Artificial intelligence promotes urban green innovation in China with regional spillovers and asymmetric climate-risk thresholds

We systematically examine the impact of artificial intelligence on urban green innovation in the context of climate risk. The results demonstrate that artificial intelligence substantially promotes green innovation, an effect primarily transmitted through expanded openness to trade and enhanced scientific-research investment, with the digital economy playing a positive moderating role. Spatial analysis reveals that artificial intelligence advances in one city generate significant positive spillovers that enhance the green innovation performance of neighboring cities. This spatial pattern exhibits pronounced regional heterogeneity: the western region displays significantly positive indirect effects, whereas the eastern and central regions are dominated by direct effects. Threshold effect analysis uncovers differentiated responses to physical climate risks: artificial intelligence’s benefits are weakened once extreme high-temperature days exceed the threshold but strengthened once extreme low-temperature days exceed their threshold, while extreme precipitation exerts no discernible threshold influence. Finally, higher transition risk is found to amplify the beneficial impact of artificial intelligence on green innovation.