Process-Aware Fairness Diagnosis for LLM Hiring Agents

Summary

Researchers introduce SCOPED-Hiring, a process-aware fairness diagnosis pipeline for LLM-based multi-agent hiring systems. It analyzes more than 311,000 structured decision trajectories across six diagnostic lenses and shows that balanced final hire rates can conceal unequal treatment during evaluation. Career gaps, proxy cues, and identity signals can affect suspicion, qualification judgments, and investigative behavior. Targeted repairs reduce total layered burden by 72.3% while shifting the hire rate by only 1.86 percentage points.