Y Combinator Startup Podcast François Chollet: The ARC Prize & How We Get to AGI
257 snips
Jul 3, 2025 
François Chollet, creator of the Keras library and founder of the ARC Prize, dives deep into the challenges of achieving true general intelligence in AI. He critiques existing models for their limitations and discusses why simply scaling up isn’t enough. Chollet emphasizes the need for adaptability and fluid intelligence over static skills. He introduces the ARC Prize as a benchmark for progress and shares insights from his research at Ndea on advancing AI's capabilities through abstraction and compositionality, laying a path towards AI that can reason and innovate.
AI Snips
Chapters
Transcript
Episode notes
Limits of Scaling Up AI
- Scaling up model size and data alone do not yield general intelligence.
- Fluid intelligence requires AI to understand new problems on the fly, not just recall memorized skills.
Test-Time Adaptation Breakthrough
- Test-time adaptation allows models to modify behavior dynamically at inference.
- This marks a shift from static knowledge recall towards fluid intelligence in AI.
Defining Intelligence Correctly
- Intelligence is a process, not just the output skill.
- True intelligence means dealing with new situations, not just repeating known tasks.