Dwarkesh podcast with Noam Brown: 130B tokens equal 4,000 years of one human thinking
daniel_mac8 · x · 2026-09-19
Dwarkesh Patel releases a new podcast episode with OpenAI researcher Noam Brown on multi-agent systems and what the current explosion of math progress signals for automating AI research. A striking framing from Dwarkesh: 130 billion tokens of concentrated cognitive effort would equal a single human thinking full-time for 4,000 years.
Topics covered:
- Multi-agent systems and Navier-Stokes
- How AI firms will work
- What math progress tells us about recursive self-improvement
- Chain-of-thought degradation
- The internal/external model gap
- How we'd know if models are actually aligned before kicking off RSI
More from AGI Musings
- Investor: in the LLM era 'did it work' replaces diplomas and test scores — StewartalsopIII · 2026-11-04
- Should AGI labs have a reverse Conway's Law — design the org from the artifact? — akbirthko · 2026-11-04
- Verification Is Not Knowledge: Science needs more than AI answers — ersinesen · 2026-11-04
- Andreessen: Agents Are Entering Enterprises as Chaos, Just Like PCs and Spreadsheets Did — Dan_Jeffries1 · 2026-11-04
- 1000+ mathematicians sign pledge to shun AI companies and models — Shesawthat · 2026-11-04
- Von Neumann meme: "You don't have to be responsible for the world that you're in" — akbirthko · 2026-11-04