面向复杂性社会场域的生成式治理新进路——基于生成式人工智能的治理方式探索  

A New Approach to Generative Governance for Complex Social Fields——Exploring Governance Methods Based on Generative Artificial Intelligence

作　　者：刘文 赵新峰 Liu Wen;Zhao Xinfeng(College of Public Administration,Shandong University of Business and Technology,Yantai,264005,China;Institute of State Governance,Beijing University,Beijing,100871,China)

机构地区：[1]山东工商学院公共管理学院,山东烟台264005 [2]北京大学国家治理研究院,北京100871

出　　处：《公共管理学报》2025年第1期102-114,173,共14页Journal of Public Management

基　　金：国家社会科学基金项目(24CZZ055)。

摘　　要：本文尝试对面向复杂性社会场域的生成式治理进路进行阐析,探讨了生成式治理蕴含的治理观念与基本原则,分析了生成式人工智能与生成式治理的内在契合性,系统论述了生成式人工智能支持下的生成式治理过程,并结合多智能体建模技术对规范分析进行了理论验证。研究认为,结构性与生成性是治理因应的两种基本思路,社会场域复杂性的加速提升,使得结构性治理难以应对并呈现出自反性。复杂性社会场域作为非线性交互网络以生成性方式展开自身,决定了复杂性社会场域治理因应的生成性进路。生成式人工智能内蕴的生成性运行方式及其展现出的涌现能力,使其能够有效因应复杂性社会场域中的问题与情境,一方面,通过对复杂交互关系的疏通以及对多元构成要素的赋能,为关于复杂性关系网络本身的引导和治理奠定新的基础;另一方面,在“去结构化”和“去主体化”的非线性交互关系中实现的跨域边界和领域的即时综合性知识生成和多模态建构,能够有效满足在复杂性场域因应中的情境性、综合性和动态演化性要求。本文为理解复杂性社会场域条件下治理的生成性转向及其展开提供了一种总体图景。This paper attempts to elucidate the generative governance approach oriented towards complex social fields.It explores the governance concepts and basic principles which are inherent in generative governance,analyzes the inherent compatibility between generative artificial intelligence and generative governance,systematically discusses the process of generative governance supported by generative artificial intelligence,and conducts theoretical verification of the normative analysis using multi-agent modeling techniques.The research suggests that structural and generativity are two basic approaches in governance adaptation.The accelerating complexity of social fields makes it difficult for structural governance to cope and presents contradictions.As complex social fields unfold themselves in a generative manner through nonlinear interaction networks,this determines the generative approach in adapting governance to complex social fields.The generative operational mode and emergent capabilities inherent in generative artificial intelligence enable it to effectively address issues and situations within complex social fields.On the one hand,by facilitating complex interactive relationships and empowering diverse constituent elements,it lays a new foundation for guiding and governing the complex relational networks themselves.On the other hand,the real-time integrated knowledge generation and multimodal construction across domain boundaries and fields,achieved within"de-structured"and"de-subjected"nonlinear interactive relationships,can effectively meet the contextual,comprehensive,and dynamically evolving requirements of adapting to complex fields.This paper provides an overall picture for understanding the generative turn in governance and its unfolding under the conditions of complex social fields.

关 键 词：生成性 复杂性社会场域 生成式治理 生成式人工智能 多智能体建模 

分 类 号：D035[政治法律—政治学]