François Chollet says OpenAI delays AGI progress by five to 10 years

Critique of current AI models

Chollet stated that large language models based on neural networks represent a dead end for achieving human-like intelligence. He argued that while these models can be improved for business utility, they do not equate to true AGI. The industry focus on these architectures has diverted resources from other potential pathways.

Funding priorities and consequences

By pouring funding into large language models at the expense of other avenues, the road to genuine artificial general intelligence faces significant obstacles. Chollet explained that this financial direction prevents exploration of alternative architectures needed for reasoning like humans.