中国人工智能算力产业发展的规制困境及其解决路径

    Regulation Dilemma and Its Solution to the Development of China's Artificial Intelligence Computing Industry

    • 摘要: 算力是人工智能发展的基础要素,也是监管人工智能的重要工具。算力治理具有二重性,既要促进算力创新发展和安全供给,又要通过算力规制有效控制人工智能风险,算力治理的矛盾和张力客观存在。研究发现:(1)算力产业发展的资金、能源投入需求大,需要建立发展激励制度以实现资源有效供给;(2)算力产业发展的安全性、稳定性要求高,软硬件供应链脆弱,需要建立安全保障制度以实现风险可控;(3)算力产业发展的高投入特性容易形成竞争壁垒和市场垄断,需要通过反垄断制度保障行业健康发展;(4)算力产业发展对技术标准和互操作性要求高,需要加强国际沟通与协调,用好涉外法治的工具箱。基于算力产业发展的制度需求,通过调控激励制度实现资源配置的均衡、加强算力安全以增强人工智能安全、调控算力市场竞争以规范人工智能发展秩序、加强算力领域涉外法治以推进人工智能国际治理等方面提高算力规制的效能。提出破解算力产业发展与算力规制的矛盾,实现安全与发展的深度融合,需要构建算力产业发展与规制的平衡机制,进行必要性评估,确有必要时才介入和调控算力的资源等制度要求;即便确有必要进行规制,也应坚持差异性监管,针对大模型大企业可以做专门性制度要求,对一般主体做基本要求,对中小型主体可以豁免部分要求;规制应进行整体性布局,整体考虑算力产业发展的社会价值与其他约束性目标背后的公共价值;规制要注重协同性推进,设立统一的监管机构主管安全与发展,建立敏捷包容、取向一致性的工作机制。

       

      Abstract: Computing power is a fundamental element in the development of artificial intelligence and an important tool for its regulation. The governance of computing power has a dual nature, which requires promoting innovative development and secure supply of computing power, as well as effectively controlling artificial intelligence risks through computing power regulation. Contradictions and tensions objectively exist. (1) The high demand for funding and energy investment in the development of computing power needs to establish a development incentive system to achieve effective resource supply; (2) the high security and stability in its development with a reliable software and hardware supply chain, needs to establish a security guarantee system to control risks; (3) the high investment in its development can easily form competitive barriers and market monopolies, and it is necessary to ensure the healthy development of the industry through anti-monopoly systems; (4) the high technical standards and interoperability in its growth needs to strengthen international communication and coordination, and make good use of the toolbox of foreign-related rule of law. Based on the institutional demand for the computing power development, regulating incentive systems are needed to balance resource allocation, enhance artificial intelligence security by strengthening computing power security, regulate the computing power market competition for better development orders of artificial intelligence, and promote international governance of artificial intelligence for more effective regulations through strengthening the rule of law in the field of computing power. To resolve the contradictions between the development of computing power and its regulation, and achieve a deep integration of security and development, it is necessary to establish a balance mechanism between development and regulation. Necessary assessments are needed and conducted when in need. Different regulations are necessary for different supervisions, such as special institutional regulations for large models and enterprises; basic ones for general entities, and partial exemptions for small and medium-sized entities. The overall layout design should be made for the regulations, taking into account the social value and the public values behind other binding goals in computing power development; synergy should be centered in promoting regulations, by establishing a unified institution to oversee its safety, development, and consistency in direction through a set of working mechanism.

       

    /

    返回文章
    返回