessing this fit include the Chi-square statistic adjusted for degrees of freedom (CMIN/df), the Goodness of Fit Index (GFI), the Comparative Fit Index (CFI), the Tucker and Lewis Index (TLI), and the Root Mean Square Error of Approximation (RMSEA).
A model is deemed to align with market data when the Chi-square test shows a P-value greater than 0.05, the CMIN/df ratio is less than or equal to 2 (or in some cases, less than or equal to 3), and the goodness-of-fit indices such as GFI, CFI, and TLI exceed 0.9, while the RMSEA is less than or equal to 0.08.
According to recent views, GFI can still be acceptable when 0.05: This means there is no statistically significant difference in variance between the two groups of values We use the t-test result under the
Step 2: Test for mean differences between the two groups of values After assessing variance differences, we move on to evaluating mean differences The t-test is used to test differences In SPSS, the figures for the t-test are obtained from the "t-test for Equality of Means" section in the "Independent Samples Test" table The test results are as follows:
- Sig < 0.05: This means there is a statistically significant difference in means between the two groups of values.
- Sig > 0.05: This means there is no statistically significant difference in means between the two groups of values.
RESEARCH RESULTS
Descriptive statistics for collected data
Firstly, the sample size was 306, and 100% of the sample belonged to Ho Chi Minh City, indicating that the research space was ensured.
Figure 4.1 Percentage frequency of student survey sample in Ho Chi Minh City
In Ho Chi Minh City, a significant majority of individuals, comprising 294 people or 96.1%, have utilized Generative AI (GenAI) for self-learning, while only 12 people, representing 3.9%, have never engaged with these tools This stark contrast highlights the growing interest and widespread adoption of GenAI for educational purposes in the region.
Figure 4.2 Percentage frequency of survey sample that had used GenAI for learning
However, the research sample was students in Ho Chi Minh City who had used Gen AI for learning Therefore, the authors removed the survey samples that chose
The research identified 294 respondents, all students from Ho Chi Minh City, who had utilized GenAI for their learning experiences.
We entered 294 coded samples into SPSS software for analysis The statistical results are presented by the group through the following charts with specific information:
Figure 4.3 Percentage frequency by Gender of the research sample
The survey included 81 male respondents (27.6%) and 213 female respondents (72.4%), highlighting a significant gender disparity influenced by the survey location and distribution method It is important to note that this sample does not accurately represent the broader population Nonetheless, the data indicates that female students are more likely to utilize Generative AI for self-learning compared to their male counterparts.
Figure 4.4 Percentage frequency by Year of study of the research sample
Year of study: Among the 294 valid respondents, the largest number was 93
The survey revealed that Juniors comprised the largest group at 31.6%, followed by Freshmen at 25.2%, and Sophomores at 23.8% Seniors represented the smallest segment at 17.7%, while other students made up just 1.7% Conducted over a relatively short period, the survey achieved a balanced representation across different academic years, indicating that the sample effectively reflects the overall student population in Ho Chi Minh City.
■ ChatGPT ■ Gemini Bing Chat ■ Claude Al Bothers
Figure 4.5 Frequency of the most frequently used Gen AI tool The most frequently used GenAI tools: ChalGPT was selected by 269 out of
294 respondents, which is 2.2 times more than Gemini (121 votes), 2.7 times more than
Bing Chat (97 votes), and 11.2 times more than Claude AI (24 votes) There were also
Despite the emergence of numerous innovative GenAI tools like Notion, ChatGPT remains the most popular choice among students The lesser-known alternatives struggle to gain traction, likely due to limited media exposure and awareness.
Testing measurement scale
4.2.1 Results of Scale Reliability Testing Using Cronbach’s Alpha Coefficient
The results of the Cronbach's Alpha reliability test arc presented in Table 4.1 as follows:
Table 4.1 Cron bach’s Alpha Results Table
Perceived Autonomy scale (PA): Cronbach’s Alpha = 0.881
|PA1] I used GenAI for self-learning because 1 wanted to.
|PA2| I feel like 1 can make a lot of input in deciding how I use GenAI in self-learning.
[PA3] I have a say regarding what input
I want to learn with GenAI.
[PA4] I have many opportunities with
Gen AI to decide for myself how to self learn.
[PA5] I feel a sense of freedom when using GenAI for self-learning.
Perceived Competence scale (PC): Cronbach’s Alpha = 0.895
[PC 111 think I am pretty good al self learning with Gen AI.
[PC2] I am pretty skillful at self learning with GenAL
|PC3| After learning with GenAI for a while, I felt pretty competent.
|PC4| I am satisfied with my Gen AI using skills for self-learning.
|PC5| I have been able to learn interesting new knowledge with Gen AI.
Perceived Relatedness scale (PR): Cronbach’s Alpha = 0.890
[PR1] I felt like I could really trust
[PR2] I'd like a chance to interact with
[PR3] I feel close to Gen AI 0.815 0.837 Appropriate
[PR4] It is likely that GenAI and I could become companions if we interacted a lot.
Intrinsic Motivation scale (IM): Cronbach’s Alpha = 0.857
[IM1] I find using Gen AI for self learning enjoyable.
|IM2] I had fun using GenAI for self learning.
[IM3] The actual process of using
GenAI for self-learning was pleasant.
[IM4] Using GenAI to address my academic inquiries is interesting.
Perceived Usefulness scale (PU): Cronbach’s Alpha = 0.890
|PU 111 believe using GenAI for self learning could be of some value to me.
[PU211 find GenAI useful for answering academic inquiries.
[PU3] Using GenAI addresses my academic inquiries more quickly.
[PU4] I would be willing to use GenAI for self-learning because it has some value to me.
[PU5] Using Gen AI for self-learning would increase my academic performance.
Self-learning Motivation scale (SLM): Cronbach’s Alpha = 0.916
[SLMl ] I enjoy self-learning 0.736 0.904 Appropriate
[SLM2] Self-learning is very important to me.
[SLM3] I always look forward to my self-learning sessions.
[SLM4] I would like to spend lots of time self-learning.
[SLM5] I would like to concentrate on learning more than any other topic.
[SLM6] I actively think about what I have learned.
[SLM7] I really try to learn how to self learn.
Table 4.1 demonstrates that all scales satisfy the necessary criteria, with the lowest total-item correlation coefficient recorded at IM4 = 0.671, surpassing the minimum threshold of 0.3 Additionally, the Cronbach's Alpha coefficients for the scales are as follows: PA = 0.881, PC = 0.895, and PR = [insert value here].
= 0.890; IM = 0.857; PU = 0.890; SLM = 0.917, all of which surpass 0.7, indicating the scales are reliable measures.
4.2.2 Results of Validity Assessment of the Scale with Exploratory Factor Analysis (EFA)
An exploratory factor analysis was conducted on 30 observed variables using the Principal Axis Factoring method and Promax rotation The analysis yielded a Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy at 0.952, indicating excellent sampling adequacy, and Bartlett's Test of Sphericity showed a significance value of 0.000, confirming the suitability of the data for factor analysis The total variance explained was 64.704%, well above the 50% threshold While most observed variables met the convergence value of greater than 0.35, the variable IM1 ("I find using generative AI enjoyable.") failed to meet the discriminant value As a result, the team opted to exclude the IM1 variable and perform the exploratory factor analysis again.
In the second attempt, exploratory factor analysis (EFA) was performed on 29 observed variables, yielding a KMO measure of 0.950, well above the acceptable threshold of 0.5 Additionally, Bartlett's Test showed a significance value of 0.000, indicating strong statistical significance, and the total variance explained reached 64.799%, surpassing the 50% benchmark All 29 observed variables satisfied the criteria for both discriminant and convergent validity, allowing for their use in subsequent factor analysis The results of this EFA are summarized in the accompanying table.
Table 4.2 Factor Loading Matrix from Exploratory Factor Analysis
[PAI] I used GenAI for self-learning 0.531 because I wanted to.
[PA2] I feel like I can make a lot of input in deciding how I use GenAI in self-learning.
[PA3] I have a say regarding what input I want to learn with GenAI.
[PA4] I have many opportunities with
GenAI to decide for myself how to self-learn.
[PA5] I feel a sense of freedom when using GenAI for self-learning.
|PC111 think I am pretty good at self learning with GenAI.
[PC2J I am pretty skillful at self learning with GenAL
[PC3J After learning with GenAI for a while, I felt pretty competent.
[PC4] I am satisfied with my GenAI using skills for self-learning.
[PC5| I have been able to learn interesting new knowledge with
[PR1] I felt like I could really trust 0.775
[PR2] I’d like a chance to interact with 0.790
0.892 [PR3] I feel close to GenAl.
[PR4] It is likely that Gen AI and I could become companions if we interacted a lot.
|1M2| I had fun using GenAI for self learning.
[IM3] The actual process of using
Gen AI for self-learning was pleasant.
|IM4| Using Gen AI to address my academic inquiries is interesting.
[PU1] I believe using GenAI for self learning could be of some value to
[PU2] I find GenAI useful for answering academic inquiries.
[PU3J Using Gen AI addresses my academic inquiries more quickly.
[PU4] I would be willing to use
Gen AI for self-learning because it has
0.386 [PU5] Using GenAI for self-learning would increase my academic performance.
[SLM2] Self-learning is very important to me.
[SLM3] I always look forward to my self-learning sessions.
[SLM4] I would like to spend lots of lime self-learning.
[SLM5] I would like to concentrate on learning more than any other topic.
[SLM6] I actively think about what Ĩ have learned.
[SLM7] I really try to learn how to self-learn.
Table 4.3 Results of the KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy 0.950
Bartlett’s Test of Sphericity Approx Chi-Square 5995.601 df 406
Confirmatory Factor Analysis (CFA) used for the following purposes:
- Evaluating the overall suitability of data based on Model Fit indicators such as: Chi-square/df, GFI, CFI, TLI, RMSEA, PCLOSE
Evaluating the measured variables reveals the effectiveness of the constructs In Exploratory Factor Analysis (EFA), all measured variables are interconnected with potential variables Conversely, Confirmatory Factor Analysis (CFA) allows researchers to define the exact number of factors required and identify the specific measurement variables linked to each potential variable.
- Evaluation of convergent validity and discriminant validity
Note: PA = Perceived Autonomy; PC = Perceived Competence; PR = Perceived
Relatedness; IM = Intrinsic Motivation; PU = Perceived Usefulness; SLM = Self learning Motivation.
The CFA results after considering the correlation between the variables showed that the model had Chi-square/df = 1.849 < 3; GFI = 0.863 > 0.8; CFI = 0.947 > 0.9;
The study demonstrates a strong model fit, with a TLI of 0.941, exceeding the threshold of 0.9, and an RMSEA of 0.054, which is below the acceptable limit of 0.06 These indicators confirm that the critical model aligns well with market data, and all scales within the research model exhibit unidirectionality.
The weight (Âj) is notably high, with the lowest value recorded at ẢSLM = 0.734, and all p-values are statistically significant at 0.000 This indicates that the variables employed to assess the concepts demonstrate convergent validity.
Table 4.4 The correlation coefficients between the concepts
Note: PA = Perceived Autonomy; PC = Perceived Competence; PR = Perceived
Relatedness; IM = Intrinsic Motivation; PU = Perceived Usefulness; SLM = Self learning Motivation.
The P-value calculations for the correlation coefficients of each pair indicate that all values are below 0.05, demonstrating that the correlation coefficients for each conceptual pair significantly differ from 1 at a 95% confidence level Consequently, the concepts examined in the model exhibit strong discriminant validity.
Table 4.5 The Summary table of measurement testing results
Using a value of Composite Reliability (CR) is greater than 0.7(>0.7) and Variance extracted (VE) is greater than 60% (>60%) The results showed that the
The study's factors demonstrate strong reliability and validity, with "Perceived Autonomy" achieving a Composite Reliability of 0.882 and a Variance extracted of 60% "Perceived Competence" follows closely with a Composite Reliability of 0.895 and a Variance extracted of 63.1% The "Perceived Relatedness" factor shows a Composite Reliability of 0.891 and a Variance extracted of 67.3% Additionally, "Intrinsic Motivation" has a Composite Reliability of 0.826 and a Variance extracted of 61.3% The "Perceived Usefulness" factor records a Composite Reliability of 0.893 and a Variance extracted of 62.5%, while "Self-learning Motivation" achieves the highest Composite Reliability of 0.918 with a Variance extracted of 61.5% Overall, these results confirm that the scales used in the study are valid.
4.2.4 Structural Equation Modeling (SEM) Analysis
Figure 4.6 The results of structural equation modeling (SEM) analysis
Note: PA = Perceived Autonomy; PC = Perceived Competence; PR = Perceived
Relatedness; IM = Intrinsic Motivation; PU = Perceived Usefulness; SLM = Self learning Motivation.
The formal theoretical model consists of six key research concepts: Perceived Autonomy (PA), Perceived Competence (PC), Perceived Relatedness (PR), Intrinsic Motivation (IM), Perceived Usefulness (PU), and Self-learning Motivation (SLM) Within this model, three independent variables (PA, PC, PR) influence two mediating variables (IM, PU), ultimately affecting the dependent variable, Self-learning Motivation (SLM).
The theoretical model demonstrates 365 degrees of freedom, with test statistics revealing a Chi-squared value of 680.862, degrees of freedom at 65, and a p-value of 0.000 The CMIN/df ratio is 1.865, indicating compliance with compatibility requirements Additional indices also satisfy the necessary criteria, including CFI at 0.861, TFI at 0.940, and another CFI at 0.946 Following the validation of these criteria, a structural equation modeling (SEM) analysis was performed to evaluate the hypotheses.
Table 4.6 The results of hypothesis testing
Note: PA = Perceived Autonomy; PC = Perceived Competence; PR = Perceived
Relatedness; IM = Intrinsic Motivation; PU = Perceived Usefulness; SLM = Self learning Motivation.
At a 95% reliability level, the analysis reveals that "Perceived Competence" (PC) does not significantly affect "Perceived Usefulness," as indicated by a Sig value exceeding 0.05 In contrast, all other hypotheses meet the significance threshold of less than 0.05, confirming their acceptance This indicates that the causal variables outlined in self-determination theory influence the mediating variables, which in turn affect the dependent variable Furthermore, the accepted hypotheses exhibit positive relationships, suggesting that the factors are proportional to one another, particularly highlighting that "Perceived Usefulness" significantly enhances "Self-learning Motivation."
From the results of hypothesis testing, the hypotheses are Significance or Non Significance:
Table 4.7 Table summarizes the conclusion of hypotheses
Hla: Perceived Autonomy has a positive (+) impact on intrinsic Motivation when using GenAl.
Hlb: Perceived Competence has a positive (+) impact on
Intrinsic Motivation when using GenAI.
Hlc: Perceived Relatedness has a positive (+) impact on
Intrinsic Motivation when using GenAI.
H2a: Perceived Autonomy has a positive (+) impact on the
H2b; Perceived Competence has a positive (+) impact on
H2c: Perceived Relatedness has a positive (+ )impact on
H3: Intrinsic motivation has a positive (+) impact on
H4: Perceived Usefulness has a positive (+) impact on
H5: Intrinsic Motivation has a positive (+) impact on Self learning Motivation.
Testing differences in self-learning motivation by gender and school
4.3.1 Influence of gender on self-learning motivation
Independent Sample T-Test analysis was conducted to test the difference in students' self-learning motivation under the influence of GenAI between 2 factors: male gender and female gender.
Table 4.8 The Sig result of Levene-test and T-test
The results of the F test indicate a significance value of 0.070, which is greater than 0.05, suggesting that there is no significant difference in variance between male and female groups Additionally, the T test results, with a significance value of 0.433, also exceed 0.05, indicating no mean difference between genders Consequently, gender does not influence students' self-learning motivation.
4.3.2 Influence of the school year on self-learning motivation
A one-way ANOVA analysis was performed to examine the differences in self-learning motivation among students influenced by GenAI, categorized by academic year: freshman, sophomore, junior, and senior.
Table 4.9 The result of Test of Homogeneity of Variances
Test of Homogeneity of Variances
Levene Statistic dfl df2 Sig.
Table 4.10 The result of Robust Tests of Equality of Means
Robust Tests of Equality of Means
The Levene test results indicate a significant difference in variance between the groups, with a significance value of 0.000 (less than 0.05) However, the Robust Tests table shows that the Welch test has a significance coefficient of 0.208 (greater than 0.05), suggesting that being in a specific academic year does not significantly impact "self-learning motivation."
Discussion
This article examines key findings from the research, highlighting three main areas: the adjustments made to measurement scales, the development of research models and hypotheses, and the degree of influence among factors within the research model.
The measurement scales utilized in this study were primarily adapted from prior research to suit the specific context of the investigation Following thorough analysis and testing, only one item, "[IM 1] I feel interested in using GenAI for self-learning," was removed from the Intrinsic Motivation scale due to its low factor loading of less than 0.35 All other observed variables were found to be suitable and successfully met the scale testing criteria.
The research model was evaluated using Confirmatory Factor Analysis (CFA), revealing a satisfactory level of fit despite a lower GFI index of 0.861, which remains within acceptable limits The quantitative study successfully established the relationships among the factors in the model, resulting in the acceptance of eight out of nine proposed hypotheses.
The analysis reveals that in Self-Determination Theory, the "Perceived Competence" factor does not directly influence students' "Perceived Usefulness" of GenAI; rather, it is shaped by "Perceived Autonomy" and "Perceived Relatedness." For students to recognize the benefits of utilizing GenAI in self-learning, they need to feel a sense of autonomy in selecting the knowledge they wish to acquire Additionally, fostering a close and familiar interaction through GenAI is essential to build trust, encouraging students to freely express their learning interests without fear of judgment, even for basic concepts Therefore, for students to appreciate the usefulness of GenAI, they must have autonomy in its usage, independent of their inherent capabilities.
The findings indicate that intrinsic motivation is significantly affected by all three components of self-determination theory, with "Perceived Autonomy" being the most influential While "Perceived Usefulness" acts as an external motivator, it does not rely on "Perceived Competence." Therefore, fostering intrinsic motivation requires the addition of "Perceived Autonomy." To enhance intrinsic motivation, it is essential that GenAI not only provides satisfaction to students but also empowers them to effectively utilize GenAI tools to acquire knowledge that aligns with their personal interests.
Finally, "Self-learning Motivation" is simultaneously influenced by both
The study highlights that "Perceived Usefulness" significantly influences "Self-learning Motivation," with a stronger impact than "Intrinsic Motivation." Additionally, "Intrinsic Motivation" plays a role in enhancing "Perceived Usefulness." To effectively boost students' motivation for self-study supported by Generative AI, it is essential to foster intrinsic motivation, which serves as a foundation for enhancing self-learning motivation.
(3) The level of impact between the factors in the research model:
The Impact of the Three factors of Self-Determination on "Perceived Usefulness" and "Intrinsic Motivation ".
Among the three components of self-determination, "Perceived Autonomy" significantly influences self-study motivation, followed by "Perceived Competence," while "Perceived Relatedness" has the least impact Students are more motivated to self-study when they can control Generative AI according to their preferences and actively engage in their learning process To maximize their results, students must also acquire knowledge on effectively utilizing Generative AI The lower influence of "Perceived Relatedness" may stem from students' reduced need for human-machine interaction compared to traditional teaching methods.
The Impact of “Perceived Usefulness”, “Intrinsic Motivation ” and “Self learning Motivation
The study reveals that both mediating factors significantly influence Self Learning Motivation, with "Perception of Usefulness" being the most impactful It indicates that students' motivation for self-study increases when they recognize the benefits of GenAI in relation to their personal interests and desired content When students perceive immediate value from these benefits, they are more likely to sustain their self-study efforts Consequently, understanding the advantages of GenAI enhances their motivation, as these benefits resonate with their personal goals and aspirations.
"Intrinsic motivation" has a positive (+) effect on the "Perceived Usefulness" Therefore, promoting intrinsic motivation will serve as a prerequisite for further enhancing students' self-learning motivation.