算法如何平等:算法歧视审查机制的建立

How Algorithms are Equal:Establishment of Algorithmic Discrimination Review Regime

人工智能时代算法在自动化决策方面发挥巨大作用的同时,也带来算法黑箱、算法歧视等负面效应。机器算法得出的决策并非人们想象的天然中立,可能会出现更为隐蔽和严重的歧视行为。规制算法歧视的理论基础在于平等权保护、国家保护义务和国家风险预防义务。实践中对算法歧视的行业自律性规制以及政府规制,均以知情同意模式为核心,立足于个人信息保护,无法有效解决算法歧视问题。应建立算法歧视审查机制,设立专门的算法审查机构,赋予其备案、调查、支持诉讼的职能。在审查标准上,针对直接歧视和间接歧视两种类型,应分别采用差别性对待标准和差别性影响标准。

In the era of artificial intelligence,algorithms play a huge role in automated decision-making,but also bring negative effects such as algorithmic black box and algorithmic discrimination.The decision made by the machine is not natural neutrality as people imagine,and there may be more hidden and serious discrimination.The theoretical basis of government regulations on algorithmic discrimination lies in equal rights protection,national protection obligations and national risk prevention obligations.In practice,industry sell-discipline regulations and government regulations that discriminate against algorithms are based on the informed consent model,based on the protection of personal information,and cannot effectively solve the problem of algorithmic discrimination.An algorithmic discrimination review regime should be established,and a special algorithm review institution should he evStablished to give it the functions of filing,investigating,and supporting litigation.In terms of review standards,for the two types of direct discrimination and indirect discrimination,the standard of differential treatment and the standard of differential impart should be adopted respectively.