COBIAS: Assessing the Contextual Reliability of Bias Benchmarks for Language Models
Large Language Models (LLMs) often inherit biases from the web data they are trained on, which contains stereotypes and prejudices. Current methods for evaluating and mitigating these biases rely on bias-benchmark datasets. These benchmarks measure bias by observing an LLM's behavior on biased statements. However, these statements lack contextual considerations of the situations they try to present. To address this, we introduce a contextual reliability framework, which evaluates model robustness to biased statements by considering the various contexts in which they may appear. We develop the Context-Oriented Bias Indicator and Assessment Score (COBIAS) to measure a biased statement's reliability in detecting bias, based on the variance in model behavior across different contexts. To evaluate the metric, we augmented 2,291 stereotyped statements from two existing benchmark datasets by adding contextual information. We show that COBIAS aligns with human judgment on the contextual reliability of biased statements (Spearman's $\rho = 0.65, p = 3.4 * 10^{-60}$) and can be used to create reliable benchmarks, which would assist bias mitigation works.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
The Curious Case of Class Accuracy Imbalance in LLMs: Post-hoc Debiasing via Nonlinear Integer Programming
Large language models (LLMs) are good knowledge bases but struggle to perform equally well for all classes in text classification. This paper investigates the case of class accuracy imbalance in LLMs, where deeply entang…
Combinatorial OptimizationIn-Context LearningLanguage ModelingLanguage Modelling+2When three experiments are better than two: Avoiding intractable correlated aleatoric uncertainty by leveraging a novel bias--variance tradeoff
Real-world experimental scenarios are characterized by the presence of heteroskedastic aleatoric uncertainty, and this uncertainty can be correlated in batched settings. The bias--variance tradeoff can be used to write t…
Active LearningLet the Fuzzy Rule Speak: Enhancing In-context Learning Debiasing with Interpretability
One of the potential failures of large language models (LLMs) is their imbalanced class performances in text classification tasks. With in-context learning (ICL), LLMs yields good accuracy for some classes but low accura…
In-Context LearningMulti Class Text Classificationtext-classificationText ClassificationAssessing Social and Intersectional Biases in Contextualized Word Representations
Social bias in machine learning has drawn significant attention, with work ranging from demonstrations of bias in a multitude of applications, curating definitions of fairness for different contexts, to developing algori…
FairnessSentenceWord EmbeddingsDecoding AI Judgment: How LLMs Assess News Credibility and Bias
Large Language Models (LLMs) are increasingly used to assess news credibility, yet little is known about how they make these judgments. While prior research has examined political bias in LLM outputs or their potential f…
Fact Checking