Knowledge tracing (KT) predicts learners’ evolving knowledge states by tracking their performance over time. While temporal dynamics have been the focus of most KT studies, spatial structures among knowledge components (KCs) remain underexplored, despite containing rich latent information. Prior work suggested that inter-KCs relationships can enhance KT performance, yet the impact of hierarchical spatial structures remains unclear. This study investigates how multilevel spatial relationships among KCs affect the performance of KT models. Using causal structure learning, we infer causal links among KCs and incorporate the resulting spatial structures into both deep learning and traditional machine learning KT models. Experimental results showed that incorporating second-order spatial structures yielded consistent performance gains. These findings underscore the value of spatial structural information in KT. Furthermore, interpretable feature analyses illustrated how spatial features shape diagnostic predictions, providing insight into factors underlying students’ learning challenges. This spatial perspective not only improves KT models’ performance but also has the potential to inform more targeted and effective instructional strategies.
Amid the rapid development of artificial intelligence (AI), this paper underscores the historical inevitability of exploring long-term development trends in education and emphasizes AI’s growing influence in this field. It is anticipated that AI will become the leading force in education in the future, and for this reason, human teachers may be required to seek a path of symbiosis with AI. Enlightened by Opinions on deepening the implementation of the “AI Plus” initiative issued by the State Council of the People’s Republic of China (2025), this paper explores the inevitability of human society’s continued development of AI and the synchronization between China’s digital education transformation and building China into a leading country in education. Mainly, the paper analyzes AI’s impact on the underlying logic of education: Driven by intellectual curiosity, students can complete the learning process from building knowledge from scratch to profound understanding through continuous interactions with AI, which in turn, leads to AI gradually replacing education’s knowledge imparting function. Consequently, this change may weaken the primacy of competency development in the long run, leading to substantive alterations in multiple dimensions of the traditional educational model. This paper clarifies the analytical logical framework covering four key dimensions—from the micro to the macro levels—proposed in White paper on China’s smart education (Ministry of Education of the People’s Republic of China, 2025), namely future teachers, future classrooms, future schools, and future learning centers, laying a theoretical foundation and ultimately providing an overarching framework to guide the subsequent research of the relevant field.
Large language models (LLMs) are increasingly used to grade written responses, yet large-scale benchmarks against human expert evaluation remain scarce, especially across languages with differing resource levels. This study evaluates ChatGPT-4o using a reranked retrieval-augmented generation framework to grade Finland’s national high-stakes matriculation examination based on 1,016 students’ open-ended responses. We examined GPT-4o’s agreement with official grades, its recognition of grading-relevant keywords, and the effect of translated responses from a low-resource language (Finnish) into a high-resource language (HRL) (English). Using descriptive statistics and correlation analyses, the results show that GPT-4o’s grades on a 0–15 scale closely matched human expert evaluations; 75.00
The integration of large language models (LLMs) into higher education is rapidly transforming learning and research processes. This study investigates the adoption of one such tool, DeepSeek, a Chinese LLM, in Bangladesh’s higher education sector. It introduces geopolitical concern as a novel construct within the extended unified theory of acceptance and use of technology (UTAUT) framework to assess how national security, strategic alignment, and technological dependency influence trust, privacy concerns, and the behavioural intention to adopt AI tools in an academic context. A quantitative survey of 202 Bangladeshi university students was conducted using the extended UTAUT model. Partial least squares structural equation modelling was employed to analyse the impact of constructs such as performance expectancy, effort expectancy, and geopolitical concerns on behavioural intention and use behaviour. Geopolitical concerns significantly shaped trust and privacy perceptions, indirectly reducing adoption intentions. Despite privacy risks, students viewed DeepSeek in a favourable light due to its free access, cultural alignment, and potential to reduce Western tech dependency. Performance expectancy and facilitating conditions were key drivers of adoption, while social influence and hedonic motivation had negligible effects. This study contributes to the integration of geopolitical factors into AI adoption frameworks, highlighting Bangladesh’s unique position amid a growing diversity of globally available AI models. It emphasises the need for culturally congruent, sovereignty-sensitive AI tools in the Global South. Policymakers and AI developers should address geopolitical sentiments and infrastructure gaps (e.g., internet reliability) to foster adoption. DeepSeek’s frugal innovation model provides a blueprint for emerging economies seeking affordable AI solutions.
Personalized learning resource recommendation aims to provide learners with appropriate learning resources to alleviate information overload caused by the explosive growth of data on online learning platforms. Current research predominantly utilizes student interaction data to enhance the quality of recommendations. However, this approach neglects the intricate dependency networks among learning resources, which directly influence the effectiveness of knowledge acquisition pathways, and lacks the capacity to model individual learning abilities and objectives. To address these limitations, this study introduces a unified learning resource recommendation method (ULRRM), which integrates multidimensional graph information and employs conceptual graphs as an intermediary framework to unify resource representations across varying levels of granularity. Specifically, a resource dependency graph is established to guide resource-dependent learning through conceptual dependency relationships, thereby encoding the topological constraints of resources. Then, a local–global dual view is constructed using session history to capture both short-term behavioral patterns and the evolution of long-term interests, thereby enabling the recommendation of learning resource sequences that incorporate multidimensional graph information. Extensive experiments conducted on real datasets demonstrate that the proposed ULRRM method surpasses baseline approaches across several widely recognized evaluation metrics.
This paper examines how Chinese secondary and tertiary English as a second language (ESL) learners engage with generative AI (GenAI) tools, such as ChatGPT, Claude, Doubao, and Pigai, not merely as writing aids but as coauthors in the academic writing process. Against the backdrop of an assessment-centered education system that emphasizes memorization and structured learning, GenAI opens new possibilities for dialogic learning and critical thinking. Drawing on case studies and current research, this paper examines how prompt engineering, both as a technical and pedagogical skill, supports digital literacy, rhetorical awareness, and metacognition. It further investigates how GenAI scaffolds language production and supports student agency, while also presenting risks such as epistemic dependency, reduced critical thinking, and ethical ambiguity. The concept of Human–AI co-learning is advanced as a theoretical framework for understanding this interaction. The paper concludes by calling for critical AI literacy, educator mediation, and culturally responsive pedagogy that reconciles traditional Chinese learning practices with reflective engagement in digital environments. By reframing GenAI from a shortcut to a scaffold, this study proposes a pedagogical model that empowers learners to reclaim authorship and engage more deeply in academic inquiry.
The course project report (CPR) is a crucial component for assessing students’ learning outcomes from courses they are studying. It assesses practical skills, academic writing, and logical thinking. In recent times, researchers have increasingly leveraged large language models (LLMs) to promote automated essay scoring (AES) in the education intelligence field due to its strong generalization and reasoning abilities. However, the existing LLM-based AES method design is based solely on writing proficiency and inevitably ignores the importance of assessment of cognitive engagement and practical competencies in CPRs. Additionally, CPR writing is a reflective process that includes knowledge-inquiry and cognition through critical thinking (CT), which have rarely been explored in the design of prompts for specific LLMs. To tackle this issue, we propose a novel, guided generative AI (GenAI) prompting framework for automated CPR assessment. It is created by integrating the Paul-Elder critical thinking concept into prompt design to enhance domain-specific knowledge transfer and the analytical capabilities of GenAI LLMs. Rather than focusing solely on language structure or writing skills, our approach emphasizes critical thinking evaluation using the Paul-Elder CT framework. Specifically, our framework—PEG-Prompt—evaluates CPR across six dimensions—structure, logic, coherence, originality, citation, and knowledge proficiency—to evaluate CPRs comprehensively from the aspects of practical competencies, analytical reasoning, and writing skills. To further enhance the CPR assessment performance of PEG-Prompt, we combine PEG-Prompt with extracted key content from reports and representative examples of few-shot scoring. Experimental results demonstrate that PEG-Prompt significantly improves the correlation between LLM-generated scores and human scores. The enhanced framework may enable students to receive helpful feedback and summaries of their CPR results through GenAI once it has been calibrated with human evaluators.
Autism is a developmental disorder that manifests in early childhood and persists throughout life, profoundly affecting social behavior and hindering the acquisition of learning and social skills in those diagnosed. With technological advancements, an increasing array of tools is being utilized to support the education of students with autism spectrum disorder (ASD), aiming to improve their educational outcomes and social capabilities. Numerous studies on autism intervention have highlighted the effectiveness of social robots in behavioral treatments. However, research on their integration into classroom settings for children with this condition remains sparse. This paper describes the design and implementation of a group experiment in a collective classroom setting mediated by a NAO robot. This involved special education teachers and the NAO robot collaboratively conducting classroom activities, aiming to foster a dynamic learning environment through interactions among teachers, the robot, and students. Conducted in a special education school, this served as a foundational study in anticipation of introducing extended robot-assisted classroom sessions at a later date. Data from the experiment suggest that ASD students in classrooms equipped with a NAO robot exhibited notably improved performance compared to those in regular classrooms. Our preliminary findings indicate that NAO robots significantly enhance focus and classroom engagement among students with ASD, potentially improving educational performance and fostering enhanced social functioning.
The integration of generative artificial intelligence (GenAI) for dissertation writing has sparked debates regarding where it can augment the writing process, which must exclusively have human intelligence at its core, and how to write GenAI prompts that produce effective output. The present study’s exploration of this topic is based on qualitative data gathered from a survey of 86 doctoral students and 7 thesis supervisors in the social sciences and humanities disciplines. We applied the AI Assessment Scale developed by Perkins et al. (2024) to evaluate GenAI’s role across various stages of doctoral dissertation writing and to explore pedagogical adaptations of GenAI to support dissertation writing in the contexts of the social sciences and humanities. Our findings indicate that GenAI can be fully utilized to improve writing mechanics, including grammar, structure, and coherence, by enhancing clarity and efficiency. GenAI also proves beneficial in analyzing larger datasets by defining a coding frame, identifying trends, and conducting sentiment analyses. GenAI can be utilized in argument structuring by organizing literature, suggesting logical ways to arrange sentences, and generating counterarguments. The participants agreed that exploring these applications saved their time and allowed them to focus on a deeper intellectual engagement. However, they recommended limiting or prohibiting GenAI use in areas that require critical reasoning, originality, and cultural context. Moreover, they underscored that AI-generated content may lack accuracy and contextual depth, thus requiring careful human validation against vague expressions. This study focuses on prompt literacy and provides a scale to utilize GenAI for doctoral dissertation writing.
Viewed as a threat to academic integrity, several questions have arisen about the dark side of AI. While AI offers several opportunities for teaching and learning, questions have also arisen regarding its impact on the future of education. This paper offers a critical perspective on the dark side of AI in education by drawing on critical social theory. It examines how AI can deepen the processes of educational inequality, power relations, and cultural identity. The findings illustrate that differential levels of economic development, digital infrastructure, cultural norms, and technological capacity significantly influence the integration and impact of AI across global education systems. To counter these risks, this paper calls for AI developers to adopt an inclusive design from the start of the application to ensure that these tools are accessible, adaptable, and affordable to the economies of the Global South. The paper further highlights the urgency of incorporating African voices, values, ethics, and worldviews into global AI governance conversations. Africa and other developing countries must not merely be sites of AI deployment but key actors in shaping the educational futures these technologies enable. Their expertise is essential to confronting colonial legacies in technological innovation and ensuring that AI advances democratic empowerment rather than digital dependency.
The integration of artificial intelligence (AI) in education is reshaping how learning is designed, delivered, and governed. This study examines five interrelated domains: integrating AI in education, learner development and personalization, curriculum and educational content, role of teachers, and governance and regulation. It argues for systemic, ethically grounded educational approaches that promote equity, lifelong learning, and collaborative policymaking.
The rapid diffusion of AI in education is commonly framed as a pedagogical, ethical, or technological challenge. This paper argues that AI constitutes a fundamentally governance-related issue, as it reshapes how authority, responsibility, and accountability are distributed within education systems. Building on governance theory and critical scholarship on digitalisation, platformisation, and datafication, the paper conceptualises AI as a systemic and transversal actor that operates across boundaries between centralised regulation and decentralised educational practice. The paper develops an analytical framework grounded in reconfigured hybrid governance models and introduces a conceptual distinction between foundational AI infrastructures (AI models), AI content, and AI tutors. Through a structured literature review and conceptual analysis, it demonstrates how existing governance arrangements—designed for earlier phases of digitalisation—are increasingly misaligned with AI-mediated education systems. The analysis highlights four key governance risks, including the homogenisation of learning processes, intensified surveillance, blurred accountability, and the erosion of student and teacher agency. In response, the paper proposes a reconfigured hybrid governance approach that differentiates governance responsibilities across system levels and AI functions. It further advances concrete policy recommendations aimed at operationalising this approach through regulatory oversight, accountability mechanisms, and the protection of educational purpose and professional autonomy. By foregrounding governance as a central analytical and policy concern, the paper contributes to current debates on how education systems can harness the benefits of AI while safeguarding democratic values and human-centred education.
The integration of artificial intelligence (AI) into education marks a critical transition, not only through the adoption of new tools but by challenging the epistemological foundations of teaching and learning. AI reshapes how knowledge is produced, mediated, and evaluated, raising critical questions around equity, agency, and accountability in increasingly data-driven environments. Its emergence compels educators and policymakers to reconsider long-standing assumptions about what counts as learning, how it is measured, and who benefits from technological change. This paper examines how AI is transforming key structures and practices across the broader education landscape, with a particular focus on school education as a strategic and emblematic site for early intervention and pedagogical innovation. While many of the transformations discussed in the paper are relevant across educational levels, schools represent a crucial point where students first experience the cognitive, social, and ethical dimensions of AI, and where systems can act early to promote inclusion, reflection, and readiness. Drawing on major European frameworks, this paper analyzes how AI is reshaping four interdependent pillars of education: curricular content, teaching paradigms, assessment systems, and governance structures. International case-based insights illustrate diverse implementation strategies while also revealing persistent challenges, such as digital inequalities, gaps in teacher preparation, and limited availability of robust mechanisms for algorithmic accountability. Adopting a conceptual, policy-informed approach, this paper synthesizes scholarly literature, European regulatory frameworks, and implementation evidence to propose a systemic view of educational transformation. Rather than framing AI as a mere driver of automation, the paper argues for a transformative approach rooted in equity, human agency, and democratic values. In practical terms, it distills policy-relevant guidance on ethics-by-design, human-in-the-loop safeguards, and capacity building for teachers and school leaders to enable responsible, system-level implementation. The conclusions highlight that, when supported by coherent policy infrastructure and teacher empowerment, school education systems can align technological innovation with inclusive, ethical, and future-oriented learning, ensuring that AI contributes to social justice rather than reinforcing existing inequalities.
The school digital renewal process (SDRP) has evolved from adoption at the infrastructure level to deep pedagogical transformation centered on personalized, competence-based learning. Traditional indicators, such as device availability or connectivity, lose relevance at advanced SDRP stages. This paper proposes a novel, evidence-based approach to constructing indicators that capture shifts in learning content and organization through an automated analysis of schools’ digital footprints using AI tools, such as publicly available digital resources. Drawing on the Bloom’s revised taxonomy and empirical data from international schools, we demonstrate the feasibility of tracking second-order changes without relying on teacher surveys. The framework supports the comparative monitoring of digital transformation aligned with the demands of the age of AI. The paper introduces a groundbreaking innovation: the use of AI tools for gathering and analyzing indicators from publicly available digital resources in schools. This approach offers a scalable and cost-efficient method of tracking and evaluating SDRP at the later stages of development.
The credit transfer system (CTS) is a complex learning and educational management system involving multiple entities such as learners, schools, and the government. Considering the external constraints, strategic assumptions, and payment assumption, and based on the benefit relationship among stakeholders, this paper attempts to construct a game-theoretic analysis framework of multi-stakeholder participation in the construction of a CTS, which focuses on the learning strategies of learners, the investment strategies of schools, and the management strategies of the government among the three entities considered in this study. Furthermore, the cost allocation and benefit demand of stakeholders to build a three-subject dynamic evolutionary game model have been analyzed. The authors also conducted a numerical simulation to analyze the decision-making mechanism of learners, schools, and the government under the CTS. The findings show that “strong participation” by learners in ability improvement, “active participation” by schools in building high-quality teaching resources, and “strong dominance” in supervision and support by the government together constitute the optimal strategy in constructing a suitable CTS.
The rapid development of artificial intelligence (AI) is accelerating the digital transformation of higher education. Today, “AI + Education” has become a key feature of Education Informatization 2.0 Action Plan in China. This study presents practical experiences in applying AI to programming courses. First, the global trends in AI-powered teaching and learning are analyzed. Key challenges in programming education that can be addressed by AI are then identified. Focusing on common teaching problems, an introductory programming course is used to demonstrate the construction of a course engine powered by large language models. This engine enables the creation of intelligent courses, driving innovation in teaching scenes, and transforming both teaching and learning methods. The exploration then extends to the design of AI-enhanced teaching and learning environments, featuring AI teaching assistants and AI learning companions. These tools provide scalable, differentiated, and personalized support for teachers. They also enable one-on-one, adaptive, and customized learning experiences for students. An integrated learning support system is proposed, which combines courses, training, competitions, testing, evaluation, and certification. The goal is to build a smart teaching ecosystem with knowledge services, personalized learning, and instructional support, as well as to realize the entire teaching process of “course-training-competition-testing-evaluation” empowered by AI for all elements and all time periods. Furthermore, the intelligent interactive virtual massive open online courses (IMOOCs) for C programming is developed. A new hybrid teaching model based on IMOOC, which integrates virtual and real elements and promotes cross-domain collaboration, has also been explored. Potential risks of overreliance on AI tools are discussed, together with strategies to address them. Finally, future trends and challenges in “AI + Higher Education” are examined. The study argues that AI will unlock new possibilities for reshaping how higher education is delivered and experienced.
Transforming engineering education in the AI era requires an evaluation of new instructional tools and a reconceptualization of the division of labor among teachers, students, and intelligent learning companion systems (ILCSs). This work explores how a retrieval-augmented generation intelligent learning companion can be embedded within a human–AI collaborative teaching model by using an analog circuit laboratory instruction as a case study. A controlled experiment compared traditional teacher-led guidance with system-supported instruction, focusing on three core dimensions: knowledge acquisition, learning effect (cognition, skill, and emotion), and flow experience (cognitive control, immersion and time transformation, loss of self-consciousness, and autotelic experience). The results indicate that while the system showed a limited impact on knowledge acquisition and emotion, it significantly enhanced skills, immersion and time transformation, and autotelic experience. These findings suggest that ILCSs serve as effective complements in practice-oriented engineering education, particularly in terms of providing personalized support and instant feedback strengthening hands-on learning and student engagement. Such companions cannot fully serve as a substitute for teacher-led conceptual scaffolding or emotional guidance. The study’s theoretical contribution lies in emphasizing the importance of role allocation in human–AI collaborative education and offers practical implications for the design of learner-centered, practice-oriented instructional models in intelligent education.
Autism, also known as autism spectrum disorder (ASD), is a neurodevelopmental condition associated with differences in emotional processing and social communication. Electroencephalogram (EEG) analysis presents a unique avenue for exploring its underlying neural evolutionary mechanisms. To this end, this study explored the similarities and differences in emotional processing between children with ASD (ASD group) and those without ASD (control group) using EEG. The final analysis included 45 children: 22 with ASD (mean age = 5.29, age range: 2–8) and 23 without ASD (mean age = 4.37, age range: 2–6). EEG signals were synchronously collected during stimulation with a series of emotional videos. The t-tests on the collected EEG data were performed to determine any statistical differences in power spectral density, sample entropy, and differential entropy values between the groups. A functional connectivity analysis was also performed for a more comprehensive understanding. SHapley Additive exPlanations (SHAP) were applied to validate the findings, ensuring their robustness and reliability. The results showed that the ASD group exhibited reduced beta-band activity in the frontal regions and enhanced delta-band activity in the temporo–occipital areas compared to the control group. Entropy analyses revealed lower brain complexity in the ASD group. Functional connectivity results showed increased high-frequency synchronization in the ASD group but more coordinated low-frequency connectivity patterns in the control group. Moreover, the application of SHAP-based analysis with XGBoost confirmed the significance and predictive value of beta- and delta-band features in the frontal and occipital regions, providing potential biomarkers for distinct emotional processing in individuals with ASD. Overall, this study holds potential to facilitate the understanding of the neuronal mechanisms underlying emotional processing in individuals with ASD and inform the development of targeted neurotherapeutic interventions.
Since the advent of the 21st century,digital technologies represented by artificial intelligence(AI),Big Data,and virtual reality(VR)are reshaping the educational ecosystem at an unprecedented pace.In parallel,research on digital education worldwide has also seen explosive growth.
