Institutional adaptability, skill-bias technological shifts, and energy efficiency in global decarbonization pathways: Exploring the role of artificial intelligence patents

The catalytic role of artificial intelligence patents (AIP) in accelerating global decarbonization pathways—through structural reductions in CO2 emissions—remains underexplored. Leveraging a panel of 29 countries (2005–2023), this study explores the effect of AIP on carbon mitigation. We find robust evidence that AIP drives significant emissions reductions, with results persisting across instrumental variable regressions, sensitivity analyses, and alternative model specifications. Mechanism tests reveal three decarbonization channels: AIP drives technological shift by favoring skilled labor and automating carbon-intensive routine tasks; institutional adaptability magnifies these decarbonization benefits; and AIP enhances systemic energy efficiency, evidenced by substantial reductions in both primary energy intensity and per capita CO2 emissions. Heterogeneity tests demonstrate that AIP's decarbonization impact is most pronounced in economies geographically proximate to the USA (the global AI innovation hub), in lower-income nations, and in countries with high fossil fuel dependency. Within AIP categories, energy-management patents yield the highest decarbonization returns. These findings underscore AI patents as critical yet underutilized tools for achieving net-zero transitions, with policy implications for targeted intellectual property incentives, global technology transfer, and institutional reforms to amplify decarbonization synergies.