The Google and Its AI Peers Guzzle Water. Yep, Guzzle
October 6, 2023
Note: This essay is the work of a real and still-alive dinobaby. No smart software involved, just a dumb humanoid.
Much has been written about generative AI’s capabilities and its potential ramifications for business and society. Less has been stated about its environmental impact. The AP highlights this facet of the current craze in its article, “Artificial Intelligence Technology Behind ChatGPT Was Built in Iowa—With a Lot of Water.” Iowa? Who knew? Turns out, there is good reason to base machine learning operations, especially the training, in such a chilly environment. Reporters Matt O’Brien and Hannah Fingerhut write:
“Building a large language model requires analyzing patterns across a huge trove of human-written text. All of that computing takes a lot of electricity and generates a lot of heat. To keep it cool on hot days, data centers need to pump in water — often to a cooling tower outside its warehouse-sized buildings. In its latest environmental report, Microsoft disclosed that its global water consumption spiked 34% from 2021 to 2022 (to nearly 1.7 billion gallons, or more than 2,500 Olympic-sized swimming pools), a sharp increase compared to previous years that outside researchers tie to its AI research.”
During the same period, Google’s water usage surge by 20% according to the company. Notably, Google was strategic about where it guzzled this precious resource: it kept usage steady in Oregon, where there was already criticism about its water usage. But its consumption doubled outside Las Vegas, famously one of the nation’s hottest and driest regions. Des Moines, Iowa, on the other hand is a much cooler and wetter locale. We learn:
“In some ways, West Des Moines is a relatively efficient place to train a powerful AI system, especially compared to Microsoft’s data centers in Arizona that consume far more water for the same computing demand. … For much of the year, Iowa’s weather is cool enough for Microsoft to use outside air to keep the supercomputer running properly and vent heat out of the building. Only when the temperature exceeds 29.3 degrees Celsius (about 85 degrees Fahrenheit) does it withdraw water, the company has said in a public disclosure.”
Though merely a trickle compared to what the same work would take in Arizona, that summer usage is still a lot of water. Microsoft’s Iowa data centers swilled about 11.5 million gallons in July of 2022, the month just before GPT-4 graduated training. Naturally, both Microsoft and Google insist they are researching ways to use less water. It be nice if environmental protection were more than an afterthought.
The write-up introduces us to Shaolei Ren, a researcher at the University of California, Riverside. His team is working to calculate the environmental impact of generative AI enthusiasm. Their paper is due later this year, but they estimate ChatGPT swigs more than 16 ounces of water for every five to 50 prompts, depending on the servers’ location and the season. Will big tech find a way to curb AI’s thirst before it drinks us dry?
Cynthia Murrell, October 6, 2023
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
