Sam Altman says concerns of ChatGPT's energy use are overblown: 'It also takes a lot of energy to train a human'

In a recent discussion at a prominent AI summit, Sam Altman addressed the ongoing concerns regarding the energy consumption of ChatGPT. He argued that comparing the energy required to train an AI model to the energy cost of a human performing a single inquiry is misleading. Altman stated, "It also takes a lot of energy to train a human," emphasizing that the investment in human education and development over many years is substantial. Altman humorously noted the extensive time and resources needed to nurture a person to the point of independent thought, stating, "It takes, like, 20 years of life, and all of the food you eat during that time before you get smart." He highlighted that this process has been ongoing for thousands of years, involving the collective evolution of humanity, which has led to significant advancements in knowledge and survival skills. He also addressed some bizarre claims circulating online, particularly the assertion that AI companies like OpenAI are consuming large amounts of water to operate ChatGPT. Altman dismissed these allegations as "totally insane," clarifying that the previous practice of using evaporative cooling in data centers is no longer in use. He refuted claims that suggested each query necessitates an excessive amount of water, highlighting that such assertions are unfounded. In earlier comments, Altman shared that an average ChatGPT query consumes approximately 0.34 watt-hours of energy, equivalent to the power needed for a lightbulb for a few minutes. He acknowledged the growing scrutiny of the AI sector's energy consumption and mentioned the industry's shift towards renewable energy sources like solar, wind, and nuclear to mitigate environmental impact. While Altman supports alternative energy, he expressed skepticism about the feasibility of space-based data centers as a solution for energy efficiency within the next decade. As a significant advocate for nuclear energy, he has invested in various projects aimed at advancing sustainable power, including a fusion energy initiative in Washington state. The conversation around data center energy usage continues to gain traction, with estimates indicating they could represent 14% of the total power demand in the US by 2050.