As the Los Angeles fires rage on, displacing nearly 180,000 people and destroying over 9,000 buildings, many social media users have pointed the blame at an unusual target: ChatGPT.
ChatGPT and other AI models have a massive carbon footprint, which contributes to climate change. Climate change is responsible for drier conditions and warming temperatures, which in turn creates the perfect environment for wildfires to spark and spread when picked up by strong winds.
So in a way, ChatGPT is part of the problem by contributing to climate change. But no, ChatGPT didn’t start the fire, and it’s not the reason why the city ran out of water.
Martin Adams, former general manager of the Los Angeles Department of Water and Power told the Los Angeles Times, “The [water] system has never been designed to fight a wildfire that then envelops a community.” Firefighters ran out of water because the system wasn’t built to pump out that much water over a sustained period of time, not because it was misappropriated by data centers.
The online conversation connecting ChatGPT to the LA wildfires was never meant literally accuse AI models of starting fires. Instead, it has tied real-time relevancy to an growing issue which is AI’s impact on the environment.
“We don’t need AI ‘art.’ We don’t need AI grocery lists. We don’t need AI self-driving cars. We don’t need ChatGPT or Gemini or Grok or DALL-E or whatever ‘revolutionary’ technology already exists inside our own human brains. We need the earth,” wrote makeup artist and activist Matt Bernstein in an Instagram post that has gained almost 500,000 likes.
So, just how much water does ChatGPT use?
Tech companies like Google, Microsoft, and OpenAI are evasive about their energy usage. But water consumption can be accessed through public records and estimated through calculations by researchers. A 2023 investigation from the Associated Press discovered the Microsoft data centers in Des Moines, Iowa used to train OpenAI’s GPT-4 needed 11.5 million gallons of water to cool its servers, which is 6 percent of the district’s total water supply.
A recent study by The Washington Post and University of California, Riverside found that a 100-word email generated by ChatGPT requires roughly the equivalent of a bottle of water, or 519 milliliters. Using ChatGPT once a week for a year by 16 million people uses 435,235,476 liters of water.
In short, that’s a lot of water. And AI’s thirst isn’t slowing down anytime soon. One 2023 study by UC Riverside estimates that AI could consume between 4.2 and 6.6 billion cubic meters of water in 2027, which is more than the annual water withdrawal of half of the UK.
What about the rest of ChatGPT’s energy consumption?
In Bernstein’s viral Instagram post, he wrote, “One search on ChatGPT uses 10x the amount of energy as a Google search. Training one AI model produces the same amount of carbon dioxide as 300 round trip flights between New York and San Francisco and five times the lifetime emissions of a car.”
These findings come from a 2019 University of Massachusetts, Amherst study quantifying the environmental cost of GPT-2, which was an early AI model from OpenAI. And ChatGPT usage has exploded since then, with 300 weekly active users according to OpenAI’s own reporting as of December 2024.
In terms of electricity input required to support its massive computing power, Sajjad Moazeni, a University of Washington assistant professor of electrical and computer engineering estimates ChatGPT usage of 1 Gigawatt per hour (GWh) a day, “which is the equivalent of the daily energy consumption for about 33,000 U.S. households.”
ChatGPT isn’t directly responsible for the Los Angeles fires, but witnessing the devastation in real time brings a visceral reality to the environmental cost of using AI to write an email.
And if you want to help but aren’t sure where to start, the Los Angeles Times has a great guide here.
