Differentiated Impacts of Artificial Intelligence on Carbon Emissions in Manufacturing: Evidence from a Quasi-Natural Experiment in China

Exploring use cases of artificial intelligence (AI) that enable manufacturing enterprises to reduce carbon emissions is crucial to achieving China's dual-carbon goals. Leveraging the rollout of China’s New Generation Artificial Intelligence Innovation and Development Pilot Zones as a quasi-natural experiment, this study uses panel data on Shanghai- and Shenzhen-listed A-share manufacturing firms from 2015 to 2023 and applies a staggered difference-in-differences (DID) design. We find that artificial intelligence (AI) adoption significantly reduces firms’ carbon emissions, with effects concentrated on high-emitting manufacturers and insignificant for low-emitting firms. The results are robust to pre-trend and placebo tests, propensity score matching (PSM), and Bacon decomposition. Mechanism evidence suggests that artificial intelligence (AI) reduces emissions by promoting green innovation and easing financing constraints, with stronger effects among state-owned firms and firms in eastern China. These findings inform differentiated carbon-mitigation policies by firm pollution intensity.