Computers and Education Artificial Intelligence最新文献

{"title":"Towards contextual-based AI: A scoping review of artificial intelligence in X reality for personalized learning","authors":"Zifeng Liu ,&nbsp;Serene Cheon ,&nbsp;Austin Stanbury ,&nbsp;Xinyue Jiao ,&nbsp;Wanli Xing ,&nbsp;Hyo Kang","doi":"10.1016/j.caeai.2025.100523","DOIUrl":"10.1016/j.caeai.2025.100523","url":null,"abstract":"<div><div>This systematic review synthesizes 54 peer-reviewed studies published between 2019 and 2025 that examine how artificial intelligence (AI) and extended reality (XR) technologies are integrated to support adaptive and personalized learning. The studies were analyzed across multiple dimensions, including learning contexts, AI applications, adaptive input parameters, software and hardware used, and evaluation methods. The findings indicate growing research interest in AI–XR integration, with the majority of studies focused on procedural training and STEM education. Across these studies, AI is frequently used in multifaceted roles, most notably as a provider of real-time adaptive feedback, conversational agent, and a generator of instructional content. Despite these promising developments, the review identifies several critical limitations. While generative AI, particularly large language models (LLMs) such as GPT, has been widely used for conversational interactions, learner profile data remains largely underutilized. Inputs such as prior knowledge and motivation are rarely incorporated. Most implementations rely on a single adaptive strategy, typically driven by performance-based measures such as pre-quiz scores or task completion. As a result, they do not fully exploit the multimodal sensing capabilities of XR platforms (e.g., eye tracking, gesture recognition, environmental tracking), which could support context-sensitive, dynamically generated 3D content aligned with when, where, and how learners need support. Current evaluations of AI–XR systems also remain dominated by short-term performance outcomes, with limited attention to knowledge transfer and critical thinking. These findings highlight key opportunities for designing context-aware, learner-centered AI–XR systems and call for future research that more fully leverages multimodal data, incorporates richer learner profile information, and is grounded in explicit pedagogical models.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100523"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926739","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial intelligence literacy at school: A systematic review with a focus on psychological foundations","authors":"Shuyan Feng,&nbsp;Astrid Carolus","doi":"10.1016/j.caeai.2026.100551","DOIUrl":"10.1016/j.caeai.2026.100551","url":null,"abstract":"<div><div>Artificial Intelligence (AI) is significantly changing school education. The increasing prevalence of AI calls for a framework of AI-related literacy specifically tailored to the educational context. A growing body of research has attempted to conceptualise AI literacy (AIL) from different disciplinary perspectives and with different foci. This systematic review aims to provide a comprehensive overview of definitions and psychological dimensions of AIL in school education by addressing the following questions: how is AIL defined and conceptualised, what are the dimensions of AIL, and what psychological dimensions are included. A total of 2642 records were identified from various databases, and 58 peer-reviewed articles were retrieved for this systematic review, which strictly followed the PRISMA guidelines. The findings propose different definitions of AIL for teachers, students and other educational professionals, and identifies dimensions that include cognitive, emotional, psychological, and behavioural constructs. More detailed, the review identifies six dimensions for teachers, such as contextual knowledge and continuous professional growth. For students, eight dimensions were identified, including AI-related thinking capacity and preparation for AI careers. Certain dimensions, such as AI knowledge and skills, AI ethics and societal implications, generative AI-specific competency, and most importantly, the psychological dimension consisting of cognitive and non-cognitive elements, were found to be shared across all target groups. Furthermore, personalisation and contextual adaptability emerged as additional key dimensions. In sum, the findings offer valuable insights for future research and practical guidance for decision-making in AI education, particularly in the areas of curriculum design, implementation, and assessment.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100551"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022839","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generative AI in higher education: A bibliometric review of emerging trends, power dynamics, and global research landscapes","authors":"Kun Dai ,&nbsp;Yabing Liu ,&nbsp;Xiaofan Zhang","doi":"10.1016/j.caeai.2026.100544","DOIUrl":"10.1016/j.caeai.2026.100544","url":null,"abstract":"<div><div>The rapid evolution of Generative Artificial Intelligence (GenAI) is reshaping higher education (HE), offering transformative opportunities for academic engagement while posing significant challenges to academic integrity, ethical frameworks, and global research power dynamics. This study maps the recent (2022–2025) research landscape of GenAI in HE through a bibliometric analysis of 2762 articles from the Web of Science Core Collection. Employing multipolarity as an analytical lens, this study examines the power dynamics within this research domain reflected by publication records from different countries (or regions). Findings highlight surging global interest in GenAI in HE, with contributions led by the US, China, and the UK, alongside rising participation from non-Western scholars and institutions. By identifying the major topics, this study uncovers a more nuanced trajectory of GenAI-related discourse in HE. By examining publication status, contributors, and research topics, this study provides insights for stakeholders navigating the complexities of GenAI integration into HE and suggests trajectories for future research in this rapidly evolving field.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100544"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022837","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing automated scoring in ILSAs with prompt compression","authors":"Ji Yoon Jung,&nbsp;Ummugul Bezirhan,&nbsp;Matthias von Davier","doi":"10.1016/j.caeai.2026.100558","DOIUrl":"10.1016/j.caeai.2026.100558","url":null,"abstract":"<div><div>Automated scoring (AS) has become increasingly prevalent in educational measurement. However, applying it to international reading assessments remains challenging, particularly due to the length and complexity of the required prompting, driven by the need to include lengthy reading passages and detailed scoring guides. Processing these lengthy inputs results in high computational costs and may impede the performance of large language models (LLMs). This study explored the potential of optimizing AS with prompt compression using OpenAI's LLM, GPT-4o. Our results show that prompt compression significantly reduces the length of reading passages and scoring guides while maintaining their essential content. Reading passages and scoring guides were compressed to approximately 18% and 15% of their original lengths, respectively. Despite this substantial compression, the AS showed remarkable performance, with an accuracy of 92.87% and a kappa score of 0.8041, closely approximating the results obtained without compression. These findings suggest optimizing AS with prompt compression can improve its efficiency and scalability, particularly in international reading assessments.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100558"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147384877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Pedagogy first, technology second: Cross-level relationships between teacher professional knowledge and student learning in artificial intelligence (AI) education","authors":"Weipeng Shen ,&nbsp;Ching-sing Chai ,&nbsp;Thomas K.F. Chiu ,&nbsp;King Woon Yau ,&nbsp;Helen Meng ,&nbsp;Irwin King ,&nbsp;Savio Wong ,&nbsp;Yeung Yam","doi":"10.1016/j.caeai.2026.100564","DOIUrl":"10.1016/j.caeai.2026.100564","url":null,"abstract":"<div><div>For current K-12 AI education, understanding how teacher professional knowledge influences student learning is essential. This study employed social cognitive theory to conceptualize teachers' instructional practice and conducted a multilevel structural equation model analysis to examine the cross-level effects of teacher AI knowledge (TAIK) and pedagogical AI knowledge (TPAIK) on student AI knowledge (SAIK), perceptions of AI for social good (SAISG), and behavioral intention to learn AI (SBIAI). Data from 46 secondary school teachers and 2832 students revealed intricate mechanisms. TAIK alone was insufficient and could slightly diminish SAISG without TPAIK. Instead, TPAIK played a fundamental role in fostering SAISG and SBIAI. Neither TAIK nor TPAIK was directly associated with SAIK. Our findings distinguished the roles of TAIK and TPAIK, which demonstrated the “Pedagogy first, technology second” guideline: TPAIK should be prioritized over TAIK for teachers' and students’ excellence in K-12 AI education.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100564"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147384882","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"How to train your dragon: Evaluating prompting and fine-tuning for GPT-based item generation in L2 listening assessment","authors":"Vahid Aryadoust,&nbsp;Joann Wong","doi":"10.1016/j.caeai.2026.100623","DOIUrl":"10.1016/j.caeai.2026.100623","url":null,"abstract":"<div><div>Recent advances in large language models (LLMs), notably GPT models, have introduced new possibilities for automatic item generation (AIG) through prompt engineering. Although iterative prompt refinement can yield measurable improvements, this process eventually plateaus, as outputs remain inconsistent or misaligned with assessment constructs. This study reports an experimental comparison of prompting and fine-tuning to advance AIG for L2 listening assessment. First, we employed prompting and refined the instruction design over three successive iterations to determine an optimised prompt. We then fine-tuned GPT-4.1 using the same optimised prompt, holding prompt design constant to isolate the effect of model adaptation. We generated a total of 40 tests and 240 multiple-choice items in the four model conditions: three prompt-only iterations and one fine-tuned iteration. To contextualize model performance against professional assessment standards, we also compared the GPT-generated items with 245 human-generated listening items used as the fine-tuning dataset. The generated items were evaluated using a hybrid two-tiered framework that combined rule-based metrics with human review to assess content validity and scalability. Fine-tuning produced stronger outcomes overall, yielding items that were more contextually grounded, linguistically coherent, and balanced than those generated through prompting alone, with some requiring minimal or no revision; however, generating higher-order items involving discourse-level reasoning remained challenging. The expert comparison showed that fine-tuned items performed comparably to human-authored items on passage dependence but remained relatively weaker in avoiding absolute language and in targeting localized spans of necessary information. Additionally, issues such as longest-correct-option bias and uneven key distribution persisted, indicating limitations inherent in LLM-generated items. These findings demonstrate the value of fine-tuning for improving item quality and stability, while underscoring the continued need for multidimensional evaluation frameworks and expert benchmarking to ensure construct-aligned, valid assessment design.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100623"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148230834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Preparing pre-service teachers for responsible generative AI use: Curriculum implications for ethics, privacy, and AI literacy","authors":"Lucas Kohnke ,&nbsp;Di Zou ,&nbsp;Chun Lai ,&nbsp;Mingyue Michelle Gu","doi":"10.1016/j.caeai.2026.100617","DOIUrl":"10.1016/j.caeai.2026.100617","url":null,"abstract":"<div><div>This study investigated what pre-service teachers should learn about generative artificial intelligence (GenAI) by backtracking from anticipated classroom dilemmas to teacher-education design needs. Through qualitative interviews with 17 pre-service teachers at a Hong Kong university, the research explored perspectives on integrating AI into education, focusing on ethical concerns and pedagogical strategies. Grounded in the interpretivist paradigm, it revealed concerns about academic integrity, privacy and the challenge of maintaining essential human skills in a GenAI-driven educational environment. Although pre-service teachers recognised GenAI's potential to enhance teaching and learning, they expressed concern about over-reliance on GenAI tools, lack of data privacy and inequitable access. The participants emphasised the need for comprehensive AI literacy training in teacher-education programmes and strategies to promote critical thinking while leveraging AI's benefits. The research suggested that teacher-preparation programmes must address these challenges by incorporating technical knowledge, pedagogical strategies for AI integration and information about the ethical challenges posed by AI use. The findings contribute to teacher-education curriculum design by identifying how GenAI literacy can be embedded across coursework, assessment training, practicum preparation and ethical decision-making.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100617"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148230819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An exploratory machine learning approach to understanding determinants of future ChatGPT use in higher education","authors":"Chaman Verma ,&nbsp;Deepak Kumar","doi":"10.1016/j.caeai.2026.100613","DOIUrl":"10.1016/j.caeai.2026.100613","url":null,"abstract":"<div><div>AI tools such as ChatGPT are used in the higher education increasingly yet the limited empirical evidence explains how the students incorporate these tools into their academic work. This study examines how students’ perceptions and demographic characteristics relate to their intended use of ChatGPT academically. The study focuses on understanding the educational meaning rather than maximizing the algorithmic performance. The survey data were collected from 166 university students for this purpose. It is analyzed to identify the key learning related constructs associated with ChatGPT use. An exploratory machine learning (ML) framework was applied to examine the relationships among these constructs while maintaining the interpretability.</div><div>The analysis suggests that learning support, accessibility and academic engagement are the important factors that were associated with students’ intentions to use ChatGPT in academic contexts. These elements indicate that students often view AI tools as resources for clarifying the concepts, supporting coursework and enhancing the learning efficiency. At the same time, the concerns related to reliability, accuracy, and academic integrity influence how students evaluate the appropriate use of such tools. Overall, the findings suggest that ChatGPT adoption in higher education is shaped largely by the pedagogical and contextual conditions rather than technological performance alone. The study provides the exploratory insights that may inform the instructional design, AI literacy initiatives, and the responsible integration of generative AI (GAI) within the higher education environments.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100613"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148167572","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generative AI-enhanced learning experiences for computational thinking: A systematic scoping review and design guidelines","authors":"Abdessalam Ouaazki ,&nbsp;Antonette Shibani ,&nbsp;Simon Knight ,&nbsp;Adrian Holzer","doi":"10.1016/j.caeai.2026.100608","DOIUrl":"10.1016/j.caeai.2026.100608","url":null,"abstract":"<div><div>Computational thinking (CT) skills are crucial to teach. The rise of Generative AI could potentially revolutionize the teaching of these skills by changing learning experiences and outcomes. Educators are rapidly integrating these tools into their educational activities; however, the best practices for doing so remain unclear. Existing reviews on GenAI are either too broad in regard to studying specific tools such as ChatGPT for learning, or too limited with terms of studying computational thinking with a focus on programming. As such, this paper aims to summarize existing knowledge and extract best practices and avenues for future research and practice around GenAI and computational thinking skills. Our contributions are two-fold: 1) Insights from a systematic scoping review of studies examining the use of GenAI to support the teaching of CT skills in diverse contexts and their reported learning outcomes, 2) Design guidelines to inform effective use of GenAI in CT education for pedagogical practice. Our results reveal a young but rapidly growing research field, with most interventions focusing on undergraduate students and basic programming tasks, often using off-the-shelf tools with limited integration. GenAI is typically used as a coder, tutor, debugger, or ideator, with mixed effects on learning outcomes. A key challenge is the tension between overreliance by beginners, who may offload thinking to GenAI, and under-utilization by advanced learners in complex projects. We derive seven guidelines to provide actionable recommendations from the findings to effectively integrate GenAI for CT while minimizing associated risks and guiding responsible AI use in education. We also suggest directions to inform the design of future systems and CT pedagogies.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100608"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148167573","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing AI literacy for educators: Where to start and to what end?","authors":"Bo Pei ,&nbsp;Jie Lu ,&nbsp;Zhaowei Zhang ,&nbsp;Priscilla Tuffour ,&nbsp;Sanghoon Park","doi":"10.1016/j.caeai.2026.100550","DOIUrl":"10.1016/j.caeai.2026.100550","url":null,"abstract":"<div><div>As AI has become an integral part in current teaching and learning practices, educators' capacities to use AI technologies effectively and responsibly are closely tied to the quality of instruction and student learning outcomes. To provide a comprehensive examination for cultivating the relevant capacities, this study conducted a systematic literature review about AI literacy for educators. Informed by Bloom's Taxonomy, we investigated the existing research with a layered progression structure from five dimensions: definitions of educators' AI literacy, fundamental knowledge for understanding of AI, AI educational practices, educators' perspectives of AI applications, and pedagogies of integrating AI. The findings of this study revealed three key dimensions (i.e., human-AI interactions, harnessing AI tools, and ethical and societal implications) of educators' AI literacy and highlighted the importance of interrelationships among these dimensions. Furthermore, our study identified the fundamental knowledge that educators need to understand, the instructional scenarios in which the AI applications are applied, the associated opportunities and challenges, and the pedagogical approaches that have been proposed to effectively scaffold educators' engagement with AI. Overall, this literature review underscores the multidimensional and context-relevance nature of AI literacy for educators, developing from the interplay of multiple competencies within specific educational contexts. Finally, the study concludes by discussing implications for providing actionable insights to design training and professional development programs that better prepare educators to navigate the AI-driven educational environments.</div></div>","PeriodicalId":34469,"journal":{"name":"Computers and Education Artificial Intelligence","volume":"10 ","pages":"Article 100550"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}