Questions arise about the environmental impact of AI production

The recent rise of artificial intelligence (AI) has caused uncertainty among many people as questions arise about what the technology’s impact on humans will be. The effects of the use of AI are expected to have a significant impact on many types of society, one of which is the environment.

AI relies on data centers to run, and these centers use a lot of electricity, which in turn can produce carbon dioxide. On top of emissions, the process of cooling these data centers takes a lot of water and creates steam.

“You’re seeing new trends coming and I think energy consumption is headed towards that [AI] now it’s amazing,” Dargan Frierson, a UW associate professor of atmospheric and climate science, said. That growth has been faster than I thought it would happen, and I think like that for a lot of people.” The environmental impact of AI is largely dependent on how electricity is generated for data centers to operate. For example, AI data centers that rely on sustainably generated electricity will have a much different environmental impact than those that rely on carbon generated electricity.

“[Companies are] trying to build​​​​these large data centers in places with a lot of renewable resources,” said Vikram Iyer, a UW associate professor of computer science and engineering. “And in some cases that’s also it puts a lot of pressure on the power grid.”

AI can analyze large amounts of information, which could have positive implications for understanding and responding to the effects of climate change, according to Léonard Boussioux, a UW professor of information systems and operations management at the School of Foster’s Business.

AI is able to identify areas where pollution is prominent in the ocean so that it can be cleaned up more efficiently, analyze icebergs to determine how much they are receding, and can guide drones to drop seeds to plant trees quickly and efficiently.

The relationship between the fields of AI and sustainability is explored in the Computing for the Environment (CS4Env) program at the UW.

“The Computing for the Environment Initiative is something we started a few years ago to try to get people from the UW involved in research projects related to the environment,” Iyer, who is director of CS4Env, said. said.

Students are also working on using AI as a way to create a sustainable world. Computing for Environmental and Social Advocacy (CESA) has worked directly at the intersection of AI and the environment.

“Our plans include integration [machine learning] algorithms for conservation data analysis and collection, modeling pollution in vulnerable communities, and learning how modeling approaches can lead to new solutions to climate change, Shreyan Mayukh Mitra, a second-year computer science student and president of CESA, said in an email.

However, not everyone is using AI consistently despite attempts to use it to help the planet.

“Besides finding better environmental solutions, there are also people using these new tools to get more fuel from the ground and get [them] get out as soon as possible,” Frierson said.

With Seattle considered one of the biggest growth areas in the AI ​​industry, it is expected that questions will continue to arise about the impact it will have on the environment. Boussouix said that through dialogue and education, he hopes to continue to spread awareness about the impact of AI.

“Just to make people aware that those [AI] models really use tons of water and electricity is a great start. It’s not enough, but it’s a start,” Boussioux said.

Contact reporter Brendan Sheehan at science@dailyuw.com. X: @brendan3sheehan

Like what you read? Support high-quality student writing by donating here.