Shaolei Ren revises widely cited AI water estimate from a full bottle to 15 milliliters per prompt

I've had a very nice back-and-forth email exchange with Shaolei Ren, who provided the original "AI uses a bottle of water per prompt" estimate with the Washington Post. I wanna be clear that he's more concerned than I am about AI's broader water use, so I don't want this to be read as him agreeing with me on my larger points, but he specifically agrees that his estimate for the water bottle per prompt is now outdated. Specifically: "The 2024 estimate was time-specific, assumption-based, and should not be used to describe general AI/ChatGPT or today’s optimized systems." His specific estimate for an average chatbot's water cost now is ~15 mL if you include the offsite costs. 33x lower than the bottle of water. So I do want to plant a flag and say "The person who provided the original sole estimate that chatbots use a bottle of water per prompt agrees that now that we have better information and more optimized systems, this is no longer the case, and models use ~15 mL each if you include the offsite costs, roughly 5 mL in the data center itself." It's silly that in 2026 educated people still believe that AI uses a bottle of water per normal prompt. No one investigating this believes it anymore.