Artificial intelligence has led to major breakthroughs in medicine and other fields, but that progress comes with a cost. The technology requires a large amount of resources, including water and electricity. That has led to concerns that this technology could harm arid states like New Mexico if it’s left unregulated.
State Rep. Christine Chandler, D-Los Alamos, expressed concerns during the interim Science and Technology Committee meeting on Monday about the resource demands of artificial intelligence.
“It’s agreed that AI uses a lot of power,” she said. “And it’s agreed that AI uses a lot of water.”
She asked what role the government should play in addressing those issues.
“This is a significant issue for us in the desert,” she said.
AI experts Patrick Bridges and Manel Martínez-Ramón defended the technology to the committee and spoke about efforts to reduce the consumption nature of AI.
Bridges is the director of the University of New Mexico’s Center for Advanced Research Computing and Martínez-Ramón is a professor in UNM’s Department of Electrical and Computer Engineering.
Currently, about 4 percent of the electricity generated in the United States goes to powering data centers that support artificial intelligence. That could increase to 9 percent by the end of the decade, Newsweek reported in August.
Chandler said that there are concerns in Los Alamos about resource adequacy as the national lab’s energy demands increase.
“We know there’s talk about…putting in a redundant electric line, which of course is being opposed by every environmental group in the state that I can think of,” she said.
Chandler said that while AI is something people want in New Mexico, it comes with a large resource footprint and is mostly unregulated in the state.
While she said that AI “has the potential for doing a lot of good things,” she expressed concerns about the cost in terms of resources. She said these costs seem to be getting swept under the carpet.
“If this were a discussion about oil, everybody in this room, well, many of us would be saying methane rules, and we can’t have them by schools and we can’t do this,” she said. “And you know what, oil brings in a lot of good to us too.”
She spoke about how oil can help heat houses, power cars and provide “huge amounts of money” to the state.
Chandler said there needs to be more discussion about the costs of bringing AI into New Mexico where there is limited water and where energy can be a “huge issue for some of us who live in little communities that have to share our electricity with a gigantic, mammoth facility.”
“We should be thinking in terms of those kinds of things, instead of saying, ‘Oh, the companies really want to be helpful, if it’s economically advantageous to them to be helpful,’” Chandler said.
New Mexico’s potential for solar and wind energy is one reason why companies are considering locating data centers to power artificial intelligence in the state. While it is not specifically AI related, one of the largest electrical customers in New Mexico is the Facebook data center near Los Lunas.
But that is only one reason why companies focused on artificial intelligence might choose New Mexico to house their data centers.
“There are reasonable ways to do cooling of big data centers in New Mexico that there aren’t always in other places. So that makes us very attractive,” Bridges said. “We have inexpensive property costs in much of the state. We don’t have many of those pesky natural disasters that a lot of other states are suffering with, I’m happy to not deal with tornadoes and hurricanes and these kinds of things.”
Power demand for data centers and artificial intelligence has soared since 2016 and that can have implications as the world looks to address climate change.
This new demand for electricity means that more resources need to be brought online to power these data centers and fossil fuel generation sources may be expanded or kept running longer to prevent electricity shortages.
Meanwhile, AI companies are working to present themselves as solutions in the fight against climate change. And, in some areas, AI is helping address challenges like the intermittent nature of wind and solar energy. AI already helps utilities better manage their generation assets to meet regional demands.
Bridges and Martínez-Ramón said that tech companies are working to make AI more energy efficient and reduce its water consumption.