AI and Industrial Policy
Aug 08-2024
AI-related industrial policy by a local government has positive correlation with AI patent filings in that particular area, and a company’s AI adoption is related to its business growth, said Prof. Li Lixing in his speech at the 186th Langrun Policy Talk. He is PKU Bo Ya Young Scholar, NSD Professor of Economics, and Director of PKU China Center for Public Finance.
By sifting through local government work reports, Prof. Li and his team found that their mentioning of AI has shot up since 2017 when a blueprint for developing new-generation AI was released by the State Council. They then applied algorithm to distinguish four types of AI industrial policy each covering a specific area: data and computing infrastructure, industry-university-research fusion, AI in governmental scenarios, and AI for transforming traditional industries. The second type, aiming to promote synergetic innovation of industries, universities, and research institutes, surged in mentioning since 2015 to 2016. Notably, ‘digital government administration’ enjoyed a large amount of mentioning in government work reports. This indicates that local governments are working on providing AI application scenarios and promoting the development of AI industry from the supply side, said Prof. Li.
To assess corporate adoption of AI, Prof. Li and his team parsed data on newly registered firms, job placements, and patent filings. They found that after policy on AI industry-university-research coordination was released, relevant patent filings by local firms spiked by 23.6% on average. To obtain further insights into the impact of policy, the research team made attempts to distinguish ‘talk’ on AI (mentioning of AI industry in policy) and ‘do’ (implementing AI policy). Initial results showed that the latter was more effective, though further research and verification would be needed.
The relationships between AI adoption and corporate growth cannot be inferred to be causal ones, as they might be subject to selection effect, cautioned Prof. Li. Therefore, current research results are not applicable in a wholesale manner.
