算法决策应用的外部风险及其公共治理路径研究

Research on the External Risks of Algorithm Decision-making Application and its Public Governance Path

算法决策应用的外部风险具有聚合性和累积性特征,侧重算法行为端,在技术运用过程中向多数不特定群体扩散,将其类型化后包括社会公共利益和实质正义受损、人的主体性地位和合法权益受挑战两方面。外部风险不同于一般侵权行为的损害后果,无法适用传统归责原则。为了有效促进算法技术和监管的良性互动,以公共治理路径为视角,将算法决策嵌入网络社会架构中分析当下治理的局限性,并提出较完善的治理方案。在思路上,应遵循分类分级的精准化治理方法,选择“软”“硬”法为协同治理工具,坚持安全和发展并重的治理理念。在路径上,应着眼于算法决策应用的生命全周期,用算法解释机制解决算法备案制失灵的问题,以风险比例确立算法决策系统的分类与规制方向,并以平台为主体多途径构建算法综合性问责体系,形成治理合力。

The external risks of algorithmic decision-making and application have the characteristics of aggregation and accumulation,focusing on the behavior of algorithms,spreading to most unspecified groups in the process of technology application,and including the damage to social public interests and substantive justice,and the challenge of people’s subjectivity and legitimate rights and interests.External risks are different from the damage consequences of general torts,and the traditional principle of attribution cannot be applied.In order to effectively promote the benign interaction between algorithm technology and supervision,this paper embeds algorithmic decision-making into the network social architecture from the perspective of public governance path,analyzes the limitations of current governance,and proposes a more complete governance scheme.In terms of thinking,we should follow the precise governance method of classification and grading,choose the“soft”and“hard”methods as collaborative governance tools,and adhere to the governance concept of paying equal attention to security and development.On the path,we should focus on the whole life cycle of algorithm decision-making application,use the algorithm interpretation mechanism to solve the problem of algorithm filing system failure,establish the classification and regulation direction of the algorithm decision-making system with risk ratio,and build a comprehensive accountability system for algorithms with the platform as the main body and multiple channels to form a governance synergy.