AI climate hopium is going to burn the planet
Far from solving the climate crisis, AI is making it worse.
During COP28 in Dubai, Kate Brandt, Chief Sustainability Officer at Google, was interviewed by the Wall Street Journal. “I think AI has a really major role to play in addressing climate change,” she said, adding that AI is an opportunity to “supercharge climate solutions in this decisive decade”. There are “three big opportunity spaces,” according to Brandt, “information, optimisation, and prediction”.
Brandt talks about “nature-based solutions”. She says that we are at an “inflection point in the carbon markets”. But, she adds, “there are opportunities for technology to play a really critical role”.
Predictably, Brandt doesn’t talk about reducing flights, addressing over-consumption, regulation of Big Polluters, or the problems associated with extractivist capitalism.
Neither does she mention the huge increase in Google’s greenhouse gas emissions in recent years. Since 2019, Google’s emissions have jumped by almost 50%.
This jump is largely a result of the large energy demand of the data centres needed for AI.
Energy and AI
Earlier this month, the International Energy Agency published a 304-page report titled “Energy and AI”. The report notes that a typical AI data centre uses as much electricity as 100,000 households. The largest ones currently under construction will consume 20 times as much.
The electricity consumed by data centres is set to more than double over the next five years. By 2030, data centres might consume 945 terawatt-hours, or slightly more than Japan’s electricity consumption today.
Data centres run 24 hours a day, seven days a week. Fossil gas plays a large role in powering data centres. Chevron is planning new fossil gas power plants to supply electricity directly to data centres.
In an article for Switchyard, journalist Mya Frazier documents the impacts of data centres in drought stricken Ohio. There are more than 100 data centres in Ohio.
The Big Tech companies signed secret deals for discounted electricity supply and tax breaks.
Rural farming communities are seeing high-voltage power lines cutting across fields, streams, and wetlands.
The 55-metre-high steel towers, need 50-metre-wide corridors. Trees are being removed to make way for the lines. Habitat of endangered or threatened bats will be destroyed.
“The lines will cross within the range of the state-endangered northern harrier hawk,” Frazier writes, “which prefers open habitats and hunts over grasslands.”
The data centres can use as much five million gallons of water per day to keep the computers cool. “How did we become so oblivious to our own peril?” Frazier asks.
Our oceans are rising and acidifying, our forests burning, our rivers flooding, our ice caps melting, our soil degrading, our harvests scorched by drought. How did we allow ourselves to be gripped by this collective death drive?
Part of the reason for this “collective death drive” is the story that AI proponents tell about AI coming to the rescue and solving the climate crisis.
AI climate hopium
IEA’s report is optimistic that AI could eventually reduce energy emissions globally — by more than the jump in emissions caused by data centres.
Perhaps the most extreme version of AI climate hopium comes from Eric Schmidt, former CEO of Google. “My own opinion is that we’re not going to hit the climate goals anyway because we’re not organised to do it, and the way to do it is the things we’re talking about now,” Schmidt said in October 2024. “I’d rather bet on AI solving the problem than constraining it and having the problem.”
But as James Temple points out, writing in the MIT Technology Review, “the climate promises of AI sound a lot like carbon offsets”. Temple writes that,
There’s something familiar about the suggestion that it’s okay to build data centers that run on fossil fuels today because AI tools will help the world drive down emissions eventually. It recalls the purported promise of carbon credits: that it’s fine for a company to carry on polluting at its headquarters or plants, so long as it’s also funding, say, the planting of trees that will suck up a commensurate level of carbon dioxide.
Big Tech companies are, of course, also big buyers of carbon offsets.
Big Tech is particularly fond of eucalyptus plantations and other tree planting schemes in Brazil, including in the savanna ecosystem of the Cerrado. As Lucy Rowland, an ecologist at the University of Exeter, told MIT Technology Review,
“Under no circumstances should planting eucalyptus ever be considered a viable project to receive carbon credits in the Cerrado.” 1
The reality is that we already have the technology to address the climate crisis. It is not a technological problem. It’s a political problem.
Rather than building ever more, and ever bigger, energy guzzling data centres, in the hope that at some point AI might solve the energy crisis, we need to find ways of keeping fossil fuels in the ground.
We urgently need to reduce emissions. Not massively increase them to feed Big Tech’s machines.
It could be decades before AI comes up with a way of addressing the climate crisis. As Temple writes, there is no mechanism to hold Big Tech accountable if it drives up emissions but never comes up with a solution.
Under the Trump administration, there’s little reason to believe that US companies, at least, will face much government pressure to use these tools specifically to drive down emissions. Absent the necessary policy carrots or sticks, it’s arguably more likely that the oil and gas industry will deploy AI to discover new fossil-fuel deposits than to pinpoint methane leaks.
And even if AI does eventually come up with ways of cutting more emissions than it creates, those future reductions will do nothing to address the heating produced by the past emissions from data centres.
In September 2024, James Temple wrote an article for the MIT Technology Review titled, “Sorry, AI won’t ‘fix’ climate change”. He concludes that,
“[T]he one thing we can state confidently about generative AI is that it’s making the hardest problem we’ve ever had to solve that much harder to solve.”
Carbon offsets, AI-driven climate solutions, green technology transition, all these frameworks assume that climate change and rocketing emissions as something solvable by techno-managerial innovation, while failing to recognize fossil fuels and massive energy consumption underpinning the contemporary technological innovation landscape. It's just hilarious to see big tech like Google propagating their climate solutions without even any conscience on their role in exacerbating the problem.
"Hopium" copyright James Hansen
AI (used to mean Artificial Insemination) is one step of Progress TOO FAR whose main work is justifying its existence.
Eucalyptus is highly inflammable.
Yes, reducing emissions is a political problem. But neo-liberal ideology promotes "market mechanisms" as being more friendly to "economic freedom" than government regulations. So we see various market mechanisms trampling each other in the mad rush to pour out money on the environment problem. AI is a mega-increase in the bad kind of GDP growth that, like arms manufacture, does nothing to increase actual physical or social progress.