News for hackers. Stuff that matters.

A comment on OpenAI O3 breakthrough high score on ARC-AGI-PUB. Parent comment

Re:OpenAI O3 breakthrough high score on ARC-AGI-PUB by Eridrus on Friday December 20, 2024 @01:43PM (#42473635)
ML is quite good at understanding and forecasting patterns when you train on the data you want to forecast. LLMs manage to do so much because we just decided to train on everything on the internet and hope that it included everything we ever wanted to know.

This tries to create patterns that are intentionally not in the data and see if a system can generalize to them, which o3 super impressively does!

[ Parent | on HN ]
Re:OpenAI O3 breakthrough high score on ARC-AGI-PUB by yunwal on Friday December 20, 2024 @02:14PM (#42473952)
ARC is in the dataset though? I mean I'm aware that there are new puzzles every day, but there's still a very specific format and set of skills required to solve it. I'd bet a decent amount of money that humans get better at ARC with practice, so it seems strange to suggest that AI wouldn't.
[ Parent | on HN ]