摘要
生成式人工智能算法治理已成为最具前沿性和紧迫性的议题之一。运用扎根理论对26份法律政策文本进行编码分析,研究发现,我国建立了以“政府层—企业层—用户层”为主的生成式人工智能算法风险法律治理架构。政府层构建安全治理机制、风险监管机制和长效规范机制以优化算法治理;企业层侧重事前预防、事中保障和事后审核以确保算法合规;用户层围绕算法素养、维权保障展开制度设计。未来应采取以下关键举措创新生成式人工智能算法风险的法律治理:加速制定算法标准和披露规则,促进政府协同监管;健全追责机制与试行监管沙盒,激发企业治理潜能;提升权利感知与改善举证责任,减轻用户维权负担;促进互动衔接与实现利益均衡,打破算法治理孤岛。
Generative artificial intelligence algorithm governance has become one of the most cutting-edge and urgent issues.Using grounded theory to encode and analyze 26 legal policy texts issued by the central government of China,the study found that China has established a legal governance framework for the risk of generative artificial intelligence algorithms based on‘government layer-enterprise layer-user layer'.The government layer builds a security governance mechanism,a risk supervision mechanism,and a long-term normative mechanism to optimize algorithmic governance;the enterprise layer focuses on ex-ante prevention,in-process guarantee and ex-post audit to ensure algorithm compliance;the user layer develops system design around algorithm literacy and rights maintenance.Looking forward to the future,the following key measures should be taken to innovate the legal governance of the risk of generative artificial intelligence algorithms:improve the formulation of algorithm standards and disclosure rules,and promoting government collaborative supervision;improve the accountability mechanism and try out the regulatory sandbox to stimulate the potential of corporate governance;improve the perception of rights and improve the burden of proof,reduce the burden of user rights protection;improve interactive convergence and achieve balance of interests,and break the island of algorithmic governance.
出处
《新媒体与社会》
2024年第2期60-73,M0007,共15页
New Media and Society
基金
国家社科基金重大专项课题“建立健全我国网络综合治理体系研究”(项目编号:20ZDA062)。
关键词
生成式人工智能
算法风险
法律治理
扎根理论
Generative Artificial Intelligence
Algorithmic Risk
Legal Governance
Grounded Theory
