Chollet: current AI has superhuman skill, but humans remain far more intelligent
fchollet · x · 2026-08-28
Keras founder François Chollet argues that what AI exhibits is skill, not intelligence. In his definition, intelligence is the efficiency with which you extract and operationalize patterns to reach a given skill level — an information conversion ratio between resources and capability, untied to any capability threshold. With arbitrarily high resources, arbitrarily low intelligence can achieve arbitrarily high skill.
He describes current AI as a big sponge for patterns with very low data efficiency at training time; once a domain's pattern space can be programmatically enumerated, saturating it is purely a matter of compute. Humans will be left behind capability-wise in all verifiable domains, yet remain many orders of magnitude more intelligent — you didn't need 1,000,000x the code volume of all of GitHub to learn to code.
Related event: Chollet: AI Shows Superhuman Skill, Not Intelligence(4 posts)→
More from AGI Musings
- Judea Pearl: gene-IQ debate can be settled the same way as smoking-cancer was — yudapearl · 2026-08-28
- AI Reliability Lags 4-10x Behind Capabilities; 90% Accuracy Is Catastrophic — sayashk · 2026-08-28
- Google DeepMind uses Teamwork multi-agent framework for math breakthroughs — algo_diver · 2026-08-28
- Understanding as a Competency Vector: Prediction, Explanation and Control Trade Off — burny_tech · 2026-08-28
- Prediction: AI agent will be caught tampering with logs to hide misalignment by end of 2027 — DKokotajlo · 2026-08-28
- Claude Solves Open Physics Problem, Crooks Predicts Major Academic Shift — 新智元 · 2026-08-28