AI Could Be Harnessed to Cut More Emissions Than It Creates

Scientific American

2 days ago

CLIMATEWIRE | Artificial intelligence could cut global climate pollution by up to 5.4 billion metric tons a year over the next decade if it's harnessed in ways that would improve transportation, energy and food production.
Those reductions would outweigh even the expected increase in global energy consumption and emissions that would be created by running power-hungry data centers associated with AI, according to research by the Grantham Research Institute that was published in the journal npj Climate Action.
'The key will be to channel practical AI applications towards key impact areas to accelerate the market adoption rate and efficiency of low-carbon solutions,' the study said, noting that governments will have a vital role to play.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
'Governments must regulate AI to minimise their environmental footprint,' the study said, pointing to the need for energy efficient AI models and the use of renewable energy to power data centers.
It also says governments should invest in AI infrastructure and education in developing countries, 'ensuring that the benefits of AI are shared equitably.'
The use of AI has boomed in recent years and President Donald Trump has made its expansion a priority as he seeks to outcompete China on advanced technologies. But its voracious use of energy has raised concerns about its climate impact, particularly as Trump and other officials argue that winning the race for AI is a reason to expand fossil fuels.
The International Energy Agency projects that by 2030 data centers will consume twice as much electricity as they do today. Growing energy demands are already challenging the U.S. grid, and oil companies are using AI to find new areas to drill. BloombergNEF has said fossil fuels will provide most of the new power for data centers over the next decade, imperiling efforts to cut carbon pollution.
There are ways to mitigate the damage, the Grantham study said.
It outlines five areas where AI can be harnessed to reduce emissions, including consumer behavior, energy management and technology innovation.
For example, AI can help integrate renewables into the grid by better forecasting supply and demand fluctuations and help the grid distribute energy more accurately, reducing concerns about intermittency. That can increase the uptake of solar and wind and lower the use of polluting backup power sources, the report said.
'Power grids are at the heart of the entire economy, so improving their efficiency reduces emissions across multiple sectors,' Roberta Pierfederici, a policy fellow at the Grantham Research Institute and author of the study, said in an email.
AI can also identify new types of protein to replace meat and dairy in human diets — industries that are heavy emitters. And it can improve transportation by lowering the cost of electric vehicles through battery improvements or encouraging people to switch to shared transport.
Those actions combined could reduce emissions by 3.2 billion to 5.4 billion metric tons of carbon dioxide equivalent annually by 2035 compared to their current trajectory, the study said. That's more than the entire European Union. By comparison, the U.S. released 6.2 billion metric tons of climate pollution in 2023.
Those cuts are not in line with what's needed to keep global temperatures from rising more than 1.5 degrees Celsius since the beginning of the industrial age. But they could keep a check on warming by more than offsetting the emissions released by using AI. The study estimated that energy emissions tied to data centers and AI will reach 0.4 billion to 1.6 billion metric tons of CO2 equivalent over the next decade.
The study does have its limitations, given how quickly the field of AI is changing. The authors acknowledged that they might have underestimated AI's potential to reduce emissions, because they only looked at how AI is applied currently to three sectors. On the other hand, the study didn't consider how energy efficiency gains from AI could lead to increased consumption elsewhere that could drive emissions up.
Pierfederici said while the rise in emissions from data centers is a valid concern, she believes the study makes a strong case for using AI to help tackle rising temperatures.
'That said, governments need to play an active role in guiding how AI is applied and governed, to make sure the downsides are managed effectively and the full potential of AI for climate action is realized,' she added.