远程教育杂志2025，Vol.43Issue(5)：20-31,112,13.DOI:10.15881/j.cnki.cn33-1304/g4.2025.05.003
从多智能体到人智学习型组织
From Multi-Agent Systems to Human-AI Learning Organizations—Learning Activities in Human-AI Communities Supported by Multi-Role Agents
摘要
Abstract
Learning communities are central to building individual and collective knowledge.However,current uses of multi-agent systems with learner groups have not yet produced true learning communities.The underlying mechanisms for such collaboration remain unclear,and common multi-agent frameworks do not meet the learning needs of human-AI communities.Drawing on learning-organization theory,this study examines how a human-AI learning community,involving teachers,learners,and software agents,forms.It designs and implements a learning system in which multi-agent collaboration supports community activities,and proposes an activ-ity model,from personal vision,human-AI dialogue,shared vision to metanoia.A quasi-experimental study in a Digital Instructional Design course was conducted.Compared with groups supported by a single agent,the human-AI learning community achieved higher learning outcomes,improved critical and creative thinking,strengthened collaboration skills,increased participation in conversations and social-network interactions,and supported intermediate-to-advanced levels of collaborative knowledge building.Future work with multi-agent systems in education should target higher-order cognitive goals,support critical and innovative thinking and deep learn-ing,and explore smart-and-digital learning environment in which multi-agent systems are embedded in online learning communities to build robust human-AI learning communities.关键词
多智能体/学习共同体/学习型组织/大语言模型/人机协同Key words
Multi-agent systems/Learning community/Learning organization/Large language model/Human-AI collaboration分类
社会科学引用本文复制引用
李海峰,王炜..从多智能体到人智学习型组织[J].远程教育杂志,2025,43(5):20-31,112,13.基金项目
本文系2024年度国家社会科学基金教育学西部项目"大语言模型会话机器人支持的西北边疆乡村教师高质量教研研究"(项目编号:XCA240326)的研究成果. （项目编号:XCA240326）
