(Credit: Ruhani Kaur/Bloomberg via Getty Images)
Unsurprisingly, OpenAI CEO Sam Altman thinks concerns about AI's impact on our water supply are overblown. But his logic is raising eyebrows: Training a large language model uses far fewer resources than raising a human, and complaints about water are "totally fake," he says.
"People talk about how much energy it takes to train an AI model, relative to how much it costs a human to do one inference query. But it also takes a lot of energy to train a human," he said during a Q&A session hosted by The Indian Express. "It takes like 20 years of life and all of the food you eat during that time before you get smart. And not only that, it took the very widespread evolution of the 100 billion people that have ever lived and learned not to get eaten by predators and learned how to figure out science and whatever, to produce you. And you took whatever you took."
A fairer question, he says, is to ask, "How much energy does [AI] take once its model is trained to answer that question versus a human? And probably, AI has already caught up on an energy efficiency basis, measured that way."
Although his comments elicited a few chuckles and head nods from those in attendance, the comparison hasn't gone down well. Sridhar Vembu,co-founder of Indian software company Zoho Corporation, was at the event and tweeted, "I do not want to see a world where we equate a piece of technology to a human being."
Indeed, Altman seems to suggest that the human experience can be distilled down to mere consumption and productivity ratios. While that might be a fair way to judge an AI, applying it to humans has dangerous implications for morality and for how technology shapes our lives.
And all this comes at a time of record investment in data centers worldwide to power AI. It's resulted in energy price increases, as well as concerns about water scarcity, noise, and pollution.
Although Altman conceded that OpenAI's previous use of evaporative cooling led to high water use, that is no longer the case, he says. Complaints that every ChatGPT query wastes gallons of water are "completely untrue, totally insane, [and have] no connection to reality."
Altman acknowledged that it's "fair" to examine AI energy consumption—"not per query, but in total" because people are using so much AI. But the answer is shifting to "nuclear or wind and solar very quickly," he says.