Andrew Lampinen on where RL creativity ends: is connecting two known math fields a new discovery?
AndrewLampinen · x · 2026-10-03
Continuing the debate on what RL teaches, Andrew Lampinen argues RL can certainly discover genuinely new things — but the boundary is not always clear. He poses a sharp example: if a model connects two areas of mathematics that have never been linked before, is that "new" or "old" if both areas already appeared in training? The question cuts to the core of whether RL merely amplifies training data or produces real novelty.
More from AGI Musings
- OpenAI and Notre Dame Law School to co-host AI-for-humanity dialogue on Dec 4 — wjscheirer · 2026-11-19
- NYT: Anthropic Shows Organizational Fractures Ahead of Historic IPO — kylekabasares · 2026-11-19
- The biggest advantage of a top lab isn't money — it's saving you years — madeofAjala · 2026-11-19
- Ex-Anthropic staffer casts superintelligence as Sauron's ring; Ladish pushes back — JeffLadish · 2026-11-19
- Proposal: A nonprofit to give visa safety to AI lab whistleblowers — AaronBergman18 · 2026-11-19
- "No innovation ever comes out of the finance department": ex-Covariant researcher on AI ROI — HanchungLee · 2026-11-19