Fatigue-Aware Learning to Defer via Constrained Optimisation

TL;DR

FALCON introduces a fatigue-aware approach to learning when AI should defer to humans, explicitly modeling workload effects with constrained optimization to improve human-AI collaboration.

Contribution

It models fatigue-induced human performance degradation within L2D using a CMDP framework and introduces a benchmark for varying fatigue dynamics.

Findings

FALCON outperforms existing L2D methods across multiple datasets.

It generalizes zero-shot to unseen fatigue patterns.

Adaptive collaboration surpasses AI-only or human-only decisions at intermediate coverage.

Abstract

Learning to defer (L2D) enables human-AI cooperation by deciding when an AI system should act autonomously or defer to a human expert. Existing L2D methods, however, assume static human performance, contradicting well-established findings on fatigue-induced degradation. We propose Fatigue-Aware Learning to Defer via Constrained Optimisation (FALCON), which explicitly models workload-varying human performance using psychologically grounded fatigue curves. FALCON formulates L2D as a Constrained Markov Decision Process (CMDP) whose state includes both task features and cumulative human workload, and optimises accuracy under human-AI cooperation budgets via PPO-Lagrangian training. We further introduce FA-L2D, a benchmark that systematically varies fatigue dynamics from near-static to rapidly degrading regimes. Experiments across multiple datasets show that FALCON consistently outperforms state-of-the-art L2D methods across coverage levels, generalises zero-shot to unseen experts with different fatigue patterns, and demonstrates the advantage of adaptive human-AI collaboration over AI-only or human-only decision-making when coverage lies strictly between 0 and 1.