Hello and welcome. Today we’re examining claims by OpenAI CEO Sam Altman about ChatGPT’s energy and water footprint. Altman wrote that a single prompt uses “0.34 Wh… equivalent to what a high-efficiency lightbulb would use in a couple of minutes” and needs “0.000085 gallons of water, roughly one-fifteenth of a teaspoon.” But Gizmodo notes he offered no evidence. Independent studies suggest GPT-3 could consume about half a liter of water for every 10 to 50 queries. Multiply that by hundreds of millions of users and you reach tens of millions of gallons per year.

Altman also insisted AI costs will “eventually converge to near the cost of electricity,” and predicted “entire classes of jobs” will disappear, offset by a wealthier world. Critics warn data centers already guzzle power and water, often cooled by fossil fuels, and that universal basic income hype won’t fix the environmental strain. As global temperatures rise, data center cooling demands could outpace any benefits AI promises.
Link to Article