Knowing Whom to Trust is not Enough: The Trust--Agreement Gap in Repeated Multi-Party LLM Negotiation
4.40T1 sourcearXiv cs.MA
Source record
Published by arXiv cs.MA (T1 source). The original is at https://arxiv.org/abs/2610.31971.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryThe paper tests whether LLM agents can use trust estimates from repeated negotiations to resist a participant covertly blocking agreement. Agents often identify the adversary, but accurate trust estimates do not reliably improve outcomes. Agreement rises by up to 23 percentage points only when agents are explicitly guided to use trust in decisions.
Why it mattersProvides a specific design constraint for multi-agent workflows: trust detection alone is insufficient, and decision procedures must explicitly convert trust assessments into negotiation behavior.
Cited by
No citations on record.