China Population, Resources and Environment, 2023, Issue 10

Authors: LYU Yue, MA Minghui, CHEN Yongchang, ZHANG Haotian

Abstract: Based on the construction of a theoretical model of artificial intelligence (AI) affecting pollution emissions decisions, this pa‐per empirically tested the impact and internal mechanisms of AI on pollution emissions by combining industrial robot data provided by the International Federation of Robotics and Chinese provincial panel data from 2006 to 2019. We found that: ① AI significantly reduced the pollution emission intensity, which was still valid after a series of robustness tests and using the instrumental variable method to overcome endogeneity. ② Our mechanism analyses showed that the pollution reduction effect of AI was mainly realized by promoting technological innovation, increasing investment in emission reduction equipment, and replacing low-skilled labor. Specifically, from the perspective of technological innovation, AI could effectively increase the number of patent applications; from the perspective of emission reduction investment, AI could significantly promote the investment in emission reduction equipment; and from the perspective of labor structure, AI could effectively replace low-skilled labor and optimize the labor input structure. ③ The heterogeneity analyses indicated that the environmental optimization effect of AI was more prominent in areas with stricter environmental performance assessments and in those with higher pollution emission intensity. ④ Further analyses showed that AI had the dual role of ‘emission reduction and output enhancement,’ meaning that it could promote the coordinated development of economic growth and environmental governance.Besides, based on firm-level micro-data, we found that AI could significantly reduce corporate pollution emission intensity. Our findings provide empirical evidence and theoretical support for AI-empowered green transformation and high-quality development.

Key words: artificial intelligence; green development; industrial robot penetration