生成式治理:大模型时代的治理新范式  

Generative Governance:A New Paradigm of Governance in the Era of Large Language Models

作　　者：米加宁[1] Mi Jianing

机构地区：[1]哈尔滨工业大学经济与管理学院,哈尔滨150006

出　　处：《中国社会科学》2024年第10期119-139,207,共22页Social Sciences in China

摘　　要：“生成式治理”这一新兴公共治理范式基于生成式人工智能技术,通过全域数据感知、跨域知识萃取、多情景策略创生,对公共事务治理中的决策制定和执行优化进行系统性赋能。泛在数据、大模型算法和大算力共同驱动了生成式治理的兴起。在理论上,生成式治理效用的发挥取决于知识的生成式“涌现”、自组织与进化、工具理性与价值理性互构。在实践上,通过数据喂料、算法迭代和模型精调,生成式大模型可以推动专业化的技术治理走向通用性的公共事务治理,在广泛、复杂和不断变化的公共事务环境中实现从微观个体到宏观系统的多尺度治理。生成式治理将揭开一个更加智能、高效和人性化的公共治理新时代的序幕。"Generative governance,"an emerging paradigm of public governance,is based on generative artificial intelligence technology.It systematically empowers decisionmakingand execution1optimizationinpublicaffairs governance through comprehensive data perception,cross-domain knowledge extraction,and multiscenario strategy generation.The rise of generative governance is jointly driven by the support of ubiquitous data,large language model algorithms,and highperformance computing.Theoretically,the efficacy of generative governance depends on the generative"emergence"of knowledge,self-organization and evolution,and the interplay between instrumental rationality and value rationality.In practice,through data feeding,algorithm iteration,and model fine-tuning,generative large language models can propel transition from specialized technical governance to general public affairs governance,achieving multi-scale governance from micro-level individuals to macro-level systems in a wide,complex,and constantly changing public affairs environment.Generative governance will herald a new era of more intelligent,efficient,and humane public governance.

关 键 词：生成式治理 生成式人工智能 公共治理 大模型 

分 类 号：D630[政治法律—政治学]