Exploring the Impact of Artificial Intelligence on Carbon Emission Efficiency: Green Innovation and Nonlinear Mechanism

The vigorous development of artificial intelligence (AI) has effectively facilitated green innovation models and exerted a profound impact on sustainable development. Utilizing panel data from 269 cities spanning the period 2011 to 2021, this study constructs a comprehensive indicator system of AI development and explores the mechanism through which AI influences carbon emission efficiency (CE) from the perspective of green innovation. The findings are as follows. First, the development of AI significantly enhances CE, and this finding remains robust after a series of robustness tests. Second, AI primarily exerts a positive effect on CE through green technological innovation and green innovation in production and consumption processes. Additionally, government-led green guidance and regulation can strengthen the positive effect of AI on CE. Notably, multiple nonlinear relationships exist between AI and CE. Specifically, AI exhibits a single-threshold effect: in the early stages, AI inhibits CE, but only after surpassing a certain threshold does it begin to exert a positive impact. Similarly, as digitalization deepens, AI's marginal contribution to CE declines. Furthermore, a dual-threshold effect is identified in the context of environmental regulation: after crossing the first threshold, AI's marginal positive effect on CE strengthens; however, beyond the second threshold, this effect weakens. Lastly, AI not only positively influences local CE but also generates positive spatial spillover effects on neighboring regions. This study offers new theoretical insights into the mechanisms linking AI, green innovation, and CE, with practical implications for government policies aimed at green emission reduction.