The LLM line of research will reach a capability plateau.
François Chollet Creator of the Keras deep-learning library and the ARC-AGI benchmark @fchollet on Xx.com · 7 Aug 2026
Recalls holding
François Chollet
Creator of the Keras deep-learning library and the ARC-AGI benchmark.
François Chollet did not write this page.
We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.
3 of François Chollet's claims fit 3 mental models. Most often: Collective action problems, Leaky abstractions, S-curves and diminishing returns.
Models they name
S-curves and diminishing returns
Most growth slows as it matures; the question is whether the curve is flattening or something new is restocking it.
Models we see in what they say
Collective action problems
What everyone would gain from, nobody provides alone, so shared goods need someone or something to hold people together.
If AI has a high chance of causing human extinction within years or decades, the only rational stance on safety monitoring and research pacing is stringent top-down government involvement and universally ratified international treaties.
François Chollet Creator of the Keras deep-learning library and the ARC-AGI benchmark If there really is a high chance of AI leading to the extinction of humanity within years/decades, then the only rational stance towards safety monitoring and research pacing should be stringent, top-down government involvement and universally ratified international treaties. @fchollet on Xx.com · 13 Sept 2026
Their wordsIf there really is a high chance of AI leading to the extinction of humanity within years/decades, then the only rational stance towards safety monitoring and research pacing should be stringent, top-down government involvement and universally ratified international treaties.
Leaky abstractions
A shortcut that hides the details works until it doesn't; when it breaks, someone has to know what is underneath.
The difficulty of software engineering is probably essentially constant across abstraction levels, because human cognition adapts to new tools.
François Chollet Creator of the Keras deep-learning library and the ARC-AGI benchmark I think the "difficulty" of software engineering is essentially constant no matter what abstraction level you move to, because human cognition adapts to new tools until it can fully utilize itself. @fchollet on Xx.com · 25 Sept 2026
Their wordsI think the "difficulty" of software engineering is essentially constant no matter what abstraction level you move to, because human cognition adapts to new tools until it can fully utilize itself.