OpenAI CEO Believes AI Is More Energy Efficient Than Humans
At the AI summit in India, OpenAI CEO Sam Altman presented an unconventional assessment of the energy efficiency of artificial intelligence. In his opinion, with proper accounting, AI and human metrics may turn out to be comparable.
Altman's key thesis is the need to consider the full "energy footprint" of human development. He suggests taking into account the energy expended on educating a person during the first 20 years of life. The head of OpenAI emphasizes that the path to intelligence requires significant energy costs even before a person reaches maturity. Before acquiring knowledge and skills, an individual goes through a long period of growth and learning—which implies the consumption of food and the use of various resources.
Moreover, behind the development of an individual stands the centuries-long evolution of humanity: according to Altman, for us to appear and reach the current level of development, the evolution of about 100 billion people was required. They learned to survive, avoid dangers, master sciences, and pass knowledge to future generations.
With this approach in mind, Altman believes that artificial intelligence may have already reached human-level energy efficiency—if the process is evaluated comprehensively, rather than only in terms of the immediate energy consumption of servers.
Additionally, the OpenAI CEO refuted widespread estimates of the resource intensity of AI systems when processing individual queries. According to him, claims that one AI query requires tens of liters of water or one and a half iPhone battery charges are inaccurate. Altman called such calculations incorrect and urged a more measured approach when assessing the environmental and energy footprint of artificial intelligence technologies.