Sam Altman: OpenAI’s New Model Passes AGI Threshold

OpenAI CEO Sam Altman says “AGI” (artificial general intelligence) has “become a very sloppy term.”

In a lengthy interview with Bloomberg News published Sunday (Jan. 5), he discussed his company’s past, as well as its plans for achieving that AI-related benchmark: artificial intelligence that can think and reason at or above human levels.

“If you look at our levels, our five levels, you can find people that would call each of those AGI, right?” Altman said. “And the hope of the levels is to have some more specific grounding on where we are and kind of like how progress is going, rather than is it AGI, or is it not AGI?”

Later in the conversation, Altman refers to something called the ARC-AGI challenge, which he refers to as “a North Star toward AGI.”

The new AI model OpenAI plans to introduce Friday (Jan. 10) passes this challenge, which requires a model to rely more on reason than training data to achieve AGI.

“They said if you can score 85% on this, we’re going to consider that a ‘pass,’ Altman said. “And our system — with no custom work, just out of the box — got an 87.5%. And we have very promising research and better models to come.”

Asked about the incoming Trump administration, Altman said that “U.S.-built infrastructure and lots of it” would be the most helpful thing the White House could do for AI this year.

“The thing I really deeply agree with the president on is, it is wild how difficult it has become to build things in the United States,” he said. “Power plants, data centers, any of that kind of stuff. I understand how bureaucratic cruft builds up, but it’s not helpful to the country in general.”

Altman also spoke about his brief — but high-profile — ouster as head of the company in November of 2023, acknowledging that the experience had traumatized him, leaving him depressed even after he returned.

“And it felt so unfair. It was just a crazy thing to have to go through and then have no time to recover, because the house was on fire,” he said.

In May of last year, former OpenAI board member Helen Toner said that Altman was fired because the board couldn’t believe what the CEO was telling it. 

She said the board learned about ChatGPT’s launch after the fact, that Altman didn’t disclose his involvement with the company’s startup fund, and that Altman repeatedly gave the board inaccurate information about the company’s formal safety processes.