算法嵌入公共治理的内在逻辑与风险规制

The Inherent Logic and Risk Regulation of Algorithm Embedding in Public Governance

算法的嵌入带来公共治理效率和效果的提升。算法嵌入公共治理至少具有三重内在逻辑,即“数据+算法+模型”的技术逻辑、提升公共治理效能的实践逻辑和推进国家治理现代化的价值逻辑。当算法参与公共治理实践时可能产生技术、法律、社会和伦理四重风险。可以在算法嵌入公共治理的事前、事中和事后三个阶段对其进行风险规制:在事前阶段,制定算法技术统一标准,建立算法影响评估机制,完善算法信息公开制度;在事中阶段,以数据权利制衡算法权力,以算法审计规制算法权力,以人工智能治理原则钳制算法权力;在事后阶段,打通异议申诉渠道,并构建科学合理的责任分配机制。与此同时,将法律、道德和文化等外部环境因素纳入考量,对算法风险进行系统约束。

The embedding of algorithms brings an improvement in the efficiency and effectiveness of public governance.The embedding of algorithms in public governance has at least three inherent logics:the technical logic of“data+algorithms+models”,the practical logic of enhancing public governance efficiency,and the value logic of promoting the modernization of national governance.When algorithms are involved in public governance practices,they could pose four-fold risks:technological,legal,social,and ethical risks.Risk regulation of algorithm embedding in public governance can be carried out in three stages:the pre-stage,the in-stage,and the post-stage.In the pre-stage,unified standard for algorithm technology should be established,an algorithm impact assessment mechanism should be developed,and an algorithm information disclosure system should be improved.In the in-stage,the algorithm power should be balanced with data rights,and regulated through algorithmic auditing,and restrained by principles of artificial intelligence governance.In the post-stage,accessibility for public objection appeals should be established,and a scientifically reasonable mechanism for assigning responsibility should be constructed.At the same time,external environmental factors such as law,morality,and culture should be taken into consideration to systematically constrain algorithmic risks.