Flutterby™! : Dwarkesh Patel interviewing Richard Stutton

Watching Dwarkesh Patel interviewing Richard Stutton — Father of Reinforcement Learning things LLMs are a dead end (Transcript), and within the first 10 minutes we see Patel falling into the "reasoning" trap, and the "Math Olympiad" trap.

Oh, cool, and now we're getting into modes of learning, and they're talking about how formal schooling is different from how humans learn(!). "I don't think learning is about training, I think learning about learning, it's an active process."

"Why are you trying to distinguish humans? Humans are animals, what we have in common is much more interesting."

Lots of good stuff in here. I love his pushback against human exceptionalism, though I'm particularly primed to that because of my childhood experience with Anthroposophy.

Richard Sutton's home page