AI-generated inquiries can be 10 times more resource-intensive than a Google search.
“AI can consume roughly 10 times as much energy as a Google search for doing the same thing, so that essentially translates into 10 times more carbon emission and 10 times more water consumption and 10 times more [of a] health burden,” said Shaolei Ren, associate professor of electrical and computer engineering at the University of California, Riverside.
Each AI data center relies on tens of thousands of servers that calculate and compute information generated by AI, and each center can use more than 100 megawatts of power daily, roughly equivalent to the consumption of 100,000 households, Ren said.
Junior Liora Kalinovskiy is researching the environmental impactsof AI for her Advanced Honors Research Program.
“I’m super interested in technology and emerging technologies like AI,” said Kalinovskiy. “I’ve always been interested in that, and I think that as a concept, AI is super, super helpful and obviously has almost unlimited capabilities. But I wanted to take a look at the other side of it because there’s both negatives and positives to this technology.”
According to Ren, AI’s main impacts can be categorized as emissions, water consumption and burden on people’s health.
AI relies on energy from fossil fuels, such as coal and natural gas, which creates a carbon emission problem that can have a long-term impact on climate change, Ren said.
AI data centers also use water to cool the heat from their servers and generate electricity.
“[Google consumed] 24 billion liters of water in 2023, so that’s roughly the same as the direct water consumption of [a] major beverage company such as PepsiCo,” Ren said.
AI can also impact people’s health because AI data centers’ use of fossil fuels emits air pollutants such as 2.5 nitrogen dioxide, sulfur dioxide and carbon dioxide, Ren said.
“These are the particles that can penetrate into people’s lungs, and these can create adverse health outcomes like asthma, coughing and heart attack,” said Ren. “Imean, many bad things.”
According to Kalinovskiy, AI is being implemented all over the place, which has made her more conscious of the negative impacts of its usage.
According to Ren, although AI has a large environmental impact, users shouldn’t be scared because when subtracting the costs from the benefits, the overall effects are still positive.
“So, generally, I think we shouldn’t really stop using AI,” said Ren. “AI is a great thing. It’s just that we don’t want to use AI for meaningless things.”
At this point, there is not much individual users can do to offset the environmental impacts caused by AI, Ren said.
If AI companies are transparent about their resource consumption, individuals can make informed decisions about their usage, but there is no such information in the public domain, Ren said.
Kalinovskiy uses AI about five to six times a week to brainstorm or explain math problems.
“AI definitely has a lot of pros and a lot of cons … it could be super useful, but also, the amount of energy it’s consuming could be detrimental,” said Kalinovskiy. “Because, let’s be honest, we’re not doing too great outside of AI with energy consumption and just climate change in general. So there definitely has to be a balance to that when exploring these new technologies.”