人工智能的快速发展使其能够将人、机器、物体等各种事物紧密联结，形成万物互联的社会形态，扮演愈发关键的媒介角色。但各国如何理解、塑造和治理人工智能的媒介应用还有待分析。本研究采取话语网络分析法，以中美欧的人工智能治理政策为研究对象，从协同治理体制的结构和权能配置两个方面分别提炼人工智能社会技术想象的想象内容、主体关系及建构逻辑。研究发现：各国治理目标和治理主体均存在一定差异，面临不同的话语封闭问题。这也说明人工智能协同治理是一种主观的政治决策。为了避免人工智能的 社会技术想象因话语封闭而陷入科林格里奇困境，本文认为，我国可以通过观察他国的治理体制，在立足自身国情的基础上，丰富人工智能社会技术想象的内容、促进想象主体的多元化。

The rapid development of artificial intelligence enabled it to closely link various things such as human-cyber-physical system, creating a social pattern of interconnection of everything positioning AI as an increasingly critical media force. However, how countries understand, shape and govern the mediated use of AI remains an area that requires further in-depth analysis. Utilizing the discourse network analysis method, this research dissects the AI governance policies of China, the United States and Europe. It specifically scrutinizes the structural composition and authority distribution within their collaborative governance systems. Through this analysis, the research aims to unravel the imagined content, subject relations, and construction logic of the AI sociotechnical imaginaries. It is found that there are some differences in the governance objectives and key governance stakeholders across countries, leading to distinct challenges related to discursive closure. It also verifies that the AI collaborative governance is a subjective political decision. To prevent the AI sociotechnical imaginaries from succumbing to Collingridge’s dilemma due to discursive closure, this research suggests that China can enrich the content of AI sociotechnical imaginaries and foster the diversity of imaginary actors by examining the governance systems of other countries, while grounding its approach in the context of its own national conditions.