GenderCARE: A Comprehensive Framework for Assessing and Reducing Gender Bias in Large Language Models

2024-08-22

Conference on Computer and Communications Security

AI Summary

This work addresses the limited practicality, insufficient inclusivity, and bias-introduction risks of existing LLM gender bias evaluation benchmarks. We propose the first end-to-end gender fairness framework covering both assessment and debiasing. Methodologically: (1) we introduce a multidimensional fairness criterion—first standardizing evaluation to include transgender and non-binary individuals; (2) we construct GenderPair, a paired benchmark enabling fine-grained, counterfactual-aware bias measurement; and (3) we design a synergistic debiasing paradigm combining counterfactual data augmentation with gender-aware directional fine-tuning. Our contributions include interpretable, robust, and task-preserving fairness optimization. Evaluated on 17 mainstream LLMs, our approach reduces gender bias by over 35% on average (up to >90%), while maintaining primary task performance within ±2%. This significantly advances both fairness and practical utility of LLMs.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Large language models (LLMs) have exhibited remarkable capabilities in natural language generation, but they have also been observed to magnify societal biases, particularly those related to gender. In response to this issue, several benchmarks have been proposed to assess gender bias in LLMs. However, these benchmarks often lack practical flexibility or inadvertently introduce biases. To address these shortcomings, we introduce GenderCARE, a comprehensive framework that encompasses innovative Criteria, bias Assessment, Reduction techniques, and Evaluation metrics for quantifying and mitigating gender bias in LLMs. To begin, we establish pioneering criteria for gender equality benchmarks, spanning dimensions such as inclusivity, diversity, explainability, objectivity, robustness, and realisticity. Guided by these criteria, we construct GenderPair, a novel pair-based benchmark designed to assess gender bias in LLMs comprehensively. Our benchmark provides standardized and realistic evaluations, including previously overlooked gender groups such as transgender and non-binary individuals. Furthermore, we develop effective debiasing techniques that incorporate counterfactual data augmentation and specialized fine-tuning strategies to reduce gender bias in LLMs without compromising their overall performance. Extensive experiments demonstrate a significant reduction in various gender bias benchmarks, with reductions peaking at over 90% and averaging above 35% across 17 different LLMs. Importantly, these reductions come with minimal variability in mainstream language tasks, remaining below 2%. By offering a realistic assessment and tailored reduction of gender biases, we hope that our GenderCARE can represent a significant step towards achieving fairness and equity in LLMs. More details are available at https://github.com/kstanghere/GenderCARE-ccs24.

Problem

• Assessing gender bias in language models
• Reducing bias without performance loss
• Including diverse gender groups in evaluation

Innovation

• Comprehensive gender bias framework
• Novel pair-based benchmark GenderPair
• Effective debiasing techniques counterfactual augmentation