AI Futures Project updates timeline: we're 5 months ahead of AI 2027, Superhuman AI Researcher seen in 2027
mhmazur · x · 2026-11-11
The AI Futures Project (Eli Lifland, Daniel Kokotajlo, et al.) published a Q3.5 timelines update arguing the intelligence explosion is on track for 2027 — now running 5 months ahead of the AI 2027 scenario.
Key points:
- METR estimate: Anthropic is making AI progress 2x its historical rate (1.3x–3x range), with "perhaps ≈80%" of current acceleration attributed to AI itself.
- Superhuman AI Researcher (SAR) — the point where a company would rather fire all human researchers — previously got 50% odds before 2029 from the most bullish forecaster; now the whole team forecasts SAR likely arrives in 2027 (absent deliberate slowdown).
- TED-AI (automating all white-collar work): 35–65% chance it arrives before 2028.
- The authors acknowledge possible modeling errors but believe the high-level conclusions hold; the METR report was the single biggest evidence source.
More from AGI Musings
- METR report: AI-driven R&D acceleration puts recursive self-improvement within reach by 2027 — birchlse · 2026-11-11
- "Retrofitting electricity to coal-powered factories": one line on AI's institutional mismatch — curious_vii · 2026-11-11
- 'AI can't be close to AGI, cancer isn't cured' — the dung beetle fallacy — danfaggella · 2026-11-11
- Yann LeCun: 'If you are interested in human-level AI, don't work on LLMs' — StasBekman · 2026-11-11
- US robotaxi deployments are closing the gap with China, new data shows — skorusARK · 2026-11-11
- Ben Todd: even Dario Amodei isn't truly AI pilled, citing Machines of Loving Grace — ben_j_todd · 2026-11-11