AI 2027 authors 'spooked' as Anthropic's S1 hints RSI threshold may be imminent
Puzzleheaded-King584 · reddit · 2026-11-05
Anthropic updated its S1 filing with new language about automated AI R&D and human oversight, noting internal estimates suggest it "may imminently" meet its Responsible Scaling Policy threshold of "dramatically accelerated progress attributable to AI" — i.e., 2x acceleration over baseline.
Daniel Kokotajlo of AIFP, lead author of AI 2027, responded that his team is "spooked": if Anthropic is really about to hit the 2x threshold, timelines to full recursive self-improvement and superintelligence may be very short. Naively calibrating their AIFM model parameters to reach 2x acceleration in January 2027 yields superintelligence sooner than depicted in AI 2027. The team is thinking harder and promises a lengthier update next week.
More from AGI Musings
- Steam Next Fest hits 8,000+ demos as AI flood stretches review times to 12 days — draginol · 2026-11-05
- Dune director Denis Villeneuve: AI only reproduces the past, human creativity steps into the unknown — adariostrange · 2026-11-05
- US AI scale faces two physical gates: delivered power and manufacturing — ingliguori · 2026-11-05
- "This time will be remembered as the start of the New Luddite movement" — Darpinian · 2026-11-05
- US policymaker's call to pause AI sparks debate on technical literacy in politics — nickbaumann_ · 2026-11-05
- How "reverse game theory" mechanism design could keep AI agents in check — HazanPrinceton · 2026-11-05