Major tech companies have been promoting the idea that generative artificial intelligence can play a substantial role in reducing greenhouse gas emissions and helping address climate change. In late 2023, for example, Google claimed that AI might cut global emissions by 5–10% by 2030—a statistic widely shared across media and policy discussions. But a new analysis by energy and climate analyst Ketan Joshi finds that most such claims have little to no solid backing from peer-reviewed studies. Of 154 specific statements about AI’s climate benefits examined in the report, only about one-quarter cited academic research, and more than one-third provided no evidence at all. Most claims also conflate older, less energy-intensive forms of machine learning with the much larger and energy-hungry generative models that have driven recent growth in the tech sector, tells Wired.com.
The report’s findings suggest that many corporate narratives about AI saving the planet are speculative at best and potentially misleading at worst. Prominent industry figures, including former Google executives and leaders of major AI research firms, have publicly framed AI as a necessary tool for tackling climate goals, even as the infrastructure needed to support large language models is expanding rapidly.
Critics argue that this enthusiasm obscures AI’s own environmental footprint. The data centers that power generative AI services consume substantial electricity and water and often rely on energy sources that emit greenhouse gases. Rather than presenting verified climate solutions, the focus on AI’s future potential may divert attention from more immediate and proven strategies for emissions reduction.
Experts quoted in the article call for greater transparency from tech firms about the real environmental costs of AI deployments and more cautious communication about expected climate impacts. Without rigorous evidence and open data on energy use and emissions, claims about AI’s climate benefits remain unsubstantiated.