Podcast: Epoch AI researcher on RSI, robotics, China model gap, and open vs closed safety
natolambert · x · 2026-09-22
Natolambert releases a new podcast with Epoch AI researcher datagenproc exploring the open questions shaping frontier AI. Timestamped chapters cover: predictions for recursive self-improvement (00:00), robotics' role in AI acceleration (18:15), how far behind Chinese models are (24:20), whether distillation explains the gap (27:39), what Chinese job postings reveal about their labs (40:58), whether open or closed models are safer (48:13), how Epoch AI operates (58:10), and what a frontier post-training recipe looks like (1:00:55).
Related event: Interconnects Podcast Debates Frontier AI's Open Questions with Epoch AI(2 posts)→
More from AGI Musings
- The Neo-Prime: Pairing Frontier AI Management with Performer Networks for Moonshots — davidfromkansas · 2026-09-22
- Capital share rising from 0.33 to 0.6 could multiply wages nearly fivefold — sudoraohacker · 2026-09-22
- Microsoft uses Windows telemetry to measure global AI adoption across ChatGPT, Claude, Gemini — robseamans · 2026-09-22
- Ben Reinhardt on healthy research orgs: bubbling cauldrons and strike teams — Ben_Reinhardt · 2026-09-22
- Did Terence Tao hint at Riemann progress? Polymath-style AI math collaboration speculation swirls — TacoCohen · 2026-09-22
- Aran Nayebi: our brains are statistical engines too, so don't dismiss LLM reasoning — aran_nayebi · 2026-09-22