New report dismisses claims around climate-friendly AI as greenwashing

Most AI leaders enthusiastically claim the new technology will soon get so smart that it will solve the problem of climate change. But a new study convincingly demonstrates that the majority of these claims can be classified as greenwashing.

The world’s largest tech firms have long claimed that even though building AI requires a huge amount of energy, the product will end up being so great that it will solve the climate crisis for us all.

For instance, Google’s former boss Eric Schmidt said already in 2024 that we shouldn’t worry about how power-hungry AI is because, actually, only AI will probably be able to mitigate the crisis.

Microsoft founder Bill Gates is also confident that AI will ultimately identify ways to help cut power consumption and drive the transition to sustainable energy.

Speaking on the sidelines of the India AI Impact summit, OpenAI CEO Sam Altman also dismissed as exaggerated some of the most viral claims about AI’s environmental toll, mockingly adding: “Humans use a lot of energy too.”

But Ketan Joshi, an energy analyst, claims in a new report that most of such claims are wrong.

Moreover, at a time when the world is asking how we’re going to power the AI boom without deepening the strains on water, energy, and climate goals, AI companies might be intentionally greenwashing the issue.

“I guarantee you’ve heard the harms of data centre expansion justified on the grounds that AI will 'solve climate change.’ These range from sci-fi claims of superintelligence through to detailed reports stacked with hundreds of examples of ‘AI for good’ helping energy, transport, and industry cut emissions,” says the analyst.

“But I’ve created a new report that, for the first time, interrogates both the logic and the evidence for this claim.”

After reviewing 154 claims about supposed AI climate benefits from tech giants, the International Energy Agency, and climate researchers, Joshi reached two central conclusions.

First, he found that most of the “benefits” tend to relate to older, smaller, and leaner forms of machine learning, such as predictive models and computer vision. This is called “traditional AI.”

However, we now know that most of the new harm is likely stemming from consumer generative AI over-deployment. Tools that chat, create images and videos, and make music are a lot more power-hungry.

This is unfair, Joshi says, and compares such claims to pulling the wool over people’s eyes. According to the analyst, there are essentially no examples where generative AI systems such as ChatGPT, Gemini, or Copilot lead to verifiable emissions reductions.

We now know that most of the new harm is likely stemming from consumer generative AI over-deployment. Tools that chat, create images and videos, and make music are a lot more power-hungry.

Secondly, the report concludes that most of the evidence that even traditional AI would help with the climate crisis lacks support.

Stunningly but predictably, of the 154 climate benefit claims that were examined in the study, only 26% cited published academic papers.

Thirty-six percent lacked any citations for their claims, while the remaining papers that included citations largely relied on corporate publications (29%). The rest cited media reports, NGOs, institutions, and unpublished research papers.

“The narrative of a gigatonne-scale shift in global emissions that ‘offsets’ the harm of generative AI is not supported by the results of this analysis,” writes Joshi, additionally reminding that the current boom is actually powered by fossil fuels.

Unlock more exclusive Cybernews content on YouTube.