Dwarkesh now expects superintelligence by 2030, citing AI speeding up AI R&D
AndyMasley · x · 2026-11-14
Dwarkesh Patel's new essay "The avalanche that hit math is coming for AI development" explains why he dramatically pulled forward his superintelligence timeline.
Core argument
- After three months of debates and watching AI progress on verifiable domains, he now believes AIs as good as the best humans at AI research are only a couple of years — possibly months — away
- Once AI research is automated, a single year will compress enormous progress, so by 2030 (plausibly as early as next year) AIs will match the best human experts in every field
- He admits it's a wild claim he didn't believe months ago
Key evidence
- METR, embedded at Anthropic for three months, concluded Anthropic's rate of progress on well-scoped tasks is 2x a year ago, with most acceleration "likely attributable to extensive use of AI to do R&D"
- OpenAI has published similar observations about AI accelerating its own research
The essay uses the avalanche that hit mathematics as an analogy for what's coming for AI development itself; video/podcast versions available.
Related event: Dwarkesh drastically shortens superintelligence timeline to before 2030(2 posts)→
More from AGI Musings
- New book repeats the myth that transformers lack recurrence, critic says it ignores Anthropic's interpretability findings — repligate · 2026-11-14
- The human/AI interregnum: nobody can maintain our systems, and AI isn't ready yet — ivan_bezdomny · 2026-11-14
- Rishabh Agarwal on physical AI: RL is 'the dumbest paradigm,' yet it works — ziv_ravid · 2026-11-14
- Recursive self-improvement has no single starting moment — Cosmos Institute essay dissects the intelligence explosion — sebkrier · 2026-11-14
- Gemini helps WSJ reporter crack her 20-year medical mystery — vivnat · 2026-11-14
- 77% of clinical drugs target the same 20% of popular targets—AI could fix this — jacobkimmel · 2026-11-14