Huge investments in artificial intelligence are forcing big tech companies to push back on their carbon neutral goals.
At that point July 2024 Environmental ReportGoogle no longer claims to be carbon neutral, instead aiming to be carbon free by 2030. Similarly, Microsoft's indirect emissions have increased by 30%, despite its ambition to be carbon negative by 2030, the company's report said. May Report.
Like others such as Meta, the company cited increased use of AI as the reason it needed new data centers to meet growing energy demands. These announcements have raised questions about the impact AI will have on the environment.
For example, AI requires significantly more energy to generate responses than a typical online search, and the heat output from servers also increases energy consumption in data centers.
“When you do an AI search, the servers get much hotter than a traditional internet search, so they require additional cooling with HVAC units and water,” said computer science major Ellie Knapp, who will graduate in 2026. “That means that as AI becomes more accessible and popular, we'll need more HVAC and water to cool the servers, which will incur environmental costs.”
Knapp worked as a technical intern on Comcast’s sustainability team this summer, and throughout his internship he used AI to measure energy consumption metrics to help maximize efficiency.
“AI is an incredibly expensive endeavor because it requires a lot of data to do pattern recognition,” Knapp said. “And that's where the concern is. [that there are] “Everybody wants to have their own competing AI capabilities, so we're seeing a lot of diversity in AI development.”
Rising energy demand threatens efforts to transition to more sustainable energy sources. The Washington PostRising demand for electricity has slowed efforts to shut down coal-fired power plants in Kansas, Nebraska, Wisconsin and South Carolina.
“Most of our electricity is generated by burning methane, so AI is one of the factors slowing progress in reducing emissions,” said Richard Feldman, associate professor and chair of the School of Environmental Science and Policy at Marist College.
International Energy Agency Due to the development of AI, data center energy consumption is expected to reach 6% of total electricity use in the United States by 2026. As of 2023, fossil fuels still produce 83% of energy in the United States, U.S. Energy Information AdministrationAI energy consumption directly impacts carbon dioxide emissions. However, Falling prices for renewable energy sources Things like solar panels could help alleviate some of the long-term problems.
“Google and everyone in our industry need to make a more concerted effort to run their facilities on sustainably produced electricity,” Feldman said.
What's more, there are ways to make AI more energy efficient through code and algorithms, and Knapp predicts financial incentives will encourage tech companies to pursue these.
“This isn't just an environmental cost to companies, it's money they're spending on energy. It's not economically sustainable, so it's going to force companies to get more creative and design AI in more ethical ways,” Knapp said.
AI also offers benefits in other areas of environmental science, such as tracking pollution and endangered species, and its ability to process large amounts of information can help fill areas with staffing shortages.
“Those two areas of environmental science don't have enough people or funding to really do the work at the level that they need to,” Feldman says, “so I'm hopeful that we're going to see AI as a powerful tool to help these people.”
As such a transformative technology, the long-term impacts of AI on the environment remain unclear. These impacts will be determined in large part by the responses of large tech companies, but individual consumers can also take notice.
“I would urge people to be more intentional about how they interact with AI tools,” Knapp says. “Instead of spamming them with multiple requests, be more thoughtful and try to minimize the number of searches.”