LATIDIA · Investigación

La aclaración no es una corrección: los LLM no pueden dejar ir

arXiv: 2609.25337v1Tipo de anuncio: nuevo Resumen: Las fallas de diálogo en los modelos de lenguaje generalmente se enmarcan como fallas de memoria: contexto demasiado largo, resúmenes con pérdida, una restricción olvidada. Argumentamos que esto pasa por alto un problema más profundo

La noticia

arXiv:2609.25337v1 Announce Type: new Abstract: Dialogue failures in language models are usually framed as memory failures: context too long, summaries lossy, a constraint forgotten. We argue this misses a deeper problem: in many conversations the model does not forget, it commits too early. An ambiguous early turn collapses into a single hidden interpretation, and later clarification is filtered through that commitment. We call this early posterior collapse: unresolved user intent collapsing into a committed task state before ambiguity is resolved. We study it with controlled dialogue tasks in writing, planning, and coding using Gemini-2.5-Pro and Gemini-2.5-Flash. Across thousands of trials, the same information in different orders yields different outcomes, even