The Dwarkesh thesis, distilled: generalization itself is a verifiable domain you can hill-climb
joshua_saxe · x · 2026-11-14
Joshua Saxe offers a pithy summary of the Dwarkesh thesis: generalization — in parameter space during training and token/latent space during in-context learning — is itself a verifiable domain that can be automatically hill-climbed, and gains there accrue to all other domains, verifiable or not. He says this framing makes him more bullish on AI progress over the next few years.
More from AGI Musings
- Going Infinite Reflection: A Silicon Valley Generation Whose Ambition Atrophied — Graham_dePenros · 2026-11-14
- A rare German film asks what home care looks like once a humanoid robot moves in — CyberRobooo · 2026-11-14
- Interconnects: Jensen's resignation is a major setback for US open-model ecosystem — Interconnects (Nathan Lambert) · 2026-11-14
- AGI endgame: either a balanced multipolar world or a centralized panopticon — CedricMakes · 2026-11-14
- "60% of people around you have some form of AI psychosis," says Alsop III — StewartalsopIII · 2026-11-14
- AI won't make everyone equal—it will split the smart from the rest worse than money — prasenx · 2026-11-14