自动决策算法的风险识别与区分规制

Risk Identification and Differentiated Control of Automatic Decision-Making Algorithm

自动决策算法既属于专业技术,又具有赋能属性,可以和既有技术相结合催化新的应用技术和场景。与之相应,立法认识论既要注意到自动决策算法作为科学技术内含的伦理性风险,同时也需要识别不同应用场景中的复数价值以及利益冲突。为了规范内生性科技风险,立法应当根据风险程度强化对算法活动的程序性控制,确立记录和报告义务,细化算法风险评估规则,完善以个人信息保护负责人为核心的二元监管体系。为了管控因赋能导致的衍生性应用风险,立法一方面应当强调人的主体性,另一方面应根据各类应用场景中的利益顺位,灵活配置权利。我国个人信息保护法规定的算法解释权在公共管理领域和商业领域意义不同,相应的规则应有能力区分公权力应用与商业应用以及是否包含知识产权等不同场景;算法结果拒绝权旨在实现个人利益、商业利益和公共利益之间的平衡,应通过例外规则保持制度弹性。如果自动决策算法在公共管理和特殊社会治理领域中的应用可能带来高风险,法秩序应施加更多的应用要求或限制,以防止算法赋能导致原有秩序功能异化。

The automated decision-making algorithm,as a specialized empowering technology,can catalyze new application technologies by utilizing existing ones.Accordingly,in legislation,its inherent ethical risks should be guarded against,and its plural values and the conflict of interests in different application scenarios should be recognized.To regulate its endogenous technological risks,legislation should improve the dual supervision system by strengthening the procedural control over its activities according to risk levels,establishing its obligation to record and report,refining the rules for the assessing the risks,and centering around the responsibility of the person in charge of protecting personal information protection.To regulate derivative risks owing to the algorithm’s catalytic function,legislation should on the one hand emphasize the subjectivity of human beings,and on the other hand flexibly assign rights according to different interests in various application scenarios.Specifically,the right to algorithm explanation,under the Personal Information Protection Law,has different meanings,and the corresponding rules should be tailored to different scenarios by differentiating the application by the public power from the one for commercial purpose,and by judging whether intellectual property rights are involved.Moreover,as the right to reject algorithmic results aims at balancing individual,commercial and public interests,its institutional flexibility should be maintained through rules of exception.In the public management or social governance,if the application of automated decision-making algorithms may bring high risks,the legal system should impose more application requirements or restrictions to prevent the alienation of society due to algorithmic empowerment.