Chollet's shift in tone is interesting. As recently as 2022, he expressed strong skepticism about LLMs solving ARC-AGI within a decade. Now, he seems to suggest that significant progress on the challenge is possible within the next 1-2 years.
Of course, he strives to maintain a consistent public persona, but this shift strongly implies that o3's performance has significantly altered his views.
This field, much like economics, is full of experts who can convincingly explain why there were wrong :)
PS: ARC-AGI is a very well-though-out/interesting/useful test.
The result announcement blog post sounded too hypey and gave the impression that he changed his tone but for the last one year he has been consistently saying that LLM+"discreet program search" is the way to go. All the top scoring submissions before o3 had followed the same bruteforce strategy. Even o3 is more or less doing the same bruteforce under the hood, maybe not discreet.
Of course, he strives to maintain a consistent public persona, but this shift strongly implies that o3's performance has significantly altered his views.
This field, much like economics, is full of experts who can convincingly explain why there were wrong :)
PS: ARC-AGI is a very well-though-out/interesting/useful test.