算法自动化决策的问题检视与制度形塑

Problem inspection and institutional shaping of algorithmic automatic decision-making

算法自动化决策存在算法偏见与泄露信息等风险,侵犯用户的平等权和隐私权,在具体场景中可能制造信息茧房,侵犯用户知情权,在实质上剥夺其自由选择权。《中华人民共和国个人信息保护法》虽然对其进行了一定规制,但其规制的对象、侵权判断标准、救济路径等仍存在模糊的地方。结合香港地区和欧盟等地的相关规定,应当明确规制的对象为服务型自动化决策,规制的内容为自动化决策结果的公平公正,重大影响的判断则应当综合分析。在数据收集方面,应当引入必要性原则,同时赋予数据主体解释权与脱离自动化决策权。

Algorithmic automatic decision-making has some risks such as algorithmic bias and information leakage, which violates the users′ right to equality and privacy.In the specific scenarios, it may create information cocoons, infringe on users′ right to know, and deprive them of their free choice in essence.Although the Personal Information Protection Law of the People′s Republic of China has regulated it to a certain extent, there are still ambiguities in the object of its regulation, the criteria for judging infringement, and the path of relief.Combined with the relevant regulations of Hong Kong region and the European Union, it should be clear that the object of regulation is service-oriented automatic decision-making, the content of regulation is the fairness and impartiality of automatic decision-making results, and the judgment of major impact should be comprehensively analyzed.In terms of data collection, the principle of necessity should be introduced, and the data subject should be given the right to interpret and detach from automatic decision-making.