How artificial intelligence affects carbon intensity: heterogeneous and mediating analyses

Under the constraints of “carbon neutrality” and “peaking carbon dioxide emissions,” artificial intelligence (AI) paves the way for achieving the dual goals of high-quality economic development and carbon emission reduction. This paper tests the effect and mechanism of AI on carbon emission intensity using panel data from 30 Chinese provincial administrative units from 2006 to 2019. Using mathematical inference and theoretical analysis, this paper concludes that AI can reduce the intensity of carbon emissions. Furthermore, this paper uses the SAC model for baseline regression, taking the spatial spillover effect into account. To ensure the benchmark results’ robustness, this paper also applies the techniques of replacing explanatory variables, endogenous analysis, exogenous shock tests, other regression models, and weight matrix tests. The study demonstrates that: The intensity of carbon emissions will decrease by 2.79% for each unit increase in the proportion of AI. Industry heterogeneity demonstrates that resource-intensive AI increases carbon emission intensity; capital-intensive AI does not currently suppress carbon emission intensity; labor-intensive AI can significantly reduce carbon emission intensity; and technology-intensive AI is a viable means of reducing carbon emission intensity. Regional heterogeneity demonstrates that AI has a varying inhibitory impact on the intensity of carbon emissions, with a higher coefficient in the northeastern and eastern regions. Mechanisms exploration concludes that AI helps to reduce the ratio of coal-based energy consumption, enhance energy use efficiency, and upgrade the industrial structure, thus reducing carbon emission intensity. Based on the aforementioned findings, this paper proposes policy measures to enhance the system for reducing carbon emissions. These measures include the implementation of intelligent technology, the enhancement of energy utilization effectiveness through the advancement of artificial intelligence, and the promotion of interdisciplinary collaboration between intelligence and industrial fields.