Learning When to Automate: Queue Control in Human-AI Service Systems

Jul 16, 2026

Upload by Levi

A human–AI service system is studied where tasks arrive sequentially and are handled via a two-stage architecture: an automated chatbot followed, when needed, by a human agent. Over a finite horizon, each task belongs to one of K heterogeneous types. The controller selects chatbot resource levels, learning unknown, type-dependent success probabilities, while unresolved tasks enter type-dependent human queues with unknown service rates. A key tradeoff is balancing reduced human congestion against increased chatbot costs. The UCB-DPP policy learns these parameters and achieves low queue-aware regret.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages43

Reading Time1 h 48 min

Abstract

Introduction

Motivation

Contributions

arXiv :2607 .060 17v 1 [ cs .LG] 7 Jul 2026 Learning When to Automate: Queue Control in Human-AI Service Systems Giovanni Montanari 1,3 , Marco Scarsini4 , and Vianney Perchet 1,2,3 1FairPlay Joint Team, Inria, France 2 Criteo AI Lab, Paris, France 3 CREST, ENSAE, Institut Polytechnique de Paris 4Department of Economics and Financial Markets, Luiss University Abstract We study a human-AI service system in which tasks arrive sequentially and are processed through a two-stage architecture: an automated chatbot followed, when necessary, by a human agent. We consider T sequentially arriving tasks, each belonging to one of K heterogeneous types. For each task the decision maker chooses how many resources to allocate to the chatbot, whose type-dependent success probabilities are initially unknown. Tasks not resolved by the chatbot enter type-dependent human-service queues, where they are processed by a human agent with unknown service rates. This model captures a central tradeoff in hybrid service systems: relying more on automation reduces human congestion but increases chatbot costs, while insufficient automation may overload the human agent. We propose the UCB-DPP policy, which combines Upper Confidence Bounds with Drift-Plus-Penalty control to learn the unknown parameters of the system while making queue-aware decisions. We prove that UCB-DPP achieves regret s(KtanTes)sah gtuhaatranteesthe prompoesaend-rpate stolicyiuitpyerofforthmeshnman-ural bsealicineesqueues. Simulations on synthetic 1 Introduction The increasing deployment of large language models (LLMs) in service systems raises a fundamental question: how should automated agents and human operators be jointly coordinated? In many practical settings, an LLM-based assistant can resolve part of the incoming workload quickly, but its performance depends on the amount of computational or operational resources allocated to it and it may still fail on difficult requests. Human agents, on the other hand, are often more reliable but slower and capacity-constrained. A central challenge is therefore to exploit automation while controlling both operational costs and congestion in the human-service system. We study this challenge through an online learning and queueing-control model for human-AI service systems. Tasks arrive sequentially over a finite horizon T and belong to one of K heterogeneous types, representing different levels of difficulty or service requirements. Each task is first processed by a chatbot. The decision maker chooses a cost level for the chatbot, interpreted as the amount of resources devoted to automated resolution. A higher cost increases the probability that the chatbot resolves the task, while tasks not resolved automatically are routed to a type-dependent human queue. The human agent can serve only one queue at a time, so the controller must also decide how to allocate limited human service capacity across task types. The key feature of this setting is that automation and human scheduling control two different sides of the queueing system. The chatbot cost decisions shape the arrival process of the human queues: using more automation reduces future human workload but increases immediate chatbot costs. Scheduling decisions, instead, determine the departure process by allocating human service capacity across queues. Thus, automation and human service cannot be optimized independently: they must be coordinated over time in response to the current congestion state. Another key difficulty is that the platform typically does not know in advance how different task types will be handled by the chatbot and by the human agent. Empirical evidence suggests that the relative effectiveness of AI and human service depends on task complexity [33], while recent work on generative AI highlights that model performance may vary sharply across tasks that appear similar [7] . These observations motivate treating chatbot effectiveness and human serv

FAQs

How are tasks processed in the proposed human–AI service system?

Tasks arrive sequentially and are first processed by an automated chatbot. If the chatbot does not resolve the task, it is routed to a type-dependent human-service queue for processing by a human agent.

What decisions does the controller make in this model?

For each task, the controller chooses a chatbot cost level that determines how many resources to allocate to automated resolution, and it also allocates limited human service capacity across task types to manage congestion.

What is the main tradeoff addressed by the paper?

Relying more on automation can reduce human congestion but increases chatbot costs, while insufficient automation may overload the human agent. The paper coordinates these effects over time while learning unknown system parameters.

Jul 16, 2026

Upload by Levi