minus-squareHaunchesTV@feddit.uktoTechnology@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-up6·3 days ago Grok Reasoning: 0% Hilarious linkfedilink
Hilarious