Structural consistency in AI governance: A PMC index assessment with evidence from China’s central-level policies

Abstract

The structural coherence of policy design has become an increasingly important issue in artificial intelligence (AI) governance. This study evaluates the structural consistency of China’s central-level AI policies issued between 2016 and 2025 (n = 54). It combines text mining to identify high-frequency policy terms and semantic co-occurrence patterns with a Policy Modeling Consistency (PMC) index framework comprising nine primary and forty-three secondary indicators. Five representative policies are then selected for detailed quantitative evaluation and visual comparison. The results show that China’s AI policy system is generally well structured, but still exhibits notable weaknesses in temporal planning, intergovernmental coordination, and incentive design. In particular, long-term policy supply remains limited, vertical coordination mechanisms are insufficiently institutionalized, and policy instruments are unevenly configured across key support dimensions. These findings suggest that future policy improvement should focus on strengthening medium- and long-term planning, enhancing coordination across governance levels, and improving the integrated design of policy instruments. Methodologically, the study demonstrates a reproducible analytical framework linking text analysis, indicator construction, quantitative evaluation, and visualization. It contributes to the literature by moving from thematic description toward structural assessment in the study of AI governance.