自动化行政算法决策的特殊风险与规制路径

Specific Risks and Regulatory Paths of Algorithmic Decision-Making in Automated Administration

摘要: 自动化算法决策的应用给行政公正原则、正当程序原则、权力与责任相一致原则带来了巨大挑战。这些挑战表现为行政决策违法或不当、决策违反法定程序、行政责任主体不明、行政机关能力弱化等风险。客观层面上，数据流转偏差、算法不透明与技术权力的侵蚀促成此类特殊风险；主观层面上，行政主体对算法的依赖扩张了风险的不可控程度。为有效控制此类新型风险、使算法行政决策更好地造福人类，可以从地位厘定、义务强化、程序控制等角度切入，继续坚持行政机关在算法决策中的第一责任人地位、坚守算法的工具主义立场，场景化、类别化规制算法决策应用程序，在强化相关主体披露义务的同时，创新式贯彻正当程序控制，完善相对人的权利保障与救济体系。

Abstract: The application of automated algorithm decision-making has brought huge challenges to the principle of administrative justice, due process, and the principle of consistency of power and responsibility, including illegal or improper administrative decision-making, violation of legal procedures, unclear administrative responsibility, and weakening of the ability of administrative agencies. On the objective respect, data flow deviation, algorithm opacity and the erosion of technical power contribute to such special risks; on the subjective ones, the administrative body’s dependence on algorithms expands the uncontrollability of risks. In order to effectively control this new type of risk and make algorithmic administrative decision-making better benefit mankind, we can cut from the perspectives of status determination, obligation enhancement, and procedural control, and continue to adhere to the position of the administrative organ as the first responsible person in algorithmic decision-making and stick to the algorithm instrumentalism, and scenario-based and categorized regulatory algorithm decision-making applications, while strengthening the disclosure obligations of relevant subjects, innovatively implement due process control, and improve the rights protection and relief system of the counterparty.