| 505 | 0 | 272 |
| 下载次数 | 被引频次 | 阅读次数 |
代理式人工智能(Agentic AI)的广泛应用标志着人工智能从“对话”到“行动”的范式转变,智能体成为驱动教学变革的重要行动者。该文针对国内外教学领域智能体应用的典型案例,归纳出其三类行动角色:作为教学代理或认知代理的知识型行动者、优化教学基础设施的组织型行动者和重构教学互动关系的关系型行动者。基于比斯塔教育哲学的教育功能理论框架,研究揭示出智能体驱动教学变革中的风险问题,即资格化的效能幻象、社会化的关系侵蚀与主体化的存在危机。最后,面向教学变革中政府、企业、学校三类关键主体,该文提出代理式人工智能时代的教育智能体风险治理策略,为我国推进“人工智能+教育”行动计划、建设教育强国提供参考。
Abstract：The widespread adoption of Agentic AI signals a paradigm shift in AI from “conversation” to “action”, with AI agents emerging as key actors in reshaping educational practices. Based on representative domestic and international application cases of AI agents in educational contexts, this study identifies three primary actor roles that AI agents assume in teaching scenarios: knowledge-oriented actors serving as teaching or cognitive agents, organizational actors that optimize educational infrastructure, and relational actors that reconstruct patterns of teaching interaction. Drawing on Biesta's framework of educational functions, this work examined the risks embedded in AI agent-driven teaching reform, including the illusion of effectiveness in qualification, the erosion of social relations in socialization, and the existential crisis of subjectification. Finally, it proposed risk governance strategies for AI agent, involving three core stakeholders in teaching transformation, i. e., government, enterprise, and schools. The findings provided theoretical and practical insights for advancing the “ AI+Education Action Plan” and building a strong educational system in China.
[1]European Schoolnet.Artificial Intelligence in School Education:An overview of policy priorities and initiatives across 23 education systems[R]. Brussels:European Schoolnet,2025.
[2]教科信[2026]1号,教育部等五部门关于印发《“人工智能+教育”行动计划》的通知[Z].
[3]李梦兴,刘妍.AI智能体能提升学习效果吗?——问题导向思维视角下学生深度问题解决能力发展效果研究[J].现代教育技术,2025,35(10):42-51.
[4]Hao Z, Cao J, et al. Mapping student-AI interaction dynamics in multi-agent learning environments:Supporting personalized learning and reducing performance gaps[J].Computers&Education,2026,241:105472.
[5]Bozkurt A.Artificial Intelligence and the Future of Personalized Education[M].Hershey:IGI Global,2026.
[6]Kremantzis M,Essien A,et al.Uncovering the Generative AI(GenAI)to Agentic AI(Ag AI)shift for business school education[J].Journal of Global Information Management,2025,33(1):1-21.
[7]Acharya D B,Kuppan K,et al.Agentic AI:Autonomous intelligence for complex goals—A comprehensive survey[J].IEEE Access,2025,13:18912-18936.
[8]Ren Y,Liu Y,et al.AI Agents and Agentic AI-navigating a plethora of concepts for future manufacturing[J].Journal of Manufacturing Systems,2025,83:126-133.
[9]Zhang S,Yu Z,et al.Role can be beneficial to mean field multi-agent reinforcement learning[J].Information Sciences,2025,721:122522.
[10]Kim Y, Baylor A L. Research-based design of pedagogical agent roles:A review,progress,and recommendations[J].International Journal of Artificial Intelligence in Education,2016,26(1):160-169.
[11]Kostopoulos G,Gkamas V,et al.Agentic AI in education:State of the art and future directions[J].IEEE Access,2025,13:177467-177491.
[12]卢国庆,徐秀莲等.协作学习中教育智能体的设计策略、角色功能与交互模式[J].电化教育研究,2026,47(1):67-74+83.
[13]王占魁.“教师代理”抑或“智能教具”——论人工智能在未来教学活动中的职能定位[J].教育发展研究,2026,46(2):18-25.
[14][15][17][22][23][24][荷]格特·比斯塔.张立平,韩亚菲译.测量时代的好教育:伦理、政治和民主的维度[M].北京:北京师范大学出版社,2019.
[16][26][27]格特·比斯塔,赵康.测量时代和人工智能时代的教育——基于主体化教育展开的对话[J].教育研究,2025,46(5):55-67.
[18]Fan Y,Tang L,et al.Beware of metacognitive laziness:Effects of generative artificial intelligence on learning motivation,processes,and performance[J].British Journal of Educational Technology,2025,56(2):489-530.
[19]王亚南.元认知的结构、功能与开发[J].南京师大学报(社会科学版),2004,(1):93-98.
[20]Ganguly A, Mehjabin N, et al. Conversational AI agents in education:An umbrella review of current utilization, challenges, and future directions for ethical and responsible use[J].AI and Ethics,2026,6(1):72.
[21]陈本皓.人机协同下的AI风险链:认知外包、能力空心化、深度思维僵化[J].广西大学学报(哲学社会科学版),2025,47(5):194-204.
[25]王小寒,石中英.教育中主体性困境与重塑:基于比斯塔“主体化”理论的分析[J].清华大学教育研究,2025,46(6):53-60.
[28]Cheng M, Lee C, et al. Sycophantic AI decreases prosocial intentions and promotes dependence[J].Science,2026,391(6792):eaec8352.
[29]Bozkurt A,Crompton H,et al.The devil is in the Det[ai] ls:AI agents,ghost students,and the crisis of verified presence in an agentic AI world[J].Open Praxis,2026,18(1):1-12.
[30]SIX P,LEAT D,et al.Towards holistic governance:the new reform agenda[M].New York:Palgrave,2002.
[31]赵晓伟,沈书生等.生成式人工智能何以赋能课堂变革——基于“认知进阶—技术支持”矩阵的融智课堂实践路径[J].教育研究,2025,46(12):40-53.
[32]章润宇,赵晓伟等.迈向人工智能大模型赋能的苏格拉底智能体:来自元分析的证据与启示[J].电化教育研究,2026,47(4):46-53.
[33]刘复兴,刘媛.数智时代教育理论变革的新维度[J].中国电化教育,2026,(1):54-59+68.
[34]Hava K,BabayiɡitÖ.Exploring the relationship between teachers’competencies in AI-TPACK and digital proficiency[J].Education and information technologies,2025,30(3):3491-3508.
[35]李晓静,胡柔嘉.我国中小学生数字技能测评框架构建与证实[J].中国电化教育,2020,(7):112-118.
[36]陈法宝,胡芳菲.经合组织和欧盟联合发布中小学教育AI素养框架草案[N].中国教育报,2025-06-26(09).
[37]中华人民共和国中央人民政府.中共中央国务院印发《教育强国建设规划纲要(2024—2035年)》[EB/OL].https://www.gov.cn/gongbao/2025/issue＿11846/202502/content＿7002799.html,2025-02-10.
基本信息:
中图分类号:G434
引用信息:
[1]李晓静,宋兆锟.智能体驱动教学变革的行动角色与风险治理[J].中国电化教育,2026,No.475(08):79-86.
基金信息:
国家社会科学基金重大项目“智能媒体对新闻真实性的挑战与治理研究”（项目编号：23&ZD213）阶段性研究成果
2026-08-06
2026-08-06