Vals RSI Index extrapolates Google models hitting 130% autonomous AI R&D by Oct 2027
HaydnBelfield · x · 2026-10-16
AI safety researcher Haydn Belfield flagged the Vals RSI Index, which scores whether models can do the research that builds the next model, benchmarked against published, human, and frontier-model references.
- Current models from Anthropic (Opus 5.5), OpenAI (GPT-6.1 Sol), Moonshot, Alibaba, Meta and others cluster at 20-40%;
- Google's extrapolation is the most aggressive: 60% by Jan 2027, 100% by Jun 2027, 130% by Oct 2027;
- OpenAI's extrapolation reaches only 60% by Aug 2028, Anthropic's 60% by Jul 2027;
- Vals' Rayan Krishnan notes it's no secret: their external RSI index predicted the same timeline.
If the extrapolation holds, models could surpass human frontier researchers at autonomous AI R&D within 2027 — a notable quantified take on recursive self-improvement timelines.
More from AGI Musings
- Miles Cranmer's talk: how to avoid 'AI Brain Rot' and intelligence atrophy — Scobleizer · 2026-10-16
- AI's fundamental design flaws are fully hidden by its products' superficial design flaws — lescarr · 2026-10-16
- IMF Official: Markets' Surprising Resilience Largely Rests on AI Optimism — kernelangus420 · 2026-10-16
- Can AI trained only on human content infer the shape of the human brain? — rickasaurus · 2026-10-16
- Gowers' tl;dr: no p(doom), but can't confidently say it's close to zero — michael_nielsen · 2026-10-16
- Fields Medalist Gowers pens long essay: can't confidently say AI extinction risk is near zero — michael_nielsen · 2026-10-16