OpenAI CEO Sam Altman on Friday defended the resource demands of artificial intelligence, calling concerns about data centers’ water use “fake” and comparing the energy used by AI systems to that of humans.
Altman was speaking on the sidelines of the India AI Impact summit in an interview with The Indian Express when he was asked to address common criticisms of AI, such as its energy and water consumption.
The CEO responded that claims circulating online that ChatGPT uses gallons of water per query were “completely untrue, totally insane,” and have “no connection to reality.”
Data centers traditionally use large amounts of water to cool electrical components and prevent overheating. While data center cooling technologies have promised reduced consumption, some newer data centers no longer rely on water at all.
Still, even with improving efficiency, a report last month from water technology company Xylem and Global Water Intelligence projected that the water drawn for cooling would more than triple over the next 25 years as computing demand rises, putting pressure on water systems.
While dismissing fears about water use, Altman said energy consumption remains a fair AI concern. “Not per query, but in total – because the world is using so much AI … and we need to move towards nuclear or wind and solar very quickly,” he said.
Asked about previous comments from Microsoft founder Bill Gates — who has suggested that the efficiency of the human brain proves that AI can evolve to also become more energy efficient over time —Altman pushed back.
“One of the things that is always unfair in this comparison is people talk about how much energy it takes to train an AI model … But it also takes a lot of energy to train a human,” he said. “It takes like 20 years of life, and all the food you eat before that time, before you get smart.”
“The fair comparison is if you ask ChatGPT a question, how much energy does it take once a 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,” he added.
The process Altman is referencing is known as inference, which refers to the use of AI models that have already been trained to create new outputs. AI inference is typically much less power-intensive than the training itself.
Altman’s comments, particularly the AI-to-human comparison, have since sparked some debate online amid growing anxiety about AI’s ability to replace human work.
Sridhar Vembu, co-founder and chief scientist of Indian software company Zoho Corporation, who was present at the summit, criticized the human-AI equivalence. “I do not want to see a world where we equate a piece of technology to a human being,” the billionaire said in an X post.
The debate comes as governments and companies pour billions into new data centers to support the computing needs of AI systems.
According to a May report by the International Monetary Fund, electricity consumption by the world’s data centers in 2023 had already reached levels comparable to Germany or France, soon after the launch of OpenAI’s groundbreaking ChatGPT AI model.
In response, some governments have been working to speed up approval processes to bring new and cheap energy online, with some environmentalists warning such moves could clash with global net-zero goals.
Some local communities in countries like the U.S. have also pushed back on development projects over fears they will strain electricity grids and raise overall electricity costs.
Last week, the City Council in San Marcos, Texas, voted down a proposed $1.5 billion data center project after months of public opposition.
Amid such pushback, many tech leaders, including OpenAI’s Altman, have argued data centers will require more energy production from diverse sources, including renewable and nuclear energy.
[/gpt3]
