DeepSeek reportedly chases true continuous learning while OpenAI bets on agent swarms
teortaxesTex · x · 2026-09-14
teortaxesTex argues that DeepSeek and OpenAI actually diverge on "continuous learning."
- DeepSeek reportedly wants genuine continuous learning in the model itself, with a roadmap to recursive self-improvement (RSI) before embodiment — and LLMs may already suffice for RSI.
- OpenAI, by contrast, seems to treat "thousands of agents + RLM" as continuous learning in the aggregate, which the author thinks is probably a correct reading.
Unofficial commentary on the two labs' differing AGI paths.
More from AGI Musings
- The Robotics Prisoner's Dilemma: Use Frontier Models and Your Moat Gets Absorbed — ChongZzZhang · 2026-09-14
- 1.2M STEM dissertations show government is the top funder of frontier PhD research — joshgans · 2026-09-14
- a16z's Josh Elman: AI collapsed build costs, not the cost of knowing what to build — a16z · 2026-09-14
- Big lab executives now publicly own up to AI loss-of-control scenarios — emmanuelvivier · 2026-09-14
- OpenAI Claims Navier-Stokes Millennium Prize Solution, Prompting a Mathematician's Rethink on AI Math — fortnow · 2026-09-14
- AI's 'rationalist' scene is really a Berkeley culture, argues OpenAI-adjacent researcher — jachiam0 · 2026-09-14