摘要
Standard development is a critical activity that facilitates innovation collaboration, regulates industrial development, and ensures the safety of science and technology. In response to the era of artificial intelligence, China is accelerating its AI standard process, giving rise to an AI standard collaboration network characterized by multi-stakeholder synergy among government, industry, academia, and research institutes, as well as broad participation from the entire industry. To analyse the stage-specific characteristics and evolutionary patterns of the AI standard collaboration network, this research extracts 1,658 drafting organizations from 271 AI standards and applies social network analysis to construct collaboration networks across three periods: the stable start and pilot exploration period (pre-2017), the top-down design and strategic layout period (2017-2024), and the comprehensive empowerment and systematic advancement period (2025-present). The findings are as follows: First, during the stable start and pilot exploration period, AI standards were few in number, narrow in scope, and limited in type. Standard drafting was carried out by a handful of specific actors, with little collaboration among them, and thus no standard collaboration network formed. Second, during the top-down design and strategic layout period, the number of AI standards increased significantly, their content gradually matured, and their types became diverse. A firm-led standard collaboration network began to take shape, with notable changes in network size, number of ties, density, average path length, clustering coefficient, and other indicators. A core-periphery structural pattern became evident, with Beijing Baidu Netcom Science Technology Co., Ltd., Huawei Technologies Co., Ltd., and the China Electronics Standardization Institute occupying advantageous positions within the network. Third, during the comprehensive empowerment and systematic advancement period, the “AI+” action has deeply integrated AI into numerous industries. The dominance of core standardisation organisations has further strengthened, emerging industry actors have risen rapidly, the influence of universities has somewhat diminished, and the network structure displays a one core with multiple strong players characteristic. Fourth, the effects of transitivity and geographic homophily on the standardisation collaboration network have weakened, whereas the influence of organisational homophily has strengthened, and the mediating effect exerts a negative influence.