基于信任的自动化决策:算法解释权的原理反思与制度重构

基于个人信息的算法自动化决策越来越普遍,我国《个人信息保护法》和相关国内外法律都引入算法解释权加以应对。但由于算法具有黑箱性、不确定性与复杂性,算法解释权是否应理解为一般性权利还是限定性权利,要求系统性解释还是个案性解释、事前解释还是事后解释、人工解释还是机器解释,都存在解释争议与适用困境。在原理层面,这一争议与困境源于个人算法控制论。应以沟通信任原则重构算法解释权,将算法解释权的性质视为一种程序性权利,而非实体性权利。算法解释权制度也应进行重构,建议根据自动化决策所处的行业领域、市场化程度、个案影响、企业能力而对其内容、程度、时间和方式作不同要求。

Automatic Decision-Making Based on Trust: Principle Reflection and System Reconstruction of the Right to Algorithm Interpretation

Algorithmic automatic decision-making based on personal information is becoming more and more common in contemporary society.The Personal Information Protection Law of China and relevant domestic and foreign laws all introduce the right to algorithm interpretation to deal with it.However,due to the black box nature,uncertainty and complexity of the algorithm,there are interpretation disputes and application dilemmas as to whether the right to algorithm interpretation should be understood as general right or limited right,systematic interpretation or case interpretation,ex ante interpretation or ex post interpretation,and manual interpretation or machine interpretation.At the principle level,this controversy and dilemma stems from individual algorithm control.The right to algorithm interpretation should be reconstructed based on the principle of communication and trust,and the nature of the right to algorithm interpretation should be regarded as a procedural right rather than a substantive right.The institution of the algorithm interpretation right should also be reconstructed.The content,degree,time and method of algorithm explanation right should be regulated according to the industry field,marketization degree,case influence and enterprise ability.