One Human, N Agents: Audit-Budget Allocation for LLM Agent Fleets under Miscalibrated, Correlated Confidence
TL;DR
A single human must audit N LLM agents under a budget of B under a budget of $B per round, guided by self-reported confidence that may be adversarially miscalibrated and by correlated errors, to locate the miscalibration threshold.
Abstract
A single human must audit $N$ LLM agents under a budget of $B \ll N$ audits per round, guided by self-reported confidence that may be adversarially miscalibrated and by correlated errors. We model this as budgeted noisy inspection over a two-level Gaussian copula and locate the miscalibration threshold $\delta^*$ past which confidence-ranked auditing is \emph{worse} than random. Two a-priori expectations reverse: $\delta^*$ \emph{rises} as the budget shrinks, and cross-family correlation is not low---shared difficulty dominates lineage. Five open-weight LLMs show operationally useless (near-constant) confidence, point estimates at or beyond the flip though CIs straddle it; a proprietary model is informative and lands below it. We give a quantitative criterion for \emph{vacuous} oversight, and replaying policies on recorded traces confirms the ordering.