LATIDIA · Investigación

Control de sesgo e inyección para la agregación de múltiples agentes

arXiv: 2610.21290v1Tipo de anuncio: nuevo Resumen: El razonamiento multiagente con modelos de lenguaje grandes (LLM) plantea una pregunta central para la toma de decisiones colectivas: cuando los agentes no están de acuerdo, ¿en qué decisión se debe confiar? Re

La noticia

arXiv:2610.21290v1 Announce Type: new Abstract: Multi-agent reasoning with large language models (LLMs) raises a central question for collective decision-making: when agents disagree, whose decision should be trusted? Resolving such disagreements is difficult because agents' decisions are susceptible to non-semantic biases such as majority pressure and response-style preferences, which can obscure the quality of the underlying reasoning. Our analysis shows how bias can be both mitigated and exploited: bias control improves initial answer selection, while bias injection elicits decision instability that serves as a reliability signal. Building on these findings, we introduce a test-time aggregation framework that requires neither validation data nor prior knowledge of relative model strength. The framework implements