Due to the deep penetration of Generative artificial intelligence(GenAI),the cultivation of pre-service teachers'teaching practice reflection competence is confronted with risks of superficial reflection and excessive reliance.Providing learning scaffolds constitutes one of the effective means to address these risks.Based on this,relying on social comparison theory,this paper designs GenAI-supported structural reflective scaffolds and explores its impact on pre-service teachers'teaching practice reflection competence through a randomized controlled experiment.Adopting methods such as questionnaire surveys,LDA topic analysis,and epistemic network analysis,this paper conducts analyses of inter-group differences in reflective thinking,quality differences in human-machine dialogues,and inter-group differences in reflection effectiveness.The findings reveal that the collaborative mechanism between intelligent technology and reflective scaffolds jointly facilitates the improvement of reflective thinking;GenAI-supported structural reflective scaffolds serve as a key factor influencing pre-service teachers'teaching practice reflection effect by standardizing and deepening the cognitive process of human-machine dialogues and improving real-time interaction quality.This paper reveals the necessity and feasibility of collaborative design between technology empowerment and scaffold support in pre-service teacher education,and provides empirical reference for cultivating pre-service teachers'teaching practice reflection competence in the intelligent era.
Unlike traditional learning platform with the core functions of resource organization,teaching interaction,and learning process support,as well as AI-augmented learning platform that merely introduce AI plugins into partial functions,AI-native learning platform emphasize the deep integration of artificial intelligence technologies into platform architecture,business workflows,and user interactions.Facing the in-depth demands that the development of personalized education places on learning platform,this paper first proposes the concept of an"AI-native learning platform"and designs a three-layer architecture consisting of the bottom layer,middle layer,and top layer based on the fundamental reshaping of the development paradigm.With large language model serving as the cognitive hub,this architecture can enable dynamic organization and intelligent scheduling of system services,providing a new idea for the intelligent,personalized and sustainable evolution of learning platform.Then,taking University G in North China as an example,this paper elaborates on the transformation pathway for existing platforms evolve toward AI-native one,including the reconstruction of the underlying data system,the middle-layer middle platform,and top-layer interaction,as well as newly-added modules for AI-native learning platform.The engineering feasibility of the architecture is validated through the construction of the eLLM training platform.Finally,this paper discusses the problems encountered during the deployment and application of AI-native learning platform and puts forwards targeted directions for future research.The architecture of AI-native learning platform proposed in this paper provides a systematic design idea and implementation pathway for the construction of smart educational infrastructure,and can serve as a reference for the application of AI agent in large-scale educational scenarios.
The disconnection between learning content and the real world constitutes one of the main problems in current education,and it also serves as an important reason influencing students'cognitive development.Grounded from the perspective of genetic epistemology,this paper interprets the fundamental connotation of learning scenarios in supporting cognitive development and analyzes the value implications of digital-intelligent technology empowering learning scenarios starting from the evolutionary logic of digital-intelligent technology.Based on four key dimensions of learning scenarios including boundary,representation,interaction,evaluation,combined with four advanced levels including cognitive activation,cognitive construction,cognitive transfer,cognitive improvement,centering on how to construct learning scenarios based on genetic epistemology mechanism,how to highlight the empowering effect of learning scenarios on cognitive advancement,and how to realize the empowerment of learning scenarios for cognitive advancement practice,the basic framework of digital-intelligence technology empowering learning scenarios is established.Finally,this paper proposes specific practical pathways from the perspectives of embodied field construction,precise resource supply,human-AI collaborative interaction,and data-driven evaluation,so as to provide theoretical basis and path guidance for overcoming application misconceptions and informing teaching practices.
Human-machine collaborative teaching has become an inevitable trend in educational reform in the intelligent era,but existing literature lacks evaluation research on primary and secondary school teachers'human-machine collaborative teaching ability.Accordingly,following the research idea of"concept definition—system construction—scale development and revision application",this paper first elucidates the connotation of human-machine collaborative teaching ability.Next,the evaluation system for teachers'human-machine collaborative teaching ability is preliminary constructed.Two rounds of expert opinions consultation via the Delphi method are conducted to revise the evaluation indicators,and the evaluation system for teachers'human-machine collaborative teaching ability including 4 first-level indicators,11 second-level indicators,and 37 third-level indicators is ultimate formed.The analytic hierarchy process and expert ranking methods are adopted to determine the weight of each indicator.Subsequently,combined with actual teaching scenarios in primary and secondary schools,this paper develops the assessment tool named"assessment scale for primary and secondary school teachers'human-machine collaborative teaching ability".The scale is revised through Rasch model test and its secondary test.Results analyzing the ability differences among primary and secondary school teachers reveal that their overall human-machine collaborative teaching ability is at a lower-than-intermediate level.In response,this paper recommends that education administrative departments conduct precise diagnosis and implement stratified training to improve primary and secondary school teachers'human-machine collaborative teaching ability.The evaluation system for teachers'human-machine collaborative teaching ability constructed in this paper can provide a structured reference for assessing teachers'professional ability in the intelligent era,and the developed assessment tool can provide reliable support for teachers'professional development and educational administration decision-making.
To advance the construction of smart education demonstration zones,it is urgent to establish a scientific and effective policy support system,for which precise quantitative evaluation of current policy schemes serves as an important prerequisite for driving policy optimization and dynamic adjustment.Based on this,the paper first proposes an analytical framework of policy supply for smart education demonstration zones,and then constructs an evaluation index system for such policy supply.Taking policy documents released by 18 national smart education demonstration zones as research objects,this paper adopts the LDA topic model and PMC index model to quantitatively evaluate policy attention allocation and policy supply characteristics,and analyzes policy implementation dilemmas of smart education demonstration zones.The results show that there are differences in policy issue selection preference,policy attention orientation,and policy attention spatio-temporal allocation among different demonstration zones.The policy supply for smart education demonstration zones shows high consistency in overall design and core-element selection,forming three types of demonstration zones include the"resource-talent"oriented type,"technology-service"oriented type,and"security-service"oriented type.In terms of specific policy implementation,each demonstration zones face multi-faceted dilemmas.Finally,from three paths of problem-driven approach,process control,and pressure response,this paper proposes dynamic adjustment strategies for policy supply of smart education demonstration zones,with a view to providing decision-making references for optimizing policy supply quality and promoting the realization of the national strategic goal of"driving overall progress through exemplary pilot sites".
With its powerful generative capabilities,generative artificial intelligence(GenAI)provides new momentum for educational innovation and transformation,yet it also gives rise to the educational information cocoons effect,exacerbating information narrowing and triggering information bias.Therefore,focusing on the GenAI-empowered collaborative learning scenarios,this paper clarifies the connotation and characteristics of the educational information cocoons effect in the GenAI era,and further explores its generation process,action mechanisms,and breakthrough pathways.The paper points out that the generation of the educational information cocoons effect in the GenAI era undergoes three stages of cognitive assimilation,generative solidification,and behavioral reinforcement.Furthermore,under the joint interaction of four elements including learners,learning platforms,learning resources,and collaborative environments,this effect triggers a series of problems,such as the standardization of generative collaborative content,the homogenization of generative collaborative groups,and the limitation of generative individual choices.To address these issues,this paper proposes the breakthrough pathways of educational information cocoons effect with the"empowerment-promotion-optimization-driving"of GenAI:empowering environmental reshaping to break the information closed-loop reinforced by group collaboration;advancing technological upgrading to eliminate the technical loop solidified by algorithmic bias;optimizing profile adjustments to interrupt the cognitive loop reinforced by behavioral preferences;and driving content generation to break the supply loop of homogenized generated content.The analysis of the educational information cocoons effect in this paper can provide theoretical references for optimizing the intelligent dissemination of educational information.
While generative artificial intelligence(GenAI)empowers higher education practice,it also induces students'cognitive evasion by dispeling desirable difficulties.Confronted with instructional intervention,such"cognitive evasion"can easily escalate into"feigned compliance",which can hardly be identified and curbed by traditional outcome-oriented assessment.Accordingly,taking the"BROKE Incident"as its starting point,this paper adopts action research,designs the analytical framework for ecological intervention,and implements three rounds of progressive ecological interventions.The findings reveal that cognitive evasion is essentially learners'abdication of deep cognitive processing,which can easily evolve into feigned compliance under supervisory pressure.Ecological intervention with three-dimensional collaboration of value,mechanism and power can reduce the benefit space for cognitive evasion and feigned compliance,thereby restoring conditions for desirable difficulties to take place.Under ecological intervention,three typical reaction patterns emerge:patterned transformation,strategic fixation,and compliance adaptation.Based on this,four pathways are proposed to address cognitive evasion induced by GenAI effectively:clarifying cognitive boundaries of human-machine collaboration,preserving process evidence of key nodes,employing diverse questioning,and engaging in reflective revision of works.The research of this paper helps sustain desirable difficulties and foster deep learning in the AI era,also can provide references for bounded human-machine collaborative learning.
Currently,the in-depth advancement of digital transformation and the continuous expansion of university asset scale have made the traditional asset management models hardly adapt to the actual needs of refined asset management in universities.In this context,the paper first analyzes the current situation and pain points of university asset management,and elucidates the advantages of digital twin technology in empowering asset management.Subsequently,this paper defines the principles to be followed in the process of digital twin technology in empowering the full-chain asset management in universities,and proposes the pathways of digital twin technology in empowering the full-chain asset management from four aspects of business processes,data structures,electronic signatures,and regulatory models.Finally,practical cases are presented to demonstrate the efficacy of digital twin technology in empowering the full-chain asset management in universities:greatly improving the efficiency of asset management business processing in universities,completing the reengineering of full-chain business processes for university asset,and realizing full-process and traceable data security management and control.The research of this paper promotes the deep integration of physical asset space and digital information space,and can provide theoretical guidance and referable practical implementation schemes for the digital transformation of asset management in universities.
Art therapy serves as an important rehabilitation approach for children with autism.However,traditional therapy models suffer from limitations such as insufficient personalization and low active participation of children,resulting in limited intervention effects.Artificial intelligence(AI),with its powerful data application capability and diverse interactive advantages,can effectively make up for the shortcomings of traditional therapy models,yet its mechanism in art therapy for children with autism remains unclear.Accordingly,this paper takes special-education teachers,experts,children with autism and their parents as research subjects,adopts the combining methods of questionnaire surveys,semi-structured interviews and behavioral observation,and conducts baseline regression analysis,mediation effect tests,moderation effect tests,and chain mediation effect test.The results show that AI(including AI data application capability and AI interaction model)can effectively improve the art therapy effect of children with autism.The fitness of personalized schemes and the participation of children exert mediating effect,while children's age and children's symptom degree exert moderating effects.AI empowers art therapy for children with autism through dual pathways,namely the data-adaptation pathway of"AI data application capability → fitness of personalized schemes",and the interactive engagement pathway of"AI interaction model → children's engagement".Combined with this conclusion,suggestions are put forward for the application of AI in art therapy.The research of this paper reveals the transmission mechanism and boundary conditions of AI-empowered art therapy,deepens the understanding of the applicability of human-machine synergy theory in special education scenarios,and can provide theoretical guidance for optimizing the practice of intelligent art therapy for children with autism.
Intangible cultural heritage(ICH)constitutes a vital carrier of fine traditional Chinese culture and serves as key content for fostering cultural understanding within international Chinese education.However,the accuracy and applicability of existing large language model(LLM)in intelligent question-answering(Q&A)of ICH still need to be improved in international Chinese education scenarios.Therefore,this paper designs an intelligent Q&A system facing ICH in international Chinese education,and proposes a lightweight pathway for the coupling of knowledge graph and LLM.Firstly,taking three representative national-level ICH items as research subjects,entities are extracted and linguistically adjusted,and Protégé is adopted to construct the knowledge graph,which is then imported into Neo4j to form the graph database.Subsequently,natural language questions are mapped onto structured retrieval through Schema constraints,Few-shot examples and Cypher queries,so as to realize the coupling of knowledge graph and LLM.Finally,the quality evaluation of answers generated by the intelligent Q&A system is conducted through both objective indicators of knowledge coverage rate and subjective scores from teachers and students'questionnaires,in order to examine the performance differences between this system and conventional LLM.Reseach results indicate that the intelligent Q&A system significantly outperforms the conventional LLM in terms of knowledge coverage,accuracy,and applicability of answers,and shows particularly prominent performance in generation of factual knowledge and procedural knowledge.The lightweight pathway proposed in this paper substantially improves the accuracy and teaching applicability of ICH content generated by intelligent Q&A systems in international Chinese education,and provides an operational practical paradigm for the effective dissemination of ICH.
The vigorous rise of large language models represented by DeepSeek and ChatGPT not only highlights the strategic value of original innovation capability,but also imposes higher demands on education and teaching for cultivating students'original innovation ability.As fundamental support for developing students'original innovation capability,big ideas have been widely recognized for their instructional value,yet prominent bottlenecks persist in their practical implementation.Accordingly,this paper conducts research following the logical thread of"background analysis—connotation retrospection—source divergence—theoretical reflection—route reconstruction":First,grounded in the era background of cultivating original innovation capability,it analyzes the intrinsic connection between big ideas and original innovation capability.Then,it traces the teaching dilemmas of big ideas and sorts out the divergence in connotations between big ideas viewed as knowledge and those viewed as thinking.Subsequently,starting from divergence in the sources and extraction methods of big ideas,this paper explores the logical starting points for generating big ideas under the guidance of different ideas.After that,it carries out systematic reflections on the scientificity and integrity of big ideas and their teaching research,revealing that the root cause of unsatisfactory teaching effects of big ideas lies in the artificial separation of knowledge acquisition and thinking development.Finally,from the basic stance of"the inherent unity of knowledge and thinking",this paper carries out the route reconstruction for big ideas and their teaching research.This paper proposes an instructional design idea centered on conceptual understanding and driven by essential questions,which can provide theoretical reference for resolving connotation divergence and overcoming practical obstacles in current big ideas teaching,and offer strong support for the development of students'original innovation capability under the background of original innovation.
With the rapid development of generative artificial intelligence(GenAI),multi-agents that integrate large language models with virtual digital humans have gradually evolved into the important collaborative entities for human-machine collaborative learning.However,the intrinsic mechanisms for how learners and multi-agents carry out effective collaboration remains unclear.Accordingly,this paper conducts an experiment on multi-agents human-machine collaborative learning in the"German Speech"course.Thirty German language learners are recruited from Z University to complete two types of human-machine collaborative learning tasks involving declarative knowledge and procedural knowledge,and their electroencephalogram(EEG)signals and human-machine dialogue text data are collected simultaneously.Through EEG analysis and latent dirichlet allocation(LDA)topicanalysis,this paper finds that the main effect of brain regions is significant,and brain-wave power of each frequency band shows certain regional differences.The human-machine dialogue tests are stably clustered into two or three topics:declarative knowledge learning focuses on knowledge interpretation and value judgment,while procedural knowledge learning involves more comparative analysis,meaning construction,and integrated application.Based on this conclusion,this paper constructs an alignment framework for human-machine collaborative cognition and EEG supported by multi-agents,and reveals the cognitive processing mechanism of human-machine collaborative learning:multi-agents-dominated external perception at the information input phase,human-machine primary-auxiliary collaborative meaning construction at the content-comprehension phase,human-machine-jointly dominated information integration at the structural-organization phase,and learner-dominated metacognitive activation at the regulation-reflection phase.Integrating implicit EEG signals with explicit dialogue texts,this paper reveals the cognitive processing characteristics and human-machine role division of human-machine collaborative learning supported by multi-agents at different stages,and can provide references for the development of multi-agent system and the design of human-machine collaborative teaching.
Artificial Intelligence Generated Digital Educational Resources(AIGDER)represent a novel resource form that leverages generative foundation models to meet user needs.Single-agent approaches suffer from weak planning and narrow perspectives,while multi-agent systems face challenges in educational contexts including human-machine intention misalignment,coordination conflicts,and unstable output quality.To address this,this paper introduces Shared Mental Models(SMM)theory,mapping its two-dimensional structure onto multi-agent technical elements to construct a multi-agent-supported AIGDER development model.Built on a core workflow of demand analysis,resource generation,application evaluation,and resource optimization,the model adopts SMM's three-stage evolution of initial formation,consensus adjustment,and fluency as its operational mechanism,driving cognitive alignment and coordination norm consistency to form a human-machine collaborative development loop.Implementation safeguard strategies are further proposed across four dimensions of technical architecture,semantic states,process control,and quality optimization to enhance stability,controllability,and sustainable evolution,offering reference for both theoretical research and practical application in AIGDER development.
With the advancement of large language models'capabilities in generation,reasoning,and behavioral simulation,the constraints that have long hindered traditional education empirical research centered on real individuals,including difficulty in sample acquisition,high experimental costs,and significant ethical risks,are encountering new breakthroughs.As quasi-real,annotatable,and controllable synthetic samples,educational virtual samples driven by large language models are emerging as a new methodological for empowering educational research.Building upon a review of the research trajectory of virtual samples,this paper systematically analyzed the theoretical logic of virtual samples in education by integrating the perspective of human-machine symbiosis,actor-network theory,and information processing theory.It further constructed a closed-loop pathway encompassing five stages of setting,memory,planning,action,and regulation-feedback,and explored its boundary expansion in survey research and experimental research.The study concluded that educational virtual samples can effectively simulate learners'cognitive responses and behavioral processes,support pre-verification of research,sample supplementation,and scenario extapolation.However,their application should adhere to the principles of authenticity,ethics,and boundary,so as to avoid substituting technological verisimilitude for educational reality.
As the fundamental resource and carrier for instructional activities in digital contexts, digital textbooks have yielded inconsistent conclusions regarding their effectiveness in existing research. To address this gap, the paper conducted a meta-analysis of 35 global experimental and quasi-experimental studies focusing on basic education. The results demonstrated that digital textbooks had a moderate positive effect on learnersu2019 cognitive performance, skill performance, and affective attitudes. Such effects were moderated by factors such as instructional model, intervention duration, educational stage, and subject area, while textbook type showed no moderating influence. In this paper, it was proposed that the development of digital textbooks should strengthen instructional embedding and disciplinary adaptability, establish an interdisciplinary collaborative textbook research and development mechanism, optimize the role positioning of teachers and students, and form a regular classroom application model for digital textbooks. These measures can facilitate the deep integration of technology and pedagogy, and boost the high-quality development of education.
The classroom atmosphere is an important factor affecting classroom learning experience,participation and teaching effectiveness.Accurate recognition of classroom atmosphere is vital for optimizing classroom teaching and promoting cognitive development.However,current classroom atmosphere recognition suffers from limitations such as one-sided representational dimensions and single data modality,which makes the existing research results difficult to directly serve classroom teaching practice.Therefore,a bimodal classroom atmosphere recognition method based on the classroom assessment scoring system was proposed in this paper:firstly,label according to classroom atmosphere indicators;then,use temporal-aware bi-directional multi-scale network integrated with multi-head self-attention and the video sliding window transformer model to conduct single-modal classroom atmosphere recognition based on audio data and based on video data,respectively;finally,adopt the random forest algorithm for dual-modal classroom atmosphere recognition to determine the classroom atmosphere level.Through a series of comparative experiments,this paper found out that the performance of classroom atmosphere recognition by integrating bimodal data was superior to that of single-modal data,and the disciplinary characteristic was an important influencing factor for classroom atmosphere recognition,and concluded that a multi-dimensional structure can more precisely represent classroom atmosphere,and bimodal data support can more accurately predict classroom atmosphere.The research in this paper can provide effective technical support for precise teaching diagnosis,and hold significant value for optimizing classroom teaching practice.
During the process of promoting educational digitalization,digital resources exert both positive and negative effects on primary school students'attention,and resource adaptation is the key to practical implementation.Grounded in the attentional resource theory and cognitive load theory,this paper first proposed a three-dimensional analysis framework of"technology-environment-individual"that affected the attention of primary school students by integrating factors influencing students'attention.Subsequently,utilizing digital resources from the National Smart Education Platform for Primary and Secondary Schools,a questionnaire survey was conducted to explore the impacts of digital resources on primary school students'attention.Through the analysis of the questionnaire data,it was found that the interactivity and multimedia characteristics of digital resources served as core driving factors affecting primary school students'attention;environmental support played a mediating role between technical characteristics and attention;and notable grade differences existed,with lower-grade students more vulnerable to information overload.On this basis,the paper designed an impact model of digital resources on primary school students'attention.It indicated that the dual effects of technical characteristics relied on cognitive load regulation,and the mediating effect of environmental support affected the allocation of attentional resources,and individual differences were the regulatory factors for the efficiency of attentional resources.Relying on this influence model,the paper revealed the dual influence mechanism of digital resources on primary school students'attention:digital resources had a significant positive promotion and negative interference dual effects on primary school students'attention,and the final net effect depended on the dynamic balance of technical characteristics,environment support and individual differences.This research of the paper provided empirical evidence for the optimized design of smart educational resources and the guidance for primary school students'attention,and helped improve the quality and efficiency of digital teaching.
Collaborative learning contributes to cultivating students'higher-order thinking,yet it often encounters practical dilemmas such as insufficient collaborative contexts and low-quality interaction.The integration of immersive virtual reality(IVR)and twin digital human technology makes it feasible to create an immersive and embodied environment for collaborative learning.However,the collaborative learning activities in related research mostly take place outside virtual spaces,which difficult to fully reflect the actual impact of IVR environment on collaborative learning.Based on this,the paper constructed a theoretical framework for Twin Digital Humans Empowered Collaborative Learning(THCL)in the IVR environment and used this framework as a guide to carry out a virtual simulation experiment on"family fire escape drill".The experiment took 40 graduate students from X University in Chongqing as the research objects,who were randomly assigned to either the THCL group or the Virtual Reality Empowered Face-to-Face Collaborative Learning(VRCL)group to engage in collaborative learning through different approaches.Through pre-test questionnaire analysis,learning outcome questionnaire analysis,and correlation analysis,this paper found that THCL demonstrated better performance in students'procedural knowledge transfer and emotional experiences,and presented a lower level of cognitive load.In addition,lower cognitive load and positive emotional experience jointly constituted important conditions for promoting knowledge transfer.This paper,through the research,revealed the mechanism of collaborative learning in IVR environment and provided empirical evidence for the joint empowerment of educational practice by IVR and twin digital human technology.
In the intelligent era, education is evolving toward personalization, adaptability, and interdisciplinary integration. However, current talent cultivation in colleges and universities is faced with the dilemma of u201Clarge quantity but not high qualityu201D, and the reform of cultivation model is constrained by the triple challenges of u201Cbreaking barriers, improving quality, and reducing unnecessary competitionu201D, making it difficult to meet the demands for high-quality talents in the intelligent era. As an innovative conception of the underlying cognitive logic, u03A9-type thinking is a new cognitive paradigm characterized by integration, openness, and inquiry. It emphasizes realizing systematic knowledge generation through an integrated cognitive architecture, creating interdisciplinary innovation opportunities via open cognitive boundaries, and empowering autonomous development and lifelong growth by an inquiry-based cognitive cycle. The cultivation of u03A9-type thinking can be implemented through four operational practical pathways, namely, reshaping the curriculum system with a modular course supermarket as the framework, reconstructing the teaching model via human-machine collaborative metacognitive practice, rebuilding the learning ecology with ubiquitous collaborative creative fields, and reshaping the evaluation system based on dynamically certified literacy development. The cultivation process is highly consistent with the logic of high-quality talent cultivation, and it is the core starting point for improving the quality and efficiency of talent cultivation. The introduction of u03A9-type thinking and the exploration of its cultivation pathways can provide cognitive support and practical solutions for higher education to achieve the leap from u201Cquantityu201D to u201Cqualityu201D in talent cultivation, promote the virtuous cycle of education, science and technology, and talents, and lay a consolidated talent foundation for the development of new quality productive forces.
Competency-oriented assessment reform in basic education must overcome the traditional emphasis on knowledge over competencies and outcomes over processes.Large language model-driven educational assessment agents offer new possibilities for multidimensional content,precise feedback,and real-time assessment,yet their development and validation in authentic multidimensional scenarios remain limited.This study developed Co-Quiz for five scenarios:assessment content generation and evaluation,knowledge and skill diagnosis,competency assessment,classroom emotion analysis,and classroom behavior analysis.It designs a multi-level collaborative functional system,formulate a configuration and development scheme,and guide sustained teacher-student-agent interaction,and implemented Co-Quiz on a low-code platform.An application strategy was then proposed and evaluated through expert ratings,questionnaires,and semi-structured interviews.Results showed high expert ratings,relatively high teacher and student satisfaction,and clear advantages in assessment effectiveness and learning support.This study offers practical guidance for developing,applying,and evaluating educational assessment agents and advancing the intelligent transformation of educational assessment.
