Comment on Announcing ARC-AGI-3 - An benchmark that tests if AI can explore, learn, and adapt in unfamiliar situations. Humans score 100%. Frontier AI scores 0.26%.
My programming tells me I’m not a bot.