Are AI and environmental technology innovations converging?

Artificial intelligence (AI) is increasingly invoked as a means of advancing climate and environmental solutions, yet how it shapes environmental technology (ET) development remains poorly understood. This study addresses that gap by asking: To what extent are AI and ETs converging, and which specific domains are driving this integration? To investigate these questions, we have constructed a new dataset of nearly 8000 AI-environmental (AI-ENVI) patents filed at the USPTO from 2003 through 2023. Drawing on this dataset, we perform semantic reclassification followed by forward citation mapping to identify influential innovations and to evaluate prominent domains of technological convergence. We find that three fields dominate: renewable energy optimization, electric vehicles, and fossil fuel efficiency/industrial decarbonization. To probe further, we examine metrics of novelty, disruptiveness, and generality. Novelty is highest in grid-level energy storage optimization; disruptiveness peaks in blockchain-based energy trading; and generality is strongest in the former as well as energy demand forecasting. Evidence that AI applications for incumbent industries attract high forward citations, and that carbon capture and storage (CCS) emerges as a disruptive subcategory, underscores the need for appropriate policy mechanisms to avoid carbon lock-in within the ET–AI nexus.