Pub Date : 2025-04-11DOI: 10.1109/TLT.2025.3560032
Biao Gao;Jun Yan;Ronghui Zhong
Digital teachers represent an innovative fusion of media and artificial intelligence (AI) within online educational environments. However, the specific ways in which the appearance anthropomorphism of digital teachers influences the delivery of different knowledge types remain insufficiently understood. Drawing on Embodied Learning Theory and Parasocial Interaction Theory, this study investigates how digital teachers' appearance (cartoonish versus realistic) interacts with knowledge types (explicit versus tacit) to affect digital learning satisfaction and usage intention, exploring the mediating roles of physical and social presence. Initially, we implemented a 2 × 2 experimental design using a large language model application, collecting data from 475 participants to empirically test our hypotheses. Subsequently, in-depth interviews were conducted with 21 Chinese university students to further validate and clarify the underlying mechanisms behind these interactions. The results indicate that digital teachers with a cartoonish appearance are more effective for delivering explicit knowledge, whereas digital teachers with a realistic appearance excel in conveying tacit knowledge. Both physical presence and social presence were found to significantly mediate these effects. This research enriches our understanding of AI-enhanced online education by highlighting the alignment effect between digital teacher appearance and the type of knowledge delivered and by uncovering the underlying psychological mechanisms. In addition, it offers practical insights for the design of digital human appearances in educational interfaces and broader AI–human interaction scenarios.
{"title":"How Digital Teacher Appearance Anthropomorphism Impacts Digital Learning Satisfaction and Intention to Use: Interaction With Knowledge Type","authors":"Biao Gao;Jun Yan;Ronghui Zhong","doi":"10.1109/TLT.2025.3560032","DOIUrl":"https://doi.org/10.1109/TLT.2025.3560032","url":null,"abstract":"Digital teachers represent an innovative fusion of media and artificial intelligence (AI) within online educational environments. However, the specific ways in which the appearance anthropomorphism of digital teachers influences the delivery of different knowledge types remain insufficiently understood. Drawing on Embodied Learning Theory and Parasocial Interaction Theory, this study investigates how digital teachers' appearance (cartoonish versus realistic) interacts with knowledge types (explicit versus tacit) to affect digital learning satisfaction and usage intention, exploring the mediating roles of physical and social presence. Initially, we implemented a 2 × 2 experimental design using a large language model application, collecting data from 475 participants to empirically test our hypotheses. Subsequently, in-depth interviews were conducted with 21 Chinese university students to further validate and clarify the underlying mechanisms behind these interactions. The results indicate that digital teachers with a cartoonish appearance are more effective for delivering explicit knowledge, whereas digital teachers with a realistic appearance excel in conveying tacit knowledge. Both physical presence and social presence were found to significantly mediate these effects. This research enriches our understanding of AI-enhanced online education by highlighting the alignment effect between digital teacher appearance and the type of knowledge delivered and by uncovering the underlying psychological mechanisms. In addition, it offers practical insights for the design of digital human appearances in educational interfaces and broader AI–human interaction scenarios.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"438-457"},"PeriodicalIF":2.9,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888395","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-10DOI: 10.1109/TLT.2025.3559623
Juan A. Muñoz-Cristóbal;Vanesa Gallego-Lema;Higinio F. Arribas-Cubero;Gabriel Rodríguez-González;Felipe Hermida-Arias;Alejandra Martínez-Monés
Orienteering has long been used in physical education due to its recognized benefits for perceptual-motor capacity, as a tool for safe and efficient movement and as a recreational activity. It also helps in the acquisition of skills in multiple domains besides physical education, such as geography, mathematics, or biology. Many teachers use this interdisciplinary nature of orienteering, complementing it with educational tasks at each control point, and using geolocation and mobile devices to avoid the cumbersome tasks related to the setting up and dismantling of physical circuits. However, the systems that allow this kind of geolocated educational orienteering activities have some limitations in their implementation of the elements of orienteering or in the educational possibilities for teachers to configure and monitor learning situations that can adapt to their learning goals. To address these challenges, this article proposes a set of design requirements to create geolocated educational orienteering systems and a mobile tool, OrientaTree, created following the said requirements. A prototype of OrientaTree has been evaluated by means of a feature analysis and a pilot study involving five teachers and 115 students. The results of the evaluation provide evidence that OrientaTree overcomes the limitations of alternative reviewed approaches to conduct geolocated educational orienteering activities. However, it could be improved to allow more configuration capabilities to permit teachers to better adapt activities to their learning goals.
{"title":"OrientaTree: A Mobile Tool for Geolocated Educational Orienteering","authors":"Juan A. Muñoz-Cristóbal;Vanesa Gallego-Lema;Higinio F. Arribas-Cubero;Gabriel Rodríguez-González;Felipe Hermida-Arias;Alejandra Martínez-Monés","doi":"10.1109/TLT.2025.3559623","DOIUrl":"https://doi.org/10.1109/TLT.2025.3559623","url":null,"abstract":"Orienteering has long been used in physical education due to its recognized benefits for perceptual-motor capacity, as a tool for safe and efficient movement and as a recreational activity. It also helps in the acquisition of skills in multiple domains besides physical education, such as geography, mathematics, or biology. Many teachers use this interdisciplinary nature of orienteering, complementing it with educational tasks at each control point, and using geolocation and mobile devices to avoid the cumbersome tasks related to the setting up and dismantling of physical circuits. However, the systems that allow this kind of geolocated educational orienteering activities have some limitations in their implementation of the elements of orienteering or in the educational possibilities for teachers to configure and monitor learning situations that can adapt to their learning goals. To address these challenges, this article proposes a set of design requirements to create geolocated educational orienteering systems and a mobile tool, OrientaTree, created following the said requirements. A prototype of OrientaTree has been evaluated by means of a feature analysis and a pilot study involving five teachers and 115 students. The results of the evaluation provide evidence that OrientaTree overcomes the limitations of alternative reviewed approaches to conduct geolocated educational orienteering activities. However, it could be improved to allow more configuration capabilities to permit teachers to better adapt activities to their learning goals.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"484-497"},"PeriodicalIF":2.9,"publicationDate":"2025-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10960751","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143943942","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-03DOI: 10.1109/TLT.2025.3557037
Xiuling He;Ruijie Zhou;Qiong Fan;Xiong Xiao;Ying Yu;Zhonghua Yan
Rapid technological advancements are reshaping pedagogical expertise development, offering novel pathways to equip educators with 21st-century professional competencies. This study proposes an innovative artificial intelligence (AI)-driven professional development approach and investigates its impact on student teachers’ competence development. In total, 28 third-year student teachers participated in tasks to mentor AI learners, applying mentor-acquired knowledge and skills. Task performance and task processes were used to delineate teacher knowledge and teaching practices, respectively, while data from professional development surveys were thoroughly analyzed to gain in-depth insights into teacher perspectives. Findings reveal that AI teaching practice significantly enhanced participants’ knowledge acquisition. Notably, high-performance groups demonstrated complex mentoring patterns emphasizing procedural mentoring. Conversely, the low-performance group preferred a more directive and factual approach, whose behavioral patterns appeared less significant. Furthermore, AI teaching practice also had a positive effect on student teachers’ perspectives toward professional knowledge and AI literacy. The findings of this study contribute to the theoretical and practical understanding of integrating AI-based learning activities into teacher education.
{"title":"Preparing Student Teachers for Professional Development: Mentoring Generative Artificial Intelligence (AI) Learners in Mathematical Problem Solving","authors":"Xiuling He;Ruijie Zhou;Qiong Fan;Xiong Xiao;Ying Yu;Zhonghua Yan","doi":"10.1109/TLT.2025.3557037","DOIUrl":"https://doi.org/10.1109/TLT.2025.3557037","url":null,"abstract":"Rapid technological advancements are reshaping pedagogical expertise development, offering novel pathways to equip educators with 21st-century professional competencies. This study proposes an innovative artificial intelligence (AI)-driven professional development approach and investigates its impact on student teachers’ competence development. In total, 28 third-year student teachers participated in tasks to mentor AI learners, applying mentor-acquired knowledge and skills. Task performance and task processes were used to delineate teacher knowledge and teaching practices, respectively, while data from professional development surveys were thoroughly analyzed to gain in-depth insights into teacher perspectives. Findings reveal that AI teaching practice significantly enhanced participants’ knowledge acquisition. Notably, high-performance groups demonstrated complex mentoring patterns emphasizing procedural mentoring. Conversely, the low-performance group preferred a more directive and factual approach, whose behavioral patterns appeared less significant. Furthermore, AI teaching practice also had a positive effect on student teachers’ perspectives toward professional knowledge and AI literacy. The findings of this study contribute to the theoretical and practical understanding of integrating AI-based learning activities into teacher education.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"458-469"},"PeriodicalIF":2.9,"publicationDate":"2025-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143896316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-31DOI: 10.1109/TLT.2025.3556527
Jingjing Chen;Rao Muhammad Aqib Hassan;Shuai Sun;Yilin Mo;Dan Zhang
The lightboard, an affordable and readily accessible tool, has become a promising approach for enhancing engagement in instructional videos. Despite its potential, previous studies have primarily highlighted the benefits of lightboard videos by evaluating learners' subjective experiences, with limited empirical research examining their impact on learning outcomes. Moreover, the psychological factors underlying the potential advantages of lightboard videos have remained largely unexplored. To address these gaps, the present study conducted an online learning task in a mathematical optimization course, randomly assigning 78 college students to three groups: lightboard, whiteboard, and no-instructor. Learning outcomes and experiences during the learning process were measured and analyzed. The results showed that the lightboard group experienced significantly lower cognitive load while achieving learning outcomes comparable to the other two groups, suggesting that lightboard videos can reduce students' cognitive load without compromising learning outcomes. Further analysis of the psychological factors revealed that cognitive load played a more critical role than perceived social presence or learning motivation in explaining learning outcomes. These findings underscore the positive impact of lightboard videos on online learning, provide insights into the underlying psychological mechanisms, and offer implications for their integration into educational practices.
{"title":"Evaluating the Impact of Lightboard Videos on College Students' Performance in a Mathematical Optimization Course","authors":"Jingjing Chen;Rao Muhammad Aqib Hassan;Shuai Sun;Yilin Mo;Dan Zhang","doi":"10.1109/TLT.2025.3556527","DOIUrl":"https://doi.org/10.1109/TLT.2025.3556527","url":null,"abstract":"The lightboard, an affordable and readily accessible tool, has become a promising approach for enhancing engagement in instructional videos. Despite its potential, previous studies have primarily highlighted the benefits of lightboard videos by evaluating learners' subjective experiences, with limited empirical research examining their impact on learning outcomes. Moreover, the psychological factors underlying the potential advantages of lightboard videos have remained largely unexplored. To address these gaps, the present study conducted an online learning task in a mathematical optimization course, randomly assigning 78 college students to three groups: lightboard, whiteboard, and no-instructor. Learning outcomes and experiences during the learning process were measured and analyzed. The results showed that the lightboard group experienced significantly lower cognitive load while achieving learning outcomes comparable to the other two groups, suggesting that lightboard videos can reduce students' cognitive load without compromising learning outcomes. Further analysis of the psychological factors revealed that cognitive load played a more critical role than perceived social presence or learning motivation in explaining learning outcomes. These findings underscore the positive impact of lightboard videos on online learning, provide insights into the underlying psychological mechanisms, and offer implications for their integration into educational practices.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"428-437"},"PeriodicalIF":2.9,"publicationDate":"2025-03-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888444","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This article addresses the growing importance of understanding how multimodal artificial general intelligence (AGI) can be integrated into educational practices. We first reviewed the theoretical foundations of multimodality in human learning, encompassing its concept and history, dual coding theory and multimedia theory, VARK multimodality, and multimodal assessment (see Section II-A). After that, we revisited the essential components of AGI, particularly focusing on the multimodal nature of AGI that distinguished it from artificial narrow intelligence. Based on its conversational functionality, multimodal AGI is considered an educational agent already tested in various educational situations (see Section II-B). How significant text, image, audio, and video modalities are for education, the technological backgrounds of AGI for analyzing and generating them, and educational applications of artificial intelligence (AI) for each modality were thoroughly reviewed (Sections III–VI). Finally, we comprehensively investigated the ethics of AGI in education, originating from the ethics of AI and specified in three strands: first, data privacy and ethical integrity, second, explainability, transparency, and fairness, and third, responsibility and decision-making. Practical implementation of ethical AGI frameworks in education was reviewed (see Section VII). This article also discusses the implications for learning theories, derived operational design principles, current research gaps, practical constraints and institutional readiness, and future directions (see Section VIII). This exploration aims to provide an advanced understanding of the intersection between AI, multimodality, and education, setting a foundation for future research and development.
{"title":"Multimodality of AI for Education: Toward Artificial General Intelligence","authors":"Gyeonggeon Lee;Lehong Shi;Ehsan Latif;Yizhu Gao;Arne Bewersdorff;Matthew Nyaaba;Shuchen Guo;Zhengliang Liu;Gengchen Mai;Tianming Liu;Xiaoming Zhai","doi":"10.1109/TLT.2025.3574466","DOIUrl":"https://doi.org/10.1109/TLT.2025.3574466","url":null,"abstract":"This article addresses the growing importance of understanding how multimodal artificial general intelligence (AGI) can be integrated into educational practices. We first reviewed the theoretical foundations of multimodality in human learning, encompassing its concept and history, dual coding theory and multimedia theory, VARK multimodality, and multimodal assessment (see Section II-A). After that, we revisited the essential components of AGI, particularly focusing on the multimodal nature of AGI that distinguished it from artificial narrow intelligence. Based on its conversational functionality, multimodal AGI is considered an educational agent already tested in various educational situations (see Section II-B). How significant text, image, audio, and video modalities are for education, the technological backgrounds of AGI for analyzing and generating them, and educational applications of artificial intelligence (AI) for each modality were thoroughly reviewed (Sections III–VI). Finally, we comprehensively investigated the ethics of AGI in education, originating from the ethics of AI and specified in three strands: first, data privacy and ethical integrity, second, explainability, transparency, and fairness, and third, responsibility and decision-making. Practical implementation of ethical AGI frameworks in education was reviewed (see Section VII). This article also discusses the implications for learning theories, derived operational design principles, current research gaps, practical constraints and institutional readiness, and future directions (see Section VIII). This exploration aims to provide an advanced understanding of the intersection between AI, multimodality, and education, setting a foundation for future research and development.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"666-683"},"PeriodicalIF":2.9,"publicationDate":"2025-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144581641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-26DOI: 10.1109/TLT.2025.3570775
Shiqi Liu;Sannyuya Liu;Lele Sha;Zijie Zeng;Dragan Gašević;Zhi Liu
Automated classification of learner-generated text to identify behavior, emotion, and cognition indicators, collectively known as learning engagement classification (LEC), has received considerable attention in fields such as natural language processing(NLP), learning analytics, and educational data mining. Recently, large language models (LLMs), such as ChatGPT, which are considered promising technologies for artificial general intelligence, have demonstrated remarkable performance in various NLP tasks. However, their capabilities in LEC tasks still lack comprehensive evaluation and improvement approaches. This study introduces a novel benchmark for LEC, encompassing six datasets that cover behavior classification (question and urgency level), emotion classification (binary and epistemic emotion), and cognition classification (opinion and cognitive presence). In addition, we propose the annotation guideline-based knowledge augmentation (AGKA) approach, which leverages GPT-4.0 to recognize and extract label definitions from annotation guidelines and applies random undersampling to select a representative set of examples. Experimental results demonstrate the following: AGKA enhances LLM performance compared to vanilla prompts, particularly for GPT-4.0 and Llama-3 70B; GPT-4.0 and Llama-3 70B with AGKA are comparable to fully fine-tuned models such as BERT and RoBERTa on simple binary classification tasks; for multiclass tasks requiring complex semantic understanding, GPT-4.0 and Llama-3 70B outperform the fine-tuned models in the few-shot setting but fall short of the fully fine-tuned models; Llama-3 70B with AGKA shows comparable performance to GPT-4.0, demonstrating the viability of these open-source alternatives; and the ablation study highlights the importance of customizing and evaluating knowledge augmentation strategies for each specific LLM architecture and task.
{"title":"Annotation Guideline-Based Knowledge Augmentation: Toward Enhancing Large Language Models for Educational Text Classification","authors":"Shiqi Liu;Sannyuya Liu;Lele Sha;Zijie Zeng;Dragan Gašević;Zhi Liu","doi":"10.1109/TLT.2025.3570775","DOIUrl":"https://doi.org/10.1109/TLT.2025.3570775","url":null,"abstract":"Automated classification of learner-generated text to identify behavior, emotion, and cognition indicators, collectively known as learning engagement classification (LEC), has received considerable attention in fields such as natural language processing(NLP), learning analytics, and educational data mining. Recently, large language models (LLMs), such as ChatGPT, which are considered promising technologies for artificial general intelligence, have demonstrated remarkable performance in various NLP tasks. However, their capabilities in LEC tasks still lack comprehensive evaluation and improvement approaches. This study introduces a novel benchmark for LEC, encompassing six datasets that cover behavior classification (question and urgency level), emotion classification (binary and epistemic emotion), and cognition classification (opinion and cognitive presence). In addition, we propose the annotation guideline-based knowledge augmentation (AGKA) approach, which leverages GPT-4.0 to recognize and extract label definitions from annotation guidelines and applies random undersampling to select a representative set of examples. Experimental results demonstrate the following: AGKA enhances LLM performance compared to vanilla prompts, particularly for GPT-4.0 and Llama-3 70B; GPT-4.0 and Llama-3 70B with AGKA are comparable to fully fine-tuned models such as BERT and RoBERTa on simple binary classification tasks; for multiclass tasks requiring complex semantic understanding, GPT-4.0 and Llama-3 70B outperform the fine-tuned models in the few-shot setting but fall short of the fully fine-tuned models; Llama-3 70B with AGKA shows comparable performance to GPT-4.0, demonstrating the viability of these open-source alternatives; and the ablation study highlights the importance of customizing and evaluating knowledge augmentation strategies for each specific LLM architecture and task.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"619-634"},"PeriodicalIF":2.9,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144272741","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The advent of generative artificial intelligence (GAI), exemplified by ChatGPT, has introduced both new opportunities and challenges in science, technology, engineering, and mathematics (STEM) and entrepreneurship education. This exploratory quasi-experimental study examined the effects of ChatGPT-assisted collaborative learning (CCL) on students’ learning performance, artificial intelligence (AI) awareness, critical thinking, and cognitive load. A total of 36 sophomore undergraduates participated in an eight-week instructional experiment, dedicating 3 h per week to applying STEM and entrepreneurship knowledge in the creation of cultural products. The experimental group (N = 21) participated in CCL, while the control group (N = 15) engaged in non-ChatGPT-assisted collaborative learning (NCCL). The results indicated that the CCL group outperformed the NCCL group in terms of learning performance, AI awareness, and cognitive load, while the NCCL group excelled in critical thinking. The findings confirm that ChatGPT offers significant potential and advantages in addressing complex problems within group collaboration and stimulating group creativity, providing new insights into fostering students’ entrepreneurial spirit and skills. However, overreliance on and misuse of ChatGPT may hinder student learning outcomes. Future research should focus on the cocreative problem-solving mechanisms between humans and machines in entrepreneurial education, particularly the interplay of knowledge, thinking, emotions, and actions in collaborative processes involving GAI.
{"title":"Human–Machine Cocreation: The Effects of ChatGPT on Students’ Learning Performance, AI Awareness, Critical Thinking, and Cognitive Load in a STEM Course Toward Entrepreneurship","authors":"Yu Ji;Zehui Zhan;Tingting Li;Xuanxuan Zou;Siyuan Lyu","doi":"10.1109/TLT.2025.3554584","DOIUrl":"https://doi.org/10.1109/TLT.2025.3554584","url":null,"abstract":"The advent of generative artificial intelligence (GAI), exemplified by ChatGPT, has introduced both new opportunities and challenges in science, technology, engineering, and mathematics (STEM) and entrepreneurship education. This exploratory quasi-experimental study examined the effects of ChatGPT-assisted collaborative learning (CCL) on students’ learning performance, artificial intelligence (AI) awareness, critical thinking, and cognitive load. A total of 36 sophomore undergraduates participated in an eight-week instructional experiment, dedicating 3 h per week to applying STEM and entrepreneurship knowledge in the creation of cultural products. The experimental group (<italic>N</i> = 21) participated in CCL, while the control group (<italic>N</i> = 15) engaged in non-ChatGPT-assisted collaborative learning (NCCL). The results indicated that the CCL group outperformed the NCCL group in terms of learning performance, AI awareness, and cognitive load, while the NCCL group excelled in critical thinking. The findings confirm that ChatGPT offers significant potential and advantages in addressing complex problems within group collaboration and stimulating group creativity, providing new insights into fostering students’ entrepreneurial spirit and skills. However, overreliance on and misuse of ChatGPT may hinder student learning outcomes. Future research should focus on the cocreative problem-solving mechanisms between humans and machines in entrepreneurial education, particularly the interplay of knowledge, thinking, emotions, and actions in collaborative processes involving GAI.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"402-415"},"PeriodicalIF":2.9,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143877647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-24DOI: 10.1109/TLT.2025.3554174
Jialun Pan;Zhanzhan Zhao;Dongkun Han
Properly predicting students'academic performance is crucial for elevating educational outcomes in various disciplines. Through precise performance prediction, schools can quickly pinpoint students facing challenges and provide customized educational materials suited to their specific learning needs. The reliance on teachers' experience to predict students' academic performance has proven to be less accurate and efficient than desired. Consequently, the past decade has witnessed a marked surge in employing machine learning and data mining techniques to forecast students' performance. However, the academic community has yet to agree on the most effective algorithm for predicting academic outcomes. Nonetheless, conducting an analysis and comparison of the existing algorithms in this field remains meaningful. Furthermore, recommendations for selecting an appropriate algorithm will be provided to interested researchers and educators based on their specific requirements. This article reviews the state-of-the-art literature on academic performance predictions using machine learning approaches in recent years. It details the variables analyzed, the algorithms implemented, the datasets utilized, and the evaluation metrics applied to assess model efficacy. What makes this work different is that relevant surveys in the past 10 years are also analyzed and compared, highlighting their contributions and review methods. In addition, we compared the accuracy of various machine learning models using popular open-access datasets and determined the best-performing algorithms among them. Our dataset and source codes are released for future algorithm comparisons and evaluations in this community.
{"title":"Academic Performance Prediction Using Machine Learning Approaches: A Survey","authors":"Jialun Pan;Zhanzhan Zhao;Dongkun Han","doi":"10.1109/TLT.2025.3554174","DOIUrl":"https://doi.org/10.1109/TLT.2025.3554174","url":null,"abstract":"Properly predicting students'academic performance is crucial for elevating educational outcomes in various disciplines. Through precise performance prediction, schools can quickly pinpoint students facing challenges and provide customized educational materials suited to their specific learning needs. The reliance on teachers' experience to predict students' academic performance has proven to be less accurate and efficient than desired. Consequently, the past decade has witnessed a marked surge in employing machine learning and data mining techniques to forecast students' performance. However, the academic community has yet to agree on the most effective algorithm for predicting academic outcomes. Nonetheless, conducting an analysis and comparison of the existing algorithms in this field remains meaningful. Furthermore, recommendations for selecting an appropriate algorithm will be provided to interested researchers and educators based on their specific requirements. This article reviews the state-of-the-art literature on academic performance predictions using machine learning approaches in recent years. It details the variables analyzed, the algorithms implemented, the datasets utilized, and the evaluation metrics applied to assess model efficacy. What makes this work different is that relevant surveys in the past 10 years are also analyzed and compared, highlighting their contributions and review methods. In addition, we compared the accuracy of various machine learning models using popular open-access datasets and determined the best-performing algorithms among them. Our dataset and source codes are released for future algorithm comparisons and evaluations in this community.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"351-368"},"PeriodicalIF":2.9,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10938259","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143856220","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-21DOI: 10.1109/TLT.2025.3572175
Simone Porcu;Alessandro Floris;Luigi Atzori
In this article, we preliminarily discuss the limitations of current video conferencing platforms in online synchronous learning. Research has shown that while the involved technologies are appropriate for collaborative video calls, they often fail to replicate the rich nature of face-to-face interactions among students and between students and professors, by constraining them to a grid of faces on screens and limiting the natural flows of conversation and nonverbal communication. We believe that a potential solution to this issue could be adopting virtual reality (VR) technologies in online synchronous teaching. To test our assumption, we developed a novel subjective assessment involving 44 electronics engineering students who attended real lessons on Internet protocols. The taught content was included in the course program and the final exam; the professor made use of slides for teaching and a blackboard to explain some exercises. Two different learning approaches were used: VR-based online synchronous learning and video-based online synchronous learning. While the former consisted in wearing a headset and participating in a virtual classroom in front of the teacher’s avatar, the latter involved watching a 2-D video of the streamed lesson through a laptop and communicating through the microphone. The opinions collected from the students included several aspects, namely, overall quality of experience, immersion, interactivity, naturalness, usability, entertainment, comfort, side effects, interaction with the teacher and students, and ease of taking notes. Key findings from Welch’s $t$-test indicate the higher interactivity ($p< 0.05$), naturalness ($p< 0.01$), entertainment ($p< 0.01$), and immersion ($p< 0.001$) perceived by students for the VR-based learning experience than the video-based one. Increased immersion was the most significant aspect, as highlighted by the lowest $p$-value. On the other hand, the level of comfort was heavily penalized ($p< 0.001$), and students were unable to take notes in the VR classroom environment easily. No significant difference ($p>0.05$) was achieved for the other considered metrics.
{"title":"Will Virtual Reality Transform Online Synchronous Learning? Evidence From a Quality of Experience Subjective Assessment","authors":"Simone Porcu;Alessandro Floris;Luigi Atzori","doi":"10.1109/TLT.2025.3572175","DOIUrl":"https://doi.org/10.1109/TLT.2025.3572175","url":null,"abstract":"In this article, we preliminarily discuss the limitations of current video conferencing platforms in online synchronous learning. Research has shown that while the involved technologies are appropriate for collaborative video calls, they often fail to replicate the rich nature of face-to-face interactions among students and between students and professors, by constraining them to a grid of faces on screens and limiting the natural flows of conversation and nonverbal communication. We believe that a potential solution to this issue could be adopting virtual reality (VR) technologies in online synchronous teaching. To test our assumption, we developed a novel subjective assessment involving 44 electronics engineering students who attended real lessons on Internet protocols. The taught content was included in the course program and the final exam; the professor made use of slides for teaching and a blackboard to explain some exercises. Two different learning approaches were used: VR-based online synchronous learning and video-based online synchronous learning. While the former consisted in wearing a headset and participating in a virtual classroom in front of the teacher’s avatar, the latter involved watching a 2-D video of the streamed lesson through a laptop and communicating through the microphone. The opinions collected from the students included several aspects, namely, overall quality of experience, immersion, interactivity, naturalness, usability, entertainment, comfort, side effects, interaction with the teacher and students, and ease of taking notes. Key findings from Welch’s <inline-formula><tex-math>$t$</tex-math></inline-formula>-test indicate the higher interactivity (<inline-formula><tex-math>$p< 0.05$</tex-math></inline-formula>), naturalness (<inline-formula><tex-math>$p< 0.01$</tex-math></inline-formula>), entertainment (<inline-formula><tex-math>$p< 0.01$</tex-math></inline-formula>), and immersion (<inline-formula><tex-math>$p< 0.001$</tex-math></inline-formula>) perceived by students for the VR-based learning experience than the video-based one. Increased immersion was the most significant aspect, as highlighted by the lowest <inline-formula><tex-math>$p$</tex-math></inline-formula>-value. On the other hand, the level of comfort was heavily penalized (<inline-formula><tex-math>$p< 0.001$</tex-math></inline-formula>), and students were unable to take notes in the VR classroom environment easily. No significant difference (<inline-formula><tex-math>$p>0.05$</tex-math></inline-formula>) was achieved for the other considered metrics.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"606-618"},"PeriodicalIF":2.9,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144232058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The integration of generative artificial intelligence (GAI) into educational settings offers unprecedented opportunities to enhance the efficiency of teaching and the effectiveness of learning, particularly within online platforms. This study evaluates the development and application of a customized GAI-powered teaching assistant, trained specifically to enhance teaching efficiency for educators and improve learning outcomes for students in online education. Using four Grade 12 courses (i.e., English, Mathematics, Financial Accounting, and Simplified Chinese), we assessed the performance of generative pretrained transformer (GPT)-4, GPT-4o, and the Trained-GPT model. Results demonstrate that the Trained-GPT achieved grading accuracy and consistency comparable to human teachers, with strong correlations observed in Mathematics (0.996) and English (0.874). While GPT-4o performed well in specific cases, its variability highlights areas for improvement. These findings underscore the potential of AI-powered teaching assistants to streamline grading, deliver timely feedback, and support scalable, high-quality online education.
{"title":"GAI Versus Teacher Scoring: Which is Better for Assessing Student Performance?","authors":"Xuefan Li;Marco Zappatore;Tingsong Li;Weiwei Zhang;Sining Tao;Xiaoqing Wei;Xiaoxu Zhou;Naiqing Guan;Anny Chan","doi":"10.1109/TLT.2025.3572518","DOIUrl":"https://doi.org/10.1109/TLT.2025.3572518","url":null,"abstract":"The integration of generative artificial intelligence (GAI) into educational settings offers unprecedented opportunities to enhance the efficiency of teaching and the effectiveness of learning, particularly within online platforms. This study evaluates the development and application of a customized GAI-powered teaching assistant, trained specifically to enhance teaching efficiency for educators and improve learning outcomes for students in online education. Using four Grade 12 courses (i.e., English, Mathematics, Financial Accounting, and Simplified Chinese), we assessed the performance of generative pretrained transformer (GPT)-4, GPT-4o, and the Trained-GPT model. Results demonstrate that the Trained-GPT achieved grading accuracy and consistency comparable to human teachers, with strong correlations observed in Mathematics (0.996) and English (0.874). While GPT-4o performed well in specific cases, its variability highlights areas for improvement. These findings underscore the potential of AI-powered teaching assistants to streamline grading, deliver timely feedback, and support scalable, high-quality online education.","PeriodicalId":49191,"journal":{"name":"IEEE Transactions on Learning Technologies","volume":"18 ","pages":"569-580"},"PeriodicalIF":2.9,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144206123","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：