ARC Prize co-founder François Chollet claims OpenAI has set back AGI progress by a decade

François Chollet, co-founder of the ARC Prize, has criticized the current industry focus on large language models (LLMs), arguing that they are a dead end for achieving artificial general intelligence (AGI). In a recent appearance on the Dwarkesh Podcast, Chollet suggested that by prioritizing LLM development at the expense of alternative architectures, companies like OpenAI have effectively delayed progress toward true AGI by five to ten years. While acknowledging that LLMs can be improved for business utility, he maintains that they lack the reasoning capabilities required for human-like intelligence. Chollet’s ARC-AGI benchmark is designed to measure genuine progress, and he notes that current models still fall significantly short of these goals. He further suggests that the industry's focus on LLM risks and marketing hype often distracts from the fundamental scientific challenges required to reach the next stage of machine intelligence.