人工智能算法歧视的法律规制:欧美经验与中国路径

Legal Regulation on Algorithmic Discrimination in AI:Euro-American Experience and Chinese Approach

算法歧视是人工智能自动化决策中,由数据分析导致的对特定群体的,系统的、可重复的、不公正对待。算法歧视正在人工智能技术应用的多个场域出现,对受害群体及整个社会有着多重不利影响。为了维护平等与公正的社会秩序,有必要对算法歧视进行法律规制。作为人工智能法律规制的先行者,在应对人工智能算法歧视方面,欧盟选择了以数据保护为中心的规制模式,而美国选择了以算法责任为中心的规制模式。我国正处于人工智能法律规制的草创阶段,在今后的法治实践中,可由算法理念、算法技术、算法审查和算法问责四个方面着手,完善我国的人工智能算法歧视法律规制框架。

Algorithmic discrimination is the systematic,repeatable,and unfair treatment to certain groups led by data analysis in the process of AI automated decision making.As pioneers of AI legal regulation,in coping with algorithmic discrimination,the EU has chosen a path centered on data protection,while the US prefers the mode focused on algorithmic accountability.China is currently at the initial stage of AI regulation.In the future,the regulatory framework of algorithmic discrimination in AI can be strengthened through aspects of algorithmic ethic,algorithmic technology,algorithmic audit and algorithmic accountability.