Researchers at the University of California, Riverside and the University of Texas, Arlington recently crunched the numbers to figure out how much water a data center operating ChatGPT would consume. They based this off of reported water consumption by Microsoft’s data centers while training OpenAI’s GPT-3, which clocked at about 185,000 gallons. For reference, that’s enough to make 320 Tesla vehicles.
How much water does ChatGPT ‘drink’ for every 20 questions it answers?
Answer: Roughly 500 ml.
It can be easy to forget about the environmental footprint of AI when you’re sitting at home using ChatGPT, where all you need to use it is the electricity that powers your computer and Internet equipment. Generative AI bots, however, are housed in data centers that have a much more significant carbon footprint.
Researchers at the University of California, Riverside and the University of Texas, Arlington recently crunched the numbers to figure out how much water a data center operating ChatGPT would consume. They based this off of reported water consumption by Microsoft’s data centers while training OpenAI’s GPT-3, which clocked at about 185,000 gallons. For reference, that’s enough to make 320 Tesla vehicles.
Using this metric, the researchers were able to calculate that ChatGPT “drinks” roughly 500 ml of fresh water for every 20 to 50 questions that it answers. That’s the same as a 16.9 oz water bottle. “While a 500 ml bottle of water might not seem too much, the total combined water footprint for inference is still extremely large” when you consider how many people are using ChatGPT, and how frequently.
Researchers at the University of California, Riverside and the University of Texas, Arlington recently crunched the numbers to figure out how much water a data center operating ChatGPT would consume. They based this off of reported water consumption by Microsoft’s data centers while training OpenAI’s GPT-3, which clocked at about 185,000 gallons. For reference, that’s enough to make 320 Tesla vehicles.
