minus-squareGrimtuck@lemmy.worldtoTechnology@lemmy.world•Announcing ARC-AGI-3 - A benchmark that tests if AI can explore, learn, and adapt in unfamiliar situations. Humans score 100%. Frontier AI scores 0.26%.linkfedilinkEnglisharrow-up4·4 days agoLLM’s are just very well-read morons. linkfedilink
LLM’s are just very well-read morons.