Most claims about AI’s climate benefits are misleading and unproven, according to a report backed by several environmental organisations. We spoke to the report’s author, Ketan Joshi, to help us make some sense of the growing rift in the climate movement around “AI”.

You’ve no doubt heard some variation of the claim that “AI” will somehow solve the climate crisis. You’ve also probably heard the warnings: that the rapid expansion of all the data centres needed to deploy more of this technology comes with considerable environmental costs. There’s a widening rift, with competing voices making wildly different arguments about the tech’s climate impacts.

Is it all just hot air? Is anyone backing up their claims with data? To try and get some answers, independent climate analyst Ketan Joshi recently published the first report of its kind. In it, he interrogates the argument that “AI” will have net-positive benefits for the climate. Supported by a consortium of environmental organisations — including Stand.earth, Beyond Fossil Fuels, and Climate Action Against Disinformation, among others — he looked into the green claims companies make about “AI.” He checked the footnotes and scoured the available data to try and come to some conclusions.

In the end, Joshi found zero verifiable evidence that the new wave of consumer generative “AI” was reducing emissions in any way, despite what the Big Tech companies would have you believe.

Yes, some forms of “AI” do have positive environmental benefits, as we get into in the conversation that follows. But Joshi found that those benefits were being misattributed to generative “AI” systems like ChatGPT, which consume up to 13 times more energy than the lower-energy technologies actually delivering the climate gains.