面向未来的中国人工智能立法：思路与重点

【摘要】 <正>人工智能治理正逐步从软法治理向硬法治理迈进，我国人工智能立法也已经被明确纳入立法规划，构建一部满足中国人工智能安全治理和产业发展需求，同时符合国际关于人工智能治理共识的法律，是我们现阶段的目标。现阶段，人工智能法的出场有其必要性，主要体现在三个层面：一是护航人工智能规范健康发展。人工智能全领域应用，带来的信任风险、公平风险、失控风险、社会风险、责任风险等，都需要通过制定法律制度来化解。更多还原

【Abstract】 Since entering the digital age, artificial intelligence technology has profoundly changed people’s production and lifestyle, but it has also brought a series of uncertain risks, which require systematic response at the institutional level to promote the standardized and healthy development of artificial intelligence technology and industry. How to grasp the legislative pace of artificial intelligence law at present? How should the positioning, adjustment objects, and governance mode of artificial intelligence law be determined? All are further questions to be answered. For this reason, in the context of the two expert proposals for AI law issued by the academic community in 2023, we invited experts involved to write articles, with a view to deepening the discussion of AI legislation and providing theoretical support for the formal introduction of China’s AI in the future. Professor Shen Weixing pointed out that to accurately grasp the orientation and direction of China’s AI legislation, it is essential to clarify key issues in the legislative process, including the concept of AI, basic principles, risk management systems, transparency and filing supervision measures, damage and relief mechanisms. Professor Zhang Linghan suggested that the current stage of AI legislation should adopt a “general principles-based” approach, characterized by “establishing general principles, outlining main lines, leaving interfaces, and advancing flexibly and timely”. Researcher Zhou Hui emphasized that the AI law should encompass three orientations: as a promotion law, a safety law and a procedural law. At the level of a promotion law, it should primarily consist of encouraging and discretionary norms to serve and support the development of the AI industry. In terms of a safety law, a two-tier risk model should be established, along with a flexible and adjustable negative list for the AI field. Regarding procedural law, it should stipulate a coordinating department for AI to enhance the operability of the rules. Professor Su Yu emphasized that the technological roadmap and application methods of artificial intelligence are quite complex, and they should be appropriately differentiated and governed through a hierarchical and multi branch framework. Professor Zheng Zhifeng believes that attention should be paid to the outer space effectiveness boundary of artificial intelligence law and its coordination with artificial intelligence element law and artificial intelligence application law internally. Professor Xu Shenjian emphasized that the construction of a comprehensive and systematic data governance system should be promoted from three levels: establishing a sound data ethics system, promoting responsible innovation in data technology, and moving towards holistic governance of data laws. Professor Zhang Yan pointed out that based on the characteristics of existing AI, when converting the scientific and technological ethics framework into an effective governance scheme, a nurturing approach rather than a deterministic approach should be adopted. Professor Li Xueyao noted that the challenges of AI ethics mainly concentrate in the deployment and application stages. It is advisable to conduct separate legislation within the framework of scientific and technological ethics governance, and establish normative requirements based on its own characteristics in terms of the responsible entity for ethical review, review initiation conditions, expert composition, and review conclusions.更多还原