AI adoption and green innovation quality: evidence from China

Using Chinese A-share listed firms from 2008 to 2023, this paper explores the association between corporate artificial intelligence (AI) adoption and the composition of green innovation quality. We measure firm AI engagement via textual keyword frequency extracted from annual reports, and adopt green invention and green utility-model patent applications as imperfect proxies reflecting differentiated green innovation thresholds rather than absolute high-low quality classification. Baseline two-way fixed-effect results reveal that AI adoption correlates more strongly with green invention filings; several AI-patent-based alternative indicators also show positive links with utility-model green patents. Our findings remain robust after switching AI measurement indicators, lagging core explanatory variables, using AI investment and patent data, and incorporating province-level time-varying economic and regulatory controls, though strict causal inference cannot be achieved with observational data. Further channel tests suggest R&D expenditure serves as the most stable supportive transmission path. The conclusions provide targeted empirical evidence for China’s dual-carbon and digital-green integrated policy design.