Artificial Intelligence, Digital Innovation, and Corporate ESG Performance

Abstract

Using a panel of 732 Chinese listed manufacturing firms from 2011–2022, this paper examines whether firm-level generative artificial intelligence (GAI) capability is associated with improved corporate ESG performance. We construct a text-based GAI measure by counting GAI-related keywords in annual reports and taking ln(count+1), and we measure ESG performance by converting Huazheng ESG ratings into a 1–9 score. Firm- and year-fixed effects results indicate a positive and statistically significant relationship between GAI and ESG outcomes. Mechanism analysis suggests that digital innovation, proxied by firms’ digital patents, partially mediates this relationship: GAI is positively related to digital innovation, and digital innovation is positively related to ESG performance. Heterogeneity tests show that the positive association is more evident among state-owned and foreign-invested firms, and it is stronger for large firms. In contrast, the moderating effect of supply chain finance is not statistically supported. A panel threshold model identifies a managerial-efficiency threshold around 6.01×108, below which the GAI–ESG link is insignificant and above which it becomes strongly positive. Robustness checks, including DDML and instrumental-variable estimation, confirm the stability of the main findings.