Abstract With several fragmented literature reviews and meta-analyses on Large-Language Models (LLMs) in education, this study provides a synthesis of these reviews focusing on the role of LLMs to facilitate different pedagogy paradigms to reveal research trends, existing gaps, and future directions. The synthesis adheres to the PRISMA guidelines and AMSTAR checklist to analyze 50 reviews to find out the trends of research in terms of publication year, geographic regions, types of reviews, types of research questions, pedagogy paradigms addressed, and challenges encountered. Findings revealed that constructivism, cognitivism, and connectivism emerged as the most frequently addressed paradigms, indicating the role of LLMs as (i) tools for learning discovery or scaffolding, where learners actively construct knowledge, (ii) tools to improve cognitive tasks (e.g., recall, comprehension, problem-solving), or (iii) tools to facilitate connections between diverse knowledge sources or enable networked learning environments. Additionally, several challenges emerge, which are related to cognitive load and processing, accuracy and comprehension, cultural sensitivity and diversity, learner autonomy and self-expression, among others. This study offers valuable insights into the evolving roles of LLMs in facilitating paradigms that contribute to a deeper understanding of the pedagogical implications, challenges, and opportunities presented by LLM adoption in education.

Background Previous research has shown that adding a pedagogical agent (PA) exhibiting humanlike embodiment cues into lecture videos has the potential to facilitate multimedia learning; however, empirical evidence regarding the impact of learning with these embodied PAs in virtual reality (VR) settings remains limited.Objectives This study aims to investigate the effect of the media environment type and the presence of embodied PA on learners' learning performance, intrinsic motivation, social presence, and cognitive load in VR lecture video settings.Methods A total of 125 participants were divided into one of four experimental conditions formed by a 2 (embodied PA: present vs. absent) & times; 2 (media environment: 2D screen vs. immersive VR) factorial between-subjects experimental design.Results and Conclusions Results indicated that (a) VR significantly enhanced intrinsic motivation and social presence; meanwhile, the interaction effects between the two variables were observed: (b) embodied PA in VR led to better learning performance and social presence than on the screen, while this effect on learning performance was reversed in no-PA conditions; moreover, (c) under screen conditions, no-PA resulted in higher learning performance and social presence than embodied PA. These findings highlight that the media environment serves as a boundary condition for embodied PA effectiveness, support the combined use of embodied PA and VR to enhance learning content, and also indicate the advantages of delivering verbal-narrated video instruction via a 2D screen when it is difficult to create embodied PA and afford VR equipment.

Smart learning environment (SLE) modules are rapidly expanding with the advancement of technology. However, the concept is still new to many educational institutions in developing countries, with scant studies conducted using the African context. To avoid being left behind in the steps towards modernization, the qualitative study proposes the need to migrate from a traditional learning environment (TLE) to a SLE. Thirty-eight (38) teachers in Ghana who transitioned from TLE to a SLE were recruited to explore their pedagogical experiences in instructing in TLE (old classrooms) and SLE (new classrooms). Generally, teachers favoured SLE compared to TLE. Teachers were motivated to instruct in smart schools because of the affordances of smart technologies. They considered the ambience created in SLE as conducive to promoting their psychological well-being, facilitating their professional development, fostering good assessment strategies, and building twenty-first-century skills in learners needed in the labour market. However, a majority of the teachers perceived self-regulation and social interactions to be more effective in TLE. The key challenges in the SLE included network lag, limited funds to purchase data bundles, malfunction of learning management systems (LMS), and difficulty in designing a Smart App tailored to instructors’ and students’ interests. It is recommended that school administrators and policymakers continually revise SLE frameworks for a successful smart education. Future researchers can investigate how to promote personalized and adaptive learning in SLE and promising tools to assist instructors in technology-enhanced learning environments.

The question of how to use artificial intelligence generated content (AIGC) properly to enhance learning among college students is a key concern for contemporary educators. Although previous studies have discussed the influence of AIGC on college teaching and student learning and its functions in this context, there remains a lack of discussions regarding ways of guiding students' use of AIGC and studies on the specific topic of helping college freshmen use AIGC properly. Based on the substitution, augmentation, modification and redefinition (SAMR) model, this study develops a progressively active teaching framework that integrates AIGC into learning. This framework is used to design learning activities for general education courses targeting freshmen. This exploratory study was conducted in the context of a 16-week course. During the teaching process, AIGC interaction log data and AIGC experience records were collected from students, following which data processing was conducted using the discourse analysis, quantitative statistical analysis, and epistemic network analysis (ENA) methods to obtain the ultimate results of this study: (1) A combination of active teaching with the SAMR model can improve the quality of interactions between students and AIGC; (2) teaching strategies rooted in active learning can enhance students' ability to use AIGC; and (3) improvements in students' technical skills strengthen the quality of their interactions with AIGC. This study makes novel contributions to the literature on active learning strategies for teachers and curriculum designers, and it offers practical guidance for educational practitioners and college students regarding the integration of AI technology into both teaching and learning.

Generative artificial intelligence (GenAI) has been increasingly used in teaching and learning. However, its effect on the learning outcome of creative problem-solving (CPS) remains inconclusive. To address this problem, this mixed-methods study examines the dual role of GenAI in college students' CPS using a natural experiment and a post-interview. By comparing outcomes of a design competition between non-GenAI student users (Year 2023, N-team(s) = 10; N-student(s) = 56) and GenAI users (Year 2024, N-teams = 18; N-student(s) = 80) at three stages (Npre+mid+post = 82), supplemented by competition mentor interviews analysis, we found that GenAI is significantly positive-associated with students' CPS performance (p < 0.001), with greater gains among students with lower baseline creativity. Additionally, not only did students' CPS performance improve significantly across the three testing stages (p < 0.001), but also the trajectory of improvement differed markedly between cohorts: students with ChatGPT support exhibited steeper gains across stages (p < 0.001). However, the content analysis of mentor interviews identified risks of idea homogenization that emerged from the submitted work and students' over-reliance on GenAI tools, with students failing to critically validate AI outputs. Mentor insights further highlight the mediating role of AI literacy and team dynamics: Teams with balanced human-AI collaboration achieved more innovation than those overly dependent on GenAI. These findings underscore GenAI's paradoxical impact: a catalyst for stimulating creativity and a potential inhibitor of independent thinking. These results suggest that teachers and instructional designers should adopt active learning-oriented instructional design strategies to balance student-AI collaboration on creativity. The study advances research discourse on human-AI co-creativity while offering actionable strategies for sustainable AI integration in education.

As global education systems are going through rapid digital transformation, identifying how smart education contributes to educational quality has become a critical policy and research imperative. This study conceptualizes smart education as a multi-layered ecosystem, integrating Bronfenbrenner’s ecological systems theory with the smart education ecosystem framework. It operationalizes six core dimensions, student development, teacher capacity, learning environment, governance, equity, and improvement culture, using internationally comparable data from 79 countries. Based on fuzzy-set Qualitative Comparative Analysis (fsQCA), the study uncovers seven sufficient configurations for achieving educational quality, which are clustered into three transformation pathways: governance-led, human capital-centered, and system-integrated models. Results reveal clear income-based divergence. High-income countries often adopt professionalized configurations with flexible governance and advanced teacher capacity, while low-income countries rely on basic enablers such as ICT infrastructure and centralized governance. Middle-income countries exhibit institutional-equity configurations that combine ethical orientation, targeted innovation, and adaptive structures. These findings demonstrate the importance of ecosystemic coherence, functional substitutability, and causal asymmetry in shaping effective smart education systems. Rather than advocating for one-size-fits-all reform models, the study highlights the need for context-sensitive configurations aligned with national capacities and priorities. The research provides a novel configurational approach to smart education, offering theoretical insights and practical guidance for designing resilient and equitable digital education systems worldwide.

Purpose: This study explores critical dimensions of emerging learning spaces, including their design requirements, associated challenges and opportunities, and best practices for integration into existing educational frameworks.Method: A systematic literature review from Web of Science, SpringerLink, and Google Scholar (2010-2023) identified 22 emerging learning spaces. A Delphi survey with 12 experts further identified the top 10 emerging spaces and highlighted their design needs, challenges, and integration strategies.Findings: Architectural, technical, and ethical issues are crucial in designing these spaces. Generative AI tools like ChatGPT could transform these environments. A framework was developed, identifying key learning scenarios such as self-directed, experiential, and immersive learning.Conclusion: The study recommends the creation of innovative learning spaces to support educational sustainability. It calls for further research to explore spaces tailored to diverse learner groups and to investigate the ethical implications of emerging technologies in education.

Self-regulated learning (SRL) has been regarded as one of the indispensable factors affecting students’ academic success in online learning environments. However, the current understanding of the mechanism/causes of SRL in online ill-structured problem-solving remains insufficient. This study, therefore, examines the configural causal effects of goal attributes, motivational beliefs, creativity, and grit on self-regulated learning. With the fuzzy sets approach (fsQCA), the proposed association was analyzed based on a sample of students (n = 88) participating in an educational design competition activity. The results uniquely revealed the predictive factors of SRL at both high and low levels. In addition, it was found that no single condition of factors leads to the prediction of high or low self-regulation. More specifically, different conditions of factors, in terms of gender, goal attributes (goal setting and achievement goals), grit, task value, creativity, and self-efficacy, can largely predict high and low self-regulated learning during ill-structured problem-solving in the context of online learning. Implications for theory and policy prescriptions were discussed to enhance self-regulated learning in online ill-structured problem-solving.

IntroductionWhile several studies investigated the effect of blended learning on students’ learning achievement, scant information exists on whether using artificial intelligence (AI) in blended learning could further contribute to the obtained effect.MethodsTo effectively address the challenges and opportunities presented by blended learning and AI, the present study conducts a meta-analysis to systematically examine the impact of AI-enhanced blended learning on students’ learning achievement, considering the significant role of multiple variables in shaping this achievement, including the type of AI technology, instruction duration, research design, and sample size as well as across different educational levels and subject areas. Specifically, 21 studies (N = 2,873 participants) were meta-analyzed.ResultsThe obtained results revealed that AI has a medium effect (g = 0.5) on students’ learning achievement in blended learning. Particularly, personalized systems in blended learning had the highest effect (i.e., large) compared to chatbots and intelligent tutoring systems. Finally, it is seen that the educational context (grade level and educational subject), as well as the experiment type (research design, intervention duration, and sample size), moderate the effect of AI on students’ learning achievement in blended learning.DiscussionThe findings of this study can help researchers and practitioners better understand the effects of AI in blended learning, thereby contributing to a better design of teaching and learning experiences accordingly.

Scant information exists about how AI with its different technologies might affect learning achievement in different educational fields across different educational levels and geographical distributions of students. Closing this gap can therefore help stakeholders understand under which learning conditions artificial intelligence in education (AIEd) might work or not, hence achieving better learning achievement. To address this research gap, this study conducted a meta-analysis and research synthesis of the effects of AI application on students’ learning achievement. Additionally, this study conducted one step forward to analyze the field of education, level of education, learning mode, intervention duration, and geographical distribution as moderating variables of the effect of AIEd. The Hedges’ g was computed for the effect sizes, where 85 quantitative studies ( N = 10,469 participants) were coded and analyzed. The results indicated that the total effect of AIEd on learning achievement is very large ( g = 1.10, p < 0.001). Particularly, chatbots achieved a very large effect, while Intelligent Tutoring Systems (ITS) and personalized learning systems had large effects. The results also show that the AIEd effect is moderated by the field of education, level of education, learning mode, intervention duration, and geographical distribution of students. The findings of this study can be useful to both researchers and practitioners as they highlight how and when AIEd integration can be effective, hence being beneficial to enhance learning achievement.

Purpose This study explores the concept of learning power (LP) from a Confucian context, through a quantitative and heuristic phenomenological study. Despite that LP has emerged in the English academic literature to include various dimensions within different frameworks, there is a knowledge gap on the concept from different cultures, particularly the Confucian culture. Teachings of Confucius in the Analects emphasize the concept of love of learning (haoxué 好学), and modern literature in the Chinese language reveals various dimensions and frameworks of the concept of LP. However, no previous reviews or connections have been conducted. Design/methodology/approach Conducted in two phases, the first phase of the study, through a systematic literature review, utilizing the PRISMA statement, of Chinese academic publications, seeks to explore the complex constructs of LP across multiple studies. The second stage of the study explores themes of LP from English translations of the Analects (17:8) by comparing different translations to understand meanings through highlighting the interpretative choices made in five significant translations from different time periods and cultural backgrounds. Findings The review reveals eight frameworks of LP from modern Chinese academic literature and codes them into three approaches of conceptualizing LP. The comparison of translations provide insightful interpretations of six key virtues related to the love of learning, namely, ren 仁 (benevolence), zhi 智 (wisdom), xìn 信 (faithfulness), zhi 直 (uprightness), yong 勇 (courage) and gang 剛 (impregnability) and six related consequences of a narrow vision in learning. Originality/value The findings of this study reveal valuable insights into themes of LP, namely, motivation to gain knowledge towards a goal, persistence towards the goal while maintaining rational actions, being sincere to learning and considering consequences, critical thinking with respect, interpersonal awareness and the building of wisdom by learning experiences. Two implications are discussed, particularly from educational philosophy and educational psychology perspectives. This paper provides insights for future research directions on the dimensions of LP in general, and from a Confucian perspective in specific.

Conflicting results exist in the literature on whether Open Educational Resources (OER) and Open Educational Practices (OEP) can improve learning performance. Additionally, limited studies, in this context, have attempted to systematically measure and understand this phenomenon. To address this research gap, this study conducts a two-level analysis based on a systematic review of the OER and OEP literature. It first conducts a meta-analysis to measure the effect of OER and OEP on learning performance. It then conducts a meta-synthesis based on the activity theory to understand what led to this effect (measured in the first phase). Specifically, 32 studies (N = 134905 participants) were quantitatively and qualitatively analyzed. The obtained results revealed that OER and OEP have significant negligible effect (g = 0.10; p < 0.05), which indicates that learners are mostly consumers of knowledge in a very traditional way. Additionally, it is found that the learning process was mainly in formal settings in classrooms using traditional technologies like websites and learning management systems. The findings of this study can help to enhance the effective adoption of OER and OEP by highlighting the confounding variables that should be considered when developing their open education initiatives.

Artificial Intelligence (AI) is transforming higher education rapidly by enabling personalized learning, enhancing administrative processes, and improving access to educational resources. However, disparities in AI adoption, particularly among women in the Asian context, raise concerns about equity, inclusivity, and access. This disparity could lead to a deficit in AI skills among women, affecting their ability to contribute as effectively as men in the future. Therefore, it is necessary to understand the current state of women's adoption of AI and the barriers they face in Asian higher education. The systematic review has been conducted using PRISMA guidelines. This review paper synthesizes the findings from the studies conducted in various contexts of Asia to present an overall picture of the state of AI adoption among women in Asia. A total of 17 studies were selected for this review, highlighting socio-cultural barriers, lack of trust, technological unawareness, biases in AI algorithms, and inadequate representation of women in AI policy formulation. Besides highlighting these barriers, the results also shed light on recommendations given by earlier studies that facilitate and encourage women to adopt AI in higher education. Based on the Asian perspective, the conclusion proposes specific recommendations for policymakers and practitioners to promote inclusive AI that empowers women in Asia to contribute more effectively to higher education.

Artificial Intelligence (AI) has rapidly emerged as a pivotal force in a multitude of sectors. This research explores the growing convergence of AI tools, such as ChatGPT, in the field of academic writing, with a focus on 136 postgraduate students from China. These students were randomized into groups that utilized AI-integrated tools in both online and offline environments. The study involved designing hybrid instructional modules that blend online and offline elements. Furthermore, an extensive questionnaire was used to evaluate the advancements of postgraduate students in domains such as digital readiness, writing self-efficacy, student engagement, and academic emotional health. The results from this investigation highlight significant improvements in several facets of postgraduate involvement, including behavioral, emotional, and cognitive engagement, as well as writing self-efficacy. The study also probes the impact of digital readiness on academic involvement, revealing a positive association with student engagement, writing self-efficacy, and academic sentiments. By augmenting students' technical proficiency, digital literacy, and acclimation to digital tools and platforms, educational institutions can bolster academic engagement, amplify writing confidence, and foster favorable academic emotion.

With the increased interest in Science of Learning (SoL) to enhance learning experiences and outcomes, several SoL definitions have been proposed from different fields and research backgrounds. This has created some ambiguity about what SoL is and how it can contribute to the educational field. To address this research issue, this study conducts a literature review of the different SoL definitions presented in the literature between 2000 and 2024. It then analyzes how these definitions evolved over time in terms of the mentioned fields, involved stakeholders, and cognitive practices. The obtained results revealed that most of the SoL definitions are rooted in the neuroscience field with an increased interest in the use of technology and Artificial Intelligence (AI) in the last few years. Particularly, it is found that some areas, such as the role of genes in learning, need more attention and were surprisingly not found in the definitions. Additionally, it is found that most of the stakeholders mentioned are generic (e.g., human, adult, etc.), depicting that learning can happen outside of schools and at any age. The findings of this study call for more cross-disciplinary collaboration on SoL implementation and impact, particularly in classroom and digital learning settings.

Despite the importance of gamification in education, there is still ongoing debate in the literature about how to design effective and useful educational gamification. This is because gamification is a complex concept that requires combining various game elements together. To further contribute to this discussion, this study first develops a gamified course, using seven game elements, where eighty-three university students were enrolled in the course during an entire semester (three months). It then builds on the complexity theory and fuzzy-set Qualitative Comparative Analysis (fsQCA) to investigate which combination of the seven game elements can explain students’ perceived usefulness of gamification in the given course. The findings revealed that no single game element leads to students perceiving gamification useful. Additionally, the evidence from this study suggests that there are ten solutions which can explain the students’ perceived usefulness of gamification in the course. Particularly, badge was the most present game element in these solutions, where it is found in eight among the ten solutions. The findings of this study can guide various stakeholders (e.g., educators, designers and developers) on how to create a useful educational gamification, hence enhancing learning experiences and outcomes.