Id: arxiv/2601.22295 Url: https://arxiv.org/abs/2601.22295 Tags: math.OC Score: 0 Title: Operating Imperfect AI: Reliability Drift and Human Congestion Domain: math.OC Source: arxiv Authors: Ziyao Wang, Svetlozar T Rachev Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-01-29 Page: https://delphi.science/r/arxiv/2601.22295 The deployment of machine learning in high-stakes services relies on ``human-in-the-loop'' architectures to mitigate algorithmic uncertainty. However, existing static policies fail to address a fundamental tension: algorithms suffer from stochastic ``reliability drift,'' while human override capacity is scarce and congestible. We formulate the management of such systems as a dynamic queueing control problem. The system state is defined by the tuple (queue backlog, reliability regime), and the control variable is a state-dependent risk threshold. We prove that the optimal escalation policy is…