Artificial intelligence and climate risk: A double machine learning approach

Abstract

We study the use and environmental impact of AI technologies. We propose a measure of the country-level AI development index. Utilizing the double machine learning method, we discover a net mitigating impact of AI on climate risk. Mechanism analysis indicates that this influence primarily stems from advancements in resource utilization efficiency, the promotion of green innovation, the reinforcement of environmental policy effectiveness, and the augmentation of green finance. Heterogeneity analysis reveals that the mitigating effect of AI on climate risks is predominantly observed in developed countries and those with better institutional environments. Our results imply that while AI overall reduces climate risks, it can also contribute to the exacerbation of climate-related inequalities.