【作 者】：
沈 苑， 汪 琼
【关 键 词 】：
人工智能教育应用； 人工智能伦理； 教育包容； 算法偏见； 数据偏见
【栏 目】：
理论探讨
【中文摘要】：
在倡导教育包容与公平的时代背景下，教育面临着以个性化、自适应为特征的智能技术所带来的偏见风险。在智能教育产品的研发过程中，潜在的设计偏见、数据偏见和算法偏见会使教育领域面临多重风险，如教育活动的公平性受阻、教育系统内部的歧视加剧、教育中的多元化特质消解和学生思想窄化。学校、科技公司、监管机构等组织应当携手共进，提前侦测偏见风险并加强治理，包括：提升研发团队的教育理解和多元眼光，让师生成为产品的合作创造者，加强偏见检测和道德评估过程，建立可解释、可审查、可纠正的系统优化机制，开展面向师生的人工智能素养教育，确立人工智能在教育中的应用范围和伦理规范，从而争取实现人机良性互动，打造多元包容的智能教育环境。
【英文摘要】：
In the context of advocating the inclusiveness and equity in education, education now faces the risk of bias raised by intelligent technologies characterized by individualization and self-adaptation. In the development of intelligent educational products, the potential design bias, data bias and algorithmic bias can expose the education sector to multiple risks, such as the hindrance of equity in educational activities, the increase of discrimination within the education system, the dissipation of the diversity in education and the narrowing of students' thoughts. Schools, technology companies, regulatory agencies and other organizations should work together to detect the risks of bias in advance and strengthen governance, including improving the educational understanding and diverse perspectives of research and development teams, and enabling teachers and students to become the co-creators of products, strengthening the process of bias detection and ethical assessment, and establishing systematic optimization mechanisms that can be explained, reviewed and corrected, carrying out AI literacy education for teachers and students, and establishing the scope and ethical norms of AI application in education, thus striving to achieve human-machine benign interaction and to create a diverse and inclusive intelligent education environment.
【下 载】：
人工智能教育应用的偏见风险分析与治理