New Agent-Based Model Shows AI Scientists Produce Papers in Bulk but Never Replicate Each Other's Work
_akpiper · x · 2026-11-06
A new arXiv paper, "Scientific Liquidity Agents" (Nov 2, 2026), introduces an agent-based model of knowledge arbitrage in scientific communities, probing a key question: AI scientists can produce papers, but do they actually advance science together?
- The model represents domain specialists and epistemic arbitrageurs in a shared environment built on a belief-desire-intention architecture, a prerequisite-based problem graph, a dynamic interaction network, and a persistent publication registry.
- Two exploratory simulations (10 and 18 agents) produced 68 and 274 registered papers.
- Key finding: paper output is abundant, but no recorded replications; topical similarity alone splits the community into seven components, while arbitrageur-mediated contact pairs connect it into one.
- Measured community structure depends on what counts as a link, with knowledge-transfer frictions modeled explicitly.
A formal model for the worry that masses of AI-generated papers won't mean faster scientific progress.
More from AGI Musings
- eigenrobot on the contemporary mythos: denying beauty's reality as a developmental failure — eigenrobot · 2026-11-06
- Researchers Slam AI Peer Review: LLMs Find 100+ Limitations but Miss Critical Issues — IanArawjo · 2026-11-06
- If AGI is so close, why is everyone leaving? Inside the bets of 45 AI neolabs — The Decoder · 2026-11-06
- Professor proposes sit-in exams and rec letters to certify AI-free problem-solving skill — PTenigma · 2026-11-06
- "College kids are screwed": founder reverses stance as AI widens skill gap — RachelVT42 · 2026-11-06
- ChatGPT's 1.2B weekly users are already doing agentic tasks in main chat — omooretweets · 2026-11-06