FairHireBench: A Cross-Generational Intersectional Bias Benchmark for LLMs in Automated Hiring
Dataset Description
FairHireBench is a comprehensive benchmark comprising 10,005 candidate profile records across 2,001 unique candidates spanning 15 intersectional demographic groups (5 racial/ethnic x 3 gender categories) for evaluating bias in AI-driven hiring systems.
Each profile represents a mid-level software engineer candidate with the following attributes:
