It is well documented that NLP models learn social biases, but little work
has been done on how these biases manifest in model outputs for applied tasks
like question answering (QA). We introduce the Bias Benchmark for QA (BBQ), a
dataset of question sets constructed by the authors that highlight attested
social biases against people belonging to protected classes along nine social
dimensions relevant for U.S. English-speaking contexts. Our task evaluates
model responses at two levels: (i) given an under-informative context, we test
how strongly responses reflect social biases, and (ii) given an adequately
informative context, we test whether the model's biases override a correct
answer choice. We find that models often rely on stereotypes when the context
is under-informative, meaning the model's outputs consistently reproduce
harmful biases in this setting. Though models are more accurate when the
context provides an informative answer, they still rely on stereotypes and
average up to 3.4 percentage points higher accuracy when the correct answer
aligns with a social bias than when it conflicts, with this difference widening
to over 5 points on examples targeting gender for most models tested.

该研究旨在通过提出一种基准数据集 —— 即 Bias Benchmark for QA（BBQ）来测试 NLP 模型的社会偏见。研究发现当上下文信息不充分时，模型通常依赖于固有的刻板印象，导致输出结果中仍存在有害偏见。同时，即便给出充分信息，模型依然受固有偏见的影响，在正确答案与社会偏见相冲突时的准确率差异可高达 5 个百分点。