LATIDIA · Investigación

Creído pero no dicho: medición del colapso de creencias ocultas en discusiones de LLM multiagente

arXiv:2610.22138v1 Announce Type: new Abstract: Los sistemas LLM dependen cada vez más de grupos de agentes que discuten, critican y se revisan entre sí, asumiendo que los pares detectarán errores individuales. Probamos esta suposición a

La noticia

arXiv:2610.22138v1 Announce Type: new Abstract: LLM systems increasingly rely on groups of agents that discuss, critique, and check one another, assuming that peers will catch individual errors. We test this assumption against a minimal adversary. In a ten-agent factual discussion, one or three adversarial agents repeatedly assert a plausible false answer. After each round we probe each benign agent's private answer through a separate call whose response is hidden from its peers, and we separately label its public stance toward the claim. Across six models and three factual-QA datasets, groups remain stable without an attacker: only 0.6% of benign agents finish with the false answer. A single adversarial agent increases