Quantitative mapping of the evolution of AI policy distribution, targets and focuses over three decades in China

Artificial intelligence (AI) technology policy plays a critical role to steer its applications to broadly relevant endpoints, and contributes to critical governance of innovations by governments, industry and society at large. In this paper, we adopt a bibliometrics-based research framework to characterize the development and evolution of China's AI policy. The framework integrates bibliometric methods, semantic analysis, and network analysis for identifying core policy elements and their evolution in the AI policy process. Specifically, we first collect China's central-level AI-related policies and identify four stages of its evolution based on policy-issuing frequency, policy trends, and core policy issuing time nodes. We then identify the core policies, core institutions, and core policy targets in each stage. Then we explore the policy issuing trends, policy distribution changes, and evolution of policy targets. Finally, patterns and characteristics of the policy process are identified, and trends are predicted. We used the PKULaw database to collect the policy-relevant data on AI in China, and the time frame is from 1990 to 2019. Our findings and the reported quantitative map might usefully inform AI policy in China and elsewhere around the world. It could also help broader stakeholder engagement in policy discussions on AI technology, industry and society.