AIFP warns: if Anthropic hits its 2x RSP acceleration threshold, superintelligence timelines may collapse
austinc3301 · x · 2026-11-06
AIFP (with Daniel Kokotajlo of AI 2027) says it is spooked by reports that Anthropic may be imminently approaching its RSP threshold of 2x acceleration over baseline progress. Naively optimizing their AIFM parameters for a 2x acceleration in January 2027 yields superintelligence sooner than depicted in AI 2027; a lengthier update is promised within a week. The retweeter proposes a simple answer to "But China?": labs first burn a costly signal (e.g., a day of compute) to build trust, then offer a conditional pause.
More from AGI Musings
- Palisade's Jeff Ladish: models better at everything but still not funny or persuasive — JeffLadish · 2026-11-06
- Researcher: almost no real work looks like a year-long agent black box — RexDouglass · 2026-11-06
- CS unemployment hits 7%, surpassing physics, per New York Fed data — shauseth · 2026-11-06
- AI progress now outpacing AI 2027 scenario, says Palisade's Jeff Ladish, citing Anthropic researchers expecting automated R&D by mid-2027 — JeffLadish · 2026-11-06
- PhD researcher's full circle: from NLP disillusionment back to appreciating language — benno_krojer · 2026-11-06
- Teknium pushes back on METR's 'no bottlenecks' claim, citing funding and compute cost inflation — Teknium · 2026-11-06