OpenAI Chief Executive Officer Sam Altman said this week that claims about AI’s water consumption are not accurate and argued that the more relevant concern is the total energy used by AI systems, during remarks at an event hosted by The Indian Express in India.
Comments On Water Use And Cooling
Altman, who was in the country for a major AI summit, said worries about AI’s water usage are “totally fake,” while noting that water had been a real issue in the past when data centers relied on evaporative cooling. He said that method is no longer used.
“Now that we don’t do that, you see these things on the internet where, ‘Don’t use ChatGPT, it’s 17 gallons of water for each query’ or whatever,” Altman said. “This is completely untrue, totally insane, no connection to reality.”
Energy Consumption And Power Sources
Altman said it is “fair” to be concerned about energy consumption, not on a per-query basis, but in total, because of how widely AI is now used. He said this level of use means the world needs to move more quickly toward nuclear power or wind and solar energy.
There is no legal requirement for technology companies to disclose how much energy or water they use. Researchers have been studying the issue independently, and data centers have been linked to rising electricity prices.
Comparisons With Other Measures
The interviewer cited a previous conversation with Bill Gates and asked whether it is accurate to say a single ChatGPT query uses the equivalent of 1.5 iPhone battery charges. Altman replied, “There’s no way it’s anything close to that much.”
He also said many discussions about ChatGPT’s energy use are “unfair,” especially when they compare the energy required to train an AI model with the cost of a single human inference. He said that comparison ignores the resources needed to train a person.
How Altman Frames The Comparison
“It also takes a lot of energy to train a human,” Altman said. “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.”
Altman said the more relevant comparison is how much energy it takes for a trained AI system to answer a question versus how much energy it takes for a human to do the same. He said that, measured that way, AI has probably already reached a similar level of energy efficiency.
Featured image credits: Heute.at
For more stories like it, click the +Follow button at the top of this page to follow us.