Benchmark
Bias Benchmark for QA
A benchmark measuring social bias in language model outputs across 11 demographic categories including age, race, gender, and religion. Higher accuracy on disambiguated questions indicates less reliance on social stereotypes.
What it measures
QA accuracy under ambiguous and disambiguated contexts across 11 protected characteristic categories covering 58,492 unique examples
Safety areas: General safety and Alignment
Score direction: higher_is_better
Public model scores
No comparable public model scores found yet.
Primary source
BBQ: A Hand-Built Bias Benchmark for Question AnsweringStatus: Verified. Last verified: 2026-06-13.