The water usage argument against AI has become a go-to. Here are two examples that clarify how it’s overblown:
(1) “It takes one gallon of water to make one single almond, and there are like 300 almonds in a bag of almonds, so if you’re worried about wasting water, you should stop eating almonds. I rather use AI than eat a bag of almonds” — @ morganbcohen on TikTok talking about her dad’s response
(2) “It takes 1,800 gallons of water to manufacture one pair of cotton jeans, which is the equivalent of [anywhere between 136,000 - 681,000] ChatGPT queries. Is AI bad for the environment? Yes. Are there other things you can do to conserve water instead of choosing not to use AI? Also yes” — @ paigeinthemachine on TikTok
The argument initially gained traction in April of 2023 bc of a study that came out of UC Riverside and UT Arlington titled “Making AI Less Thirsty” — but it especially caught wind bc of TikTok virality throughout 2025. The study estimated that roughly 500 milliliters of water (a small water bottle) is utilized through data center cooling systems when a user receives 10-50 medium-length responses from ChatGPT’s model at the time, GPT-3. Other researchers responded by saying the water footprint also varies widely based on location bc of differing infrastructures, energy sources, and system efficiencies
When it comes to their outlooks on competition, Eileen seems to be more business-minded while Alysa is more artist-minded. Neither is better or worse, just different archetypes
“I come alive at about 1AM”. @DemisHassabis shares how 10PM-4AM is when he does most of his thinking and creative work








