OpenAI’s Sam Altman said humanity is close to reaching superintelligence and he believes ChatGPT is “already more powerful than any human who has ever lived” in certain respects.
While artificial intelligence researchers and developers have different definitions of advanced AI, superintelligence refers to theoretical AI systems that far surpass humans’ abilities across all cognitive domains.
OpenAI is pursuing artificial general intelligence that performs at least as well as humans alongside a slew of American and international competitors.
Mr. Altman’s views of the powerful AI models’ promise and his hopes for its future carry major national security consequences as his company is making products for U.S. intelligence and military customers.
He wrote Tuesday on his blog that humanity is close to building superintelligence and AI makers have recently built systems that are smarter than people.
“In some big sense, ChatGPT is already more powerful than any human who has ever lived,” Mr. Altman wrote. “Hundreds of millions of people rely on it every day and for increasingly important tasks; a small new capability can create a hugely positive impact; a small misalignment multiplied by hundreds of millions of people can cause a great deal of negative impact.”
Mr. Altman predicted agents that “can do real cognitive work” will arrive in 2025, systems that “can figure out novel insights” will emerge in 2026, and by 2027 society “may see the arrival of robots” doing physical tasks.
The OpenAI CEO has a large financial incentive in hyping the expectations for his company’s products under development.
In December 2024, word spread that if OpenAI achieves AGI or a certain profit threshold, the company’s revenue-sharing agreement with Microsoft would reportedly change to let OpenAI keep more of its own profit. Microsoft is a large benefactor of OpenAI, having invested billions of dollars.
OpenAI and Microsoft have their own internal definition of AGI that put a price tag on the technology at $100 billion in profit, according to a 2024 report from tech publication the Information. The companies’ terms were not made public but said to be part of a 2023 agreement.
OpenAI is not yet near the profit threshold. The company has an annualized revenue of $10 billion, according to a CNBC report on Monday.
In 2025, Microsoft also said it made a new agreement with OpenAI that extended its partnership through the duration of 2030. Microsoft said the arrangement maintained its access to OpenAI’s intellectual property and kept its revenue-sharing agreements in place.
Mr. Altman’s decision to publish about the future of superintelligence on his personal blog and social media accounts rather than with the institutional backing of OpenAI’s website is notable.
He has previously reserved major predictions for his own blog that he later appeared to tamp down.
In January, Mr. Altman took to his blog to declare that OpenAI was confident it knew how to build AGI and that AI agents would join the workforce in 2025.
Later that month, Mr. Altman said on X that online hype was getting out of control and OpenAI had not built AGI yet nor was preparing to deploy it.
The timing of Mr. Altman’s latest personal reflections on powerful AI comes as some of his competitors make moves to compete with OpenAI.
Meta is reportedly planning to join forces with Scale AI in a multibillion-dollar project with designs on creating a lab to reach superintelligence. News of the closely held plans spread on Tuesday, hours before Mr. Altman sought to draw attention back to his own agenda.
• Ryan Lovelace can be reached at rlovelace@washingtontimes.com.