The impact of AI on carbon emissions: evidence from 66 countries

Abstract

This study aims to address debate in previous studies on whether AI has a positive or negative effect on carbon emission reduction. We used quantile regression and PSTR models to study the diverse impacts of AI on carbon emissions in 66 countries from 1993–2019. There were three main findings in this paper. First, the impact of AI on carbon emissions varies across countries, and its effect on carbon reduction is mainly found in high-carbon emission and high-income countries. Second, the industrial structure environment of different countries affects the role of AI in carbon reduction, with its marginal effect in limiting emissions decreasing with the rise of secondary industrial structures. Third, the impact of AI varies in countries based on their different demographic structures. The marginal effect of AI on carbon emission reduction increases in places with older populations. This study offers unique insight into the heterogeneous impact of AI on CO2 emissions. Our analysis confirms the importance of industrial and demographic structures in promoting carbon emission reduction. We provide effective policy recommendations for economic development and environmental governance.