ARC Prize co-founder François Chollet: LLM focus set back AGI progress by 5–10 years

ARC Prize co-founder François Chollet said on the Dwarkesh Podcast that large language models are a dead end for AGI and that pouring funding into them has set progress back by up to a decade. He noted no LLM has come close to passing the ARC-AGI benchmark.

• Chollet calls LLMs a dead end for reaching AGI
• LLM funding bias set AGI progress back 5–10 years
• No model has come close to passing ARC-AGI
• ARC Prize was founded to measure true AGI progress