何以透明,以何透明:人工智能法透明度规则之构建

Why and How to Achieve Transparency:Establishing Rules for Artificial Intelligence Law

透明度作为人工智能风险治理的一项基本要求,其现实基础在于人工智能应用引发的信息不对称以及这种不对称所蕴含的对个体受保护利益或社会秩序的侵害风险。当下,信息不对称主要存在于人工智能系统分别作为产品、交互性媒介及自动化决定系统而影响现实世界等三类场景之中。相应地,透明度规则的主体内容应为人工智能系统提供者、运营者为满足相对人透明度需求而负有的信息披露义务。考虑到技术保密性对于产业发展的关键作用,且对外披露亦不是增加系统安全性的有效方式,对人工智能技术应以不强制披露为原则。诸项透明度规则既可归于“人工智能风险治理”主题之下,同时又分别属于个人信息保护法、消费者权益保护法、产品质量法、合同法、程序法乃至网络安全法规则,彰显出人工智能法的领域法特征。

Transparency,a basic requirement of AI risk governance,stems from the information imbalance caused by AI applications and the potential risks to individual rights or social order that come with such imbalance.At present,information imbalances are prevalent in three key scenarios where AI systems function as products,interactive platforms,and automated decision-making systems in the real world.Accordingly,the core focus of transparency regulation should center on the obligation of AI system providers and operators to disclose information to address the transparency needs of all parties involved.Recognizing the importance of technological confidentiality in industrial development and the limited effectiveness of external disclosure in enhancing system security,a principle of non-compulsory disclosure should guild AI technology practices.The transparency rules can be classified within the framework of"AI risk governance"while also intersecting with the law related to personal information protection,consumer rights,product standards,contracts,legal procedures,and cybersecurity,underscoring the diverse legal landscape governing AI practices.