AI for Mitigating Climate Change: An Account of Uses, Difficulties, Morality, and Prospects

Abstract

Artificial intelligence (AI) is now being used as a core technology in global climate change mitigation efforts, with evidence showing that AI data-center cooling optimization reduced energy use by 40 percent in data centers, the share of renewables in the grid increased from around 27 to 29 percent between 2019 and 2020, and renewable generation grew by more than 8 percent in 2021.The same literature that celebrates these gains also reveals the costs of AI: training large-scale models like GPT-3 comes with a high energy cost, and climate-relevant data are not evenly distributed between regions, potentially leading to a disproportionate benefit for those already well-resourced.Existing reviews usually distinguish between the technical capability and the governance risk of AI, and try to discuss these two strands separately without much synthesis.This narrative review fills this void by summarizing the contribution of AI in four interrelated areas: optimization of renewable energy, tracking of carbon emissions, environmental monitoring, and sustainable development in urban and agriculture sectors and highlighting that this is not an automatic benefit to the climate.The data that trains AI, the energy efficiency of the systems that run AI, and transparency and equity of institutions that govern AI are the three main factors that determine its net effect, according to evidence.Therefore, this review argues that AI functions best not as a climate solution in itself but as a force multiplier for climate policy, capable of accelerating well-governed strategies and equally capable of amplifying inequity and emissions where governance is absent.