論文發(fā)表

近日，全球人工智能創(chuàng)新治理中心秘書長、復(fù)旦發(fā)展研究院副研究員姚旭合作論文《全球人工智能治理：多元主體結(jié)構(gòu)與互動機制》發(fā)表于《信息安全與通信保密》（2025年11月）。

研究基于人工智能技術(shù)快速迭代并深度嵌入全球社會結(jié)構(gòu)的背景，檢視了傳統(tǒng)的以國家為中心的單邊治理模式存在的困境。研究提出，全球人工智能治理正呈現(xiàn)出國際組織、主權(quán)國家、科技企業(yè)與公眾社會多元主體協(xié)同參與、跨層級互動的新格局。為此，本研究立足“規(guī)則—能力—認知”的分析框架，系統(tǒng)梳理四類關(guān)鍵行為體在制度規(guī)則、資源協(xié)作與公共空間三個維度的互動機制，揭示治理權(quán)力從剛性控制向軟性重構(gòu)的演進邏輯，為理解全球AI治理的范式轉(zhuǎn)型與體系重構(gòu)提供了理論支撐。

作者/Authors

姚旭、朱政宇

（姚旭，全球人工智能創(chuàng)新治理中心秘書長、復(fù)旦大學(xué)發(fā)展研究院副研究員，主要研究方向為人工智能、科技產(chǎn)業(yè)和數(shù)據(jù)治理）

摘要/Abstract

在人工智能技術(shù)快速發(fā)展并深度嵌入社會結(jié)構(gòu)的背景下，全球人工智能治理呈現(xiàn)出多主體參與、跨層級協(xié)作的新格局。以國際組織、國家主體、科技企業(yè)與公眾4類關(guān)鍵治理行為體為分析主線，梳理其在人工智能治理中的角色演變與互動機制。通過對制度規(guī)則互動、資源協(xié)作互動與公共空間互動 3 種主要關(guān)系類型的系統(tǒng)歸納，形成“規(guī)則—能力—認知”閉環(huán)，揭示多元主體在標(biāo)準(zhǔn)制定、能力組織與社會反饋等治理環(huán)節(jié)中的協(xié)同路徑與結(jié)構(gòu)特征。不同主體在互動過程中不斷推動人工智能治理從單邊規(guī)制向協(xié)商共治演進，逐步形成一個具備網(wǎng)絡(luò)特征與適應(yīng)能力的多元參與治理體系。這有助于深化對全球人工智能治理中行為體關(guān)系結(jié)構(gòu)與互動邏輯的理解，為構(gòu)建更具包容性和有效性的治理架構(gòu)提供基礎(chǔ)性支撐。

Against the backdrop of rapid advances in artificial intelligence(AI) and its deep embedding in social structures, global AI governance is taking shape as a new landscape characterized by multi–actor participation and cross–level collaboration. Focusing on four key categories of governance actors—states, international organizations, technology companies, and the public—this paper traces the evolution of their roles and the mechanisms of their interaction in AI governance. By systematically summarizing three main types of relationships—interaction through institutional rules, resource collaboration, and public sphere—it constructs a “rules–capabilities–cognition” closed loop, revealing the pathways of coordination and structural features through which diverse actors engage in standard–setting, capacity building, and societal feedback. In the process of interaction, these actors continuously propel AI governance from unilateral regulation toward negotiated co–governance, gradually forming a multi–actor governance system with networked characteristics and adaptive capacities. This study helps to clarify the relational structures and interaction logics among actors in global AI governance and provides a foundational basis for building more inclusive and effective governance frameworks.

關(guān)鍵詞/Key Words

人工智能治理；多元主體；制度規(guī)則；資源協(xié)作；公共空間

AI governance ; multiple actors ; institutional rule ; resource collaboration ; public sphere

原文鏈接/Link

https://kns.cnki.net/kcms2/article/abstract?v=vxt3_kKBHYNXjkzbYTQhxK2OmFnzNYUZHDXBm4h5kEwlR26VYlZVuJJYwPl8xf1U0xheo43OtuC8UyeNwNQQHXwMOvbnXssEM68IiqjN4ufO2FaTpGklGpTd9n_YCM4TYcVqzvSOxA3_l1YyF7K-9m5_EwOCP-E8RPO_KRG_uvnb78fPJqE6Dg==&uniplatform=NZKPT&language=CHS

來源｜全球人工智能創(chuàng)新治理中心

排版 | 趙浩博