机器学习决策中就业歧视的生成与法治进路

The Generation of Employment Discrimination in Machine Learning Decision-making and Its Legal Approach

在线阅读下载PDF

导出

摘要机器学习决策在用工匹配领域发挥着日益重要的作用,而中国关于其在就业歧视方面的立法还未展开。结合算法技术原理和算法应用实践对自动化决策进行风险分析,就业歧视可能沿着“机器学习”这一技术路径嵌入算法决策,具体包括:机器学习“无意识”的就业歧视源头和决策者“有意识”的就业歧视手段,其将造成被决策者举证艰难的就业歧视结果。面对“机器学习决策”中愈加隐蔽、间接、复杂的就业歧视,现有法律法规在适用中面临困境,需要进行适应性改进,具体包括:劳动法中“就业歧视范畴”的完善与明确;个人信息保护法中“自动化决策解释”的软解释路径;个人信息保护法中“自动化决策拒绝权”的范畴细化;工会法中“技术改造监督”的制度接轨;算法管理规定中“算法备案”的拓展引申。 Machine learning decisions are playing an increasingly important role in the field of employment matching,but China has not yet initiated legislation on their potential for employment discrimination.By combining the technical principles of algorithm technology and the practical application of algorithms to conduct risk analysis on automated decision-making,employment discrimination may be embedded in algorithmic decisions along the"machine learning"technical path,specifically including:the"unconscious"source of employment discrimination in machine learning and the"conscious"means of employment discrimination by decision-makers,which will result in difficult-to-prove employment discrimination for the decision subjects.In the face of increasingly concealed,indirect,and complex employment discrimination in"machine learning decisions",the existing laws and regulations are encountering difficulties in application and need to be adaptively improved,specifically including:the improvement and clarification of the"scope of employment discrimination"in labor law;adopting a soft interpretation path for"explanations of automated decisions"in the Personal Information Protection Law;refining the scope of"the right to refuse automated decisions"in the Personal Information Protection Law;aligning the"supervision of technological transformation"in the Trade Union Law with the current situation;and extending the"algorithm filing"provisions in algorithm management regulations.

作者罗熠琛 LUO Yichen(School of Law,Wuhan University,Wuhan 430072,China)

机构地区武汉大学法学院

出处《北京工业大学学报(社会科学版)》北大核心 2025年第5期108-122,共15页 Journal of Beijing University of Technology (Social Sciences Edition)

基金国家社会科学基金重大项目(24&ZD112) 研究阐释党的二十届三中全会精神国家社会科学基金重大专项(24ZDA025)。

关键词机器学习自动化决策就业歧视训练数据算法解释 machine learning automated decision-making employment discrimination training data algorithmic explanation

分类号 D922.5 [政治法律—民商法学]

作者简介罗熠琛(1999-),男,武汉大学法学院博士研究生。