In a blog post, Altman stated: “We are now confident we know how to build AGI as we have traditionally understood it”
By Nileena Sunil
OpenAI CEO Sam Altman has refuted the claim that OpenAI’s o3-mini has reached the milestone of Artificial General Intelligence (AGI) in an X post. This has caused some confusion since it seemed to contradict his previous statements.
AGI refers to a type of artificial intelligence (AI) that has the ability to self-teach and solve problems it was never trained for, without manual intervention. In a blog post, Altman stated: “We are now confident we know how to build AGI as we have traditionally understood it.”
READ: OpenAI unveils Sora, an AI tool that transforms text Into video (December 11, 2024)
He had also suggested the same in a Bloomberg interview where he said that OpenAI’s o3, which is currently being safety-tested, had “passed the ARC-AGI challenge, the leading benchmark for AGI.” He said the company was setting its sights on superintelligence, “which is leaps and bounds beyond AGI, just as AGI is to AI.”
A post published by AI writer Gwern Branwen also claimed that OpenAI was on the verge of a major breakthrough, with its new reasoning models being able to produce data needed to train more advanced models.
However, Altman took to X on Jan. 20 to write, “twitter hype is out of control again. We are not gonna deploy AGI next month, nor have we built it. We have some very cool stuff for you but pls chill and cut your expectations 100x.”
The o3 model was introduced as part of the 12 Days of OpenAI in December. It was the successor to the o1 model, introduced last September.
READ: OpenAI concludes holiday campaign “Shipmas”: Here’s what you missed (December 23, 2024)
A report by François Chollet, an independent software engineer and AI researcher, states that while the ARC-AGI challenge is a critical benchmark for such breakthroughs, it is not an “acid test” for AGI, “passing ARC-AGI does not equate to achieving AGI, and, as a matter of fact, I don’t think o3 is AGI yet; o3 still fails on some very easy tasks, indicating fundamental differences with human intelligence,” he said. It remains yet to be seen when the development of AGI would be realized, if ever.