• 中国科学学与科技政策研究会
  • 中国科学院科技政策与管理科学研究所
  • 清华大学科学技术与社会研究中心
ISSN 1003-2053 CN 11-1805/G3

科学学研究 ›› 2024, Vol. 42 ›› Issue (8): 1596-1606.

• 科学学理论与方法 • 上一篇    下一篇

三重逻辑下 AI 技术治理制度供给质效提升研究

贾婷1, 陈强2   

  1. 1. 同济大学经济与管理学院
    2.
  • 收稿日期:2023-06-27 修回日期:2023-10-07 出版日期:2024-08-15 发布日期:2024-08-15
  • 通讯作者: 贾婷
  • 基金资助:
    国家社科基金项目《面向新征程的国家创新体系整体效能提升研究》

A study on the quality and efficiency improvement of AI technology governance institutional supply under triple logic

  • Received:2023-06-27 Revised:2023-10-07 Online:2024-08-15 Published:2024-08-15

摘要: 从任务型围棋程序AI-AlphaGo到生成式大语言模型Chat GPT,人工智能技术的功能外延在不断进化,相关制度供给也应随着AI所引发的全域性变革快速迭代响应。研究梳理了我国人工智能技术治理的制度供给现状,发现其存在技术发展与风险治理的总体困境、多元需求与供给不足的目标困境、理论响应与实践转化的执行困境,制度供给的有效性不足。研究还发现,从理论层面来看,“国家、市场及社会”三重制度逻辑与制度供给间存在着天然的内洽性;从实践层面来看,三重制度逻辑与人工智能技术治理需求也存在明显的耦合性。人工智能技术治理制度供给的质效提升,一方面要将制度逻辑内嵌于人工智能技术治理框架中,以精准把握确定性治理的总体需求,保证供给质量;另一方面,要在精准把握新需求、明确治理目标的基础上,辅之以“试探性”治理路径,来敏捷应对人工智能技术发展的不确定性和复杂性,提升供给效率。对三重逻辑下人工智能技术治理制度供给质效提升的路径思考,将助益于整体把握我国人工智能技术治理的重点需求和发展方向。

Abstract: From the task-based Weiqi program AI-AlphaGo to the generative large language model ChatGPT, the functional outreach of AI technology is constantly evolving. Therefore, the relevant institutional supply should respond quickly and iteratively to the domain-wide changes triggered by AI. The study sorts out the current situation of institutional supply of AI technology governance in China, that a system of institutional provision with guidance and regulation complementing each other has been put in place. However, some dilemmas remain, including the overall dilemma between technology development and risk governance, the objective dilemma between diversified demand and insufficient supply, and the execution dilemma between theoretical response and practical transformation, which reveals the insufficient effectiveness of institutional supply. The study finds that there is natural internal self-compatibility between the triple institutional logic of "state, market and society" and institutional supply from the theoretical level; there is also an obvious coupling between the triple institutional logic and the demands of AI technology governance from the practical level. To improve the quality and efficiency of AI technology governance institutional supply, on the one hand, the institutional logic should be embedded in the framework of AI technology governance, in order to accurately grasp the overall demand for deterministic governance and ensure the quality of supply. Firstly, the nation logic of institutional supply proposes that AI technology should be developed on the basis of guaranteeing national scientific and technological security and social stability. It is necessary to externally enhance the awareness of scientific and technological strategic forward layouts and strengthen the ability to secure network security. Also, the existing social and ethical norms, laws and policy systems should be adjusted internally to avoid a "regulatory vacuum" and ensure the safety and controllability of AI technology development and application by clarifying the responsibilities of the main actors. Secondly, the market logic of institutional supply advocates the use of regulation to promote the healthy and orderly development of the AI industry based on respecting the law of market development. It defines safety testing and assessment standards for AI systems, products and services and forms a broad consensus of social participation. Thirdly, the society logic of institutional supply calls for strengthening the ethical governance of science and technology and enhancing the well-being of the people, which emphasizes creating more opportunities for the intelligent participation of marginalized social strata by connecting their needs through system innovation. On the other hand, on the basis of accurately grasping the new demands and clearly defining the governance goals, it is necessary to supplement the "exploratory" governance path to agilely respond to the uncertainty and complexity of the development of AI technology, and thus to enhance the efficiency of supply. Firstly, carry out dynamic practices in an iterative and gradual manner, and apply more guiding policies for dynamic management. Secondly, verify the effectiveness of governance in a small-scale pilot program, and seek only certain solutions to key issues. Finally, respond to uncertainty with informal governance, and reduce the constraints of the legitimacy process. Reflections on the path of improving the quality and efficiency of AI technology governance institutional supply under the triple logic will help to grasp the key needs and development direction of AI technology governance in China as a whole.