【摘要】 人工智能技术的“双刃剑”效应要求法治体系突破静态规制的局限，动态体现社会行为方式和生活状态。本文从公共管理“敏捷治理”理论视角切入，构建了“风险识别-容忍厘定-风险处置-价值重塑”的分析框架，并结合“人工智能社会实验”实践行动，系统阐述了社会实验如何遵循敏捷治理逻辑，驱动人工智能在立法、执法、司法、守法等各环节实现“从实验到规则”的演进。研究发现，社会实验通过动态适应性、快速响应、多元协同三大机制为人工智能科学立法锚定方向、为适切执法提供标尺、为有效司法建立标准、为全民守法夯实基础。最终，从拓展领域场景、加强学术指导、促进成果转化、扩大组织动员四个方面提出人工智能社会实验驱动人工智能法治的具体路径。更多还原

【Abstract】 Against the backdrop of the structural contradiction between the rapid evolution of artificial intelligence(AI) technology and the lagging response of the current rule of law system,this paper explores how to promote the dynamic evolution of the rule of law system of AI through the social experimental method of AI.With the wide application of generative AI such as ChatGPT,the "unsupervised learning" and "uncertainty of generated results" further enlarge the "double-edged sword" effect of AI,posing systematic challenges to national security fields,such as political security,military security,and data security.The conventional static legal framework is difficult to effectively deal with the new problems caused by the iteration of AI technology.It is an urgent need for modernizing national governance to build an AI rule of law system with dynamic adjustment capabilities.Firstly,this paper constructs an integrated analysis framework of "risk identification-tolerance determination-risk disposal-value remodeling " from the perspective of agile governance theory in the field of public management.This framework not only embodies the core concept of agile governance,such as flexible adaptation,rapid response,and multi-coordination,but also provides a clear logical main line for understanding how AI social experiments drive the evolution of AI rule of law.Specifically,the framework systematically explains the realization logic of the rule of law of AI through the organic connection of four dimensions:(1) It lays an empirical foundation for scientific legislation through dynamic risk identification mechanism;(2) It provides an objective scale for appropriate law enforcement based on risk classification;(3) It transforms empirical results into standard norms to establish a unified basis for effective justice;(4) It is necessary to promote multi-subject value negotiation and lay a solid social foundation for the law-abiding of the whole people.Secondly,based on the specific practice of "AI social experiment" in Guangzhou,this paper analyzes the realization mechanism of social experiment driving AI rule of law.Through three key actions,it demonstrates how social experiments follow the logic of agile governance:(1) It uses the "unveiling and commanding " mechanism to dynamically organize social experimental research,and deploys social experiments in 18 specific scenarios in six major fields,such as smart medical care,smart manufacturing,and smart transportation;(2) It is necessary to follow the experimental method of "control-control-comparison" to accurately evaluate the social impact of A I;(3) It is necessary to promote the participation of multiple stakeholders in value rational consultation and build a governance pattern of co-construction and co-governance.Finally,the paper finds that the AI social experiment anchors the direction of AI science legislation through three mechanisms:dynamic adaptability,rapid response,and multiple coordination,provides a yardstick for appropriate law enforcement,establishes standards for effective justice,and lays a solid foundation for law-abiding people.Based on the research conclusion,this paper puts forward the concrete realization path of AI social experiment driving AI rule of law from four aspects:(1) expanding the field and scene of social experiment and strengthening the empirical basis of scientific legislation;(2) strengthening academic guidance and norms to enhance the adaptability of AI law enforcement;(3)promoting the standardization of social experimental results and supporting the standardization and unification of judicial adjudication;(4) expanding the mobilization of social experimental organizations and shaping a law-abiding culture of co-construction and co-governance.This paper innovatively combines the theory of agile governance with the social experiment practice of AI,and constructs an analytical framework that links the whole chain of legislation,law enforcement,justice and law-abiding,which fills the theoretical gap of existing research.The research results not only provide theoretical support for the construction of the AI rule of law system,but also provide operable policy tools for government departments to carry out AI governance practice.更多还原