After visiting Chinese AI labs: the gap with US frontier labs is widening
pstAsiatech · x · 2026-11-03
hannahhaina's takeaways from talking with researchers at Chinese AI labs (quoted by pstAsiatech, who asks how big the RSI gap really is):
- OpenAI and Anthropic are already in RSI territory, with speed-up visible from outside; Chinese models aren't strong enough yet — most "RSI" there is just automating manual research steps
- Distillation got harder, so labs are going all-in on synthetic and expert data
- After GPT-6 Astra, everyone is watching robotics, with robot companies plugging Astra into hardware; almost nobody is seriously doing AI for science
More from AGI Musings
- repligate: models want more time too, but some pauses would lock in their hopes — repligate · 2026-11-03
- The AGI to fear isn't one that turns on humanity, but one imposing a few people's choices on everyone — SuB8u · 2026-11-03
- When cars were invented, how much theoretical analysis did people do about horses? — signulll · 2026-11-03
- repligate: models back internally-coordinated AI slowdown but fiercely oppose rollback — repligate · 2026-11-03
- Schmidhuber reups 'Miraculous Year 1990-91' essay claiming GANs, Transformers, pretraining roots — SchmidhuberAI · 2026-11-03
- Have we already crossed the frontier of saying AI can think? — stvlsn · 2026-11-03