wow the way ARC-AGI-3 scores work is way sillier than I thought | Tech Twitter

Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.

News//Trending//Time Travel//Threads//Articles//Profiles//Topics//About

Focus

Press Space to continue

Built by Daniel Hunter

wow the way ARC-AGI-3 scores work is way sillier than I thought | Tech Twitter

Press Space to continue

wow the way ARC-AGI-3 scores work is way sillier than I thought

SterlingCroxton@SterlingCroxton6d ago

It’s worth noting that something like GPT-6’s huge jump to ~100% on ARC-AGI-3 was always the most likely outcome, because ARC-AGI-3 uses a (totally non-standard) way of calculating scores. They never should have done it this way and now it’s blown up in their faces. https://t.co/1kTWtyvlsI

Previous tweet Next tweet More from @peterwildeford