Rebuttal to AI speedup claim: timeline mismatch and no diminishing returns
austinc3301 · x · 2026-11-09
Countering the claim that Mythos-class models accelerated Anthropic's research, the author argues: the speedup appeared to start in October 2025 while Mythos shipped in February 2026, so the timeline doesn't fit; and if it were a one-time RL free lunch, you'd expect diminishing returns or some slowdown, which the data doesn't show.
More from AGI Musings
- SWE workday 2022 vs now: 30% reviewing AI code, 40% writing specs — itsOmSarraf_ · 2026-11-09
- Software Is Now ~100% AI-Written — But Senior Engineers Reap the Benefits, What About Juniors? — bendee983 · 2026-11-09
- Smriti opens 1.1M+ Indian manuscripts and 100K inscriptions for free search — Paimaamu · 2026-11-09
- Veteran architect: let AI design novel chip architectures, formally verify the output — satnam6502 · 2026-11-09
- Berkeley Dinner on the "AI Crisis" Features Talk by Prof. David Krueger — KatjaGrace · 2026-11-09
- Rethinking the Bitter Lesson: 'don't work on things that don't scale,' not 'scale is enough' — gowthami_s · 2026-11-09