AI can help in the fight against climate change, but if it's not deployed wisely, artificial intelligence could mess things up more

• Scientists are exploring ways artificial intelligence could help respond to the climate crisis.
• There are ways AI is already helping the planet but the tech has to be deployed wisely, experts say.
• Failing to do so risks exacerbating the inequalities that a changing climate will itself drive.

Priya Donti thinks humanity will need artificial intelligence to help us avoid some of the most damaging consequences of a hotter planet.

Donti, the cofounder and executive director of the nonprofit Climate Change AI, studies how power grids can better incorporate renewable energy. She expects AI will play a part.

"It is hard to envision a future decarbonized power grid that does not fundamentally have AI and machine learning as a key workhorse there," Donti told Insider.

She said the question of AI and sustainability often gets framed as how AI can supercharge efforts to fight the climate crisis versus the technology's rapacious appetite for energy-intensive computing power. Donti said we should instead be thinking more about how to wisely deploy AI.

Questions around AI in general have been mushrooming in the six months since ChatGPT became publicly available, in what's now seen as a black swan event. The White House, for its part, this week issued a request for information on how AI might shape the nation's goal of halving greenhouse-gas emissions by 2030, among others.

In sustainability circles, there appears to be more attention to the ways the technology can give environmental efforts a digital assist.

AI can use satellites to look for poaching and illegal logging. It's being used by the United Nations and companies like Google to speed response times for and even predict disasters such as floods. A California startup called Refiberd is relying on AI-powered robotics to sift through discarded textiles with 95% accuracy so they can be recycled.

And there's an app that uses AI to predict when and where crop-damaging locusts might sweep across the horn of Africa and other parts of the continent. The app, called Kuzi, is named after the Swahili word for the wattled starling, which feasts on the insects.

Yet for all these wins, Donti said, AI also has been used to sell consumers more stuff and to help oil and gas companies extract more fossil fuels — work that the tech companies providing this computing horsepower have been criticized for.

Donti, an incoming professor at the Massachusetts Institute of Technology, said we need to add AI into the sustainability arsenal, though that doesn't mean the bots will be taking over things like operating the grid. Instead, Donti expects AI will handle some background processes, but humans will still be making the big calls about how something like the power system operates.

She likens it to humans steering a car down the street: Drivers are still in charge but modern vehicles automate many lower-level functions.

Donti cautioned that simply adding AI to an environmental problem won't necessarily mean it gets solved. And doing so could worsen a situation if spending on AI draws disproportionate funding away from something fundamental but effective like beefing up insulation in drafty buildings.

"AI is a really powerful tool for accelerating climate action," she said, adding, "It's not going to solve every problem."

There are also risks that AI could help the already powerful further consolidate their influence. An entrenched player in the energy market like an oil or gas company could use AI to help discover new caches of fossil fuels and keep prices low; that could make it harder for clean-energy competitors. Donti said AI has an "accelerating effect" that often lets an entity become more effective at achieving its objective.

"If you're now only having this technology that's more accessible to actors in society that are more powerful, it potentially exacerbates gaps because now there's a certain subset of people in society who can use AI and its multiplicative effect and other people who can't," she said.

Phil De Luna, a former director of the National Research Council of Canada and the author of a book on how AI can help scientists more quickly discover useful materials, also raised concerns about a fair deployment of the benefits of AI. He said it needs to be developed so that the people hardest hit by a changing climate — often those who are poorer or living in the Global South — can gain ground.

Even AI tools that are deployed with good intent risk exacerbating societal inequities, he said. For example, De Luna said AI could be used in regions where wildfires are common to determine where to develop firebreaks. Many homeowners could benefit, but if those tools aren't available to residents of poorer areas, those groups could suffer more.

"If artificial intelligence is used in sustainability to create moats and technology barriers and profit, then we will have failed in the actual purpose, which is to prevent climate change, and ultimately loss of life," De Luna said.

De Luna is most excited about using AI with laboratory robots to hasten the discovery of materials and technologies that could mean better batteries, new hydrogen materials, and improved systems for capturing carbon-dioxide emissions. Tech like solar panels and wind turbines took decades — far too long — to go from invention to widespread use, De Luna noted.

"We have less than 30 years to find and scale the next clean technology," he said. AI will prove essential to that. "It's the tool that we need, now more than ever, for us to reach our climate goals."