Speculation: if agents can solve Navier-Stokes, OpenAI may already use them to improve its own training stack
teortaxesTex · x · 2026-09-09
X user teortaxesTex amplifies a speculation: if frontier agents can crack decades-old math problems like Navier-Stokes within days, there is no reason the same approach wouldn't work on ML algorithms and the training stack itself. He suggests OpenAI could plausibly run 100K such agents to improve its own training algorithms — recursive self-improvement in practice. Speculative discussion, no evidence offered.
Related event: Massive agent clusters spark talk of AI-accelerated research(3 posts)→
More from AGI Musings
- Commenter doubles down: safety delays won't change for the next model either — Darpinian · 2026-09-09
- Alignment researcher reaffirms 2023 essay: AI alignment is fundamentally tractable — QuintinPope5 · 2026-09-09
- Chamath's 8090 says 'the singularity is here' — while hiring across all GTM roles — GarrisonLovely · 2026-09-09
- She tried AI voice mode in the shower and now draws a hard 'no AI' line — alliekmiller · 2026-09-09
- Physicist Daniel Whiteson: 'solving mathematics' is a nonsense phrase — DanielWhiteson · 2026-09-09
- Eric Jang predicts recursive self-improvement by end of 2026 — ericjang11 · 2026-09-09