构建人工智能教育应用风险防范“金字塔”——欧盟《人工智能法案》解读  

Building a “Pyramid” for Risk Prevention in AI Education Applications:Interpretation of the EU AI Act

作　　者：周丹 苏福根[2] 龚超 袁中果[4] Dan ZHOU;Fugen SU;Chao GONG;Zhongguo YUAN(Institute of Education,Tsinghua University,Beijing 100084;Research Division,Center for Education Management Information of the Ministry of Education,Beijing 100816;Research Center for Japanese Studies,Tsinghua University,Beijing 100084;RDFZ Chaoyang School,Beijing 100028)

机构地区：[1]清华大学教育研究院,北京100084 [2]教育部教育管理信息中心研究处,北京100816 [3]清华大学日本研究中心,北京100084 [4]中国人民大学附属中学朝阳学校,北京100028

出　　处：《中国教育信息化》2025年第2期79-88,共10页Chinese Journal of ICT in Education

基　　金：2024年度国家社会科学基金教育学一般项目“数智环境下基于大语言模型的个性化学习设计与评价研究”(编号:BCA240055);教育部教育管理信息中心国外教育研究2023年度委托课题“国外人工智能教育战略研究及其数据库建设”(编号:MOE-CIEM-20230017)。

摘　　要：欧盟发布的《人工智能法案》提出采用分级管控、全生命周期监管等措施提高人工智能的安全性。基于该法案提出的风险监管理念,结合全球人工智能教育应用的风险防范案例,从以下几个方面分析了面对风险时可采取的防范对策:保护受教育者的基本权益,避免算法歧视;保护师生个人隐私数据,杜绝数据泄露;确保教育应用的透明性和可解释性,防范信息污染。最后,建议适时出台人工智能教育应用的分类分级指导政策,构建分类分级与风险防范并行的监管制度;高度重视师生隐私信息保护,加强师生人工智能素养教育,积极构建教育可信数据空间;在人工智能教育产品开发的过程中,应遵从人工智能应用风险防范和伦理规范,确保所开发的教育产品符合伦理准则和价值观。The EU's AI Act proposes such measures as hierarchical control and full lifecycle supervision to improve the security of AI.Based on the risk regulation concept proposed in the bill and combined with risk prevention cases of global AI education applications,this article analyzes the preventive measures that can be taken when facing risks from the following aspects:Protect the basic rights and interests of learners and avoid algorithmic discrimination;Protect personal privacy data of teachers and students,and prevent data leakage;Ensure transparency and interpretability of educational applications,and prevent information pollution.Finally,it is recommended to timely introduce classification and grading guidance policies for AI education applications,and establish a regulatory system that combines classification and grading with risk prevention.We attach great importance to the protection of teachers' and students' privacy information,strengthen the education of teachers' and students' AI literacy,and actively build a trustworthy data space for education.In the development process of AI education products,it is necessary to follow the risk prevention and ethical norms of AI applications,ensuring that the developed educational products comply with ethical standards and values.

关 键 词：欧盟 人工智能法案 人工智能教育 风险防范 可信数据空间 

分 类 号：G434[文化科学—教育学]