BiasFreeBench: a Benchmark for Mitigating Bias in Large Language Model Responses

2025/09/30 by Xin Xu, Xu, Xin, Xiaoqiao He +9 · 1 citation
Computer Science · Social Sciences · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Computational and Text Analysis Methods #Computers and Society (cs.CY) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Natural Language Processing Techniques #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2510.00232

openalex publication_date 2025/09/30 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Existing studies on bias mitigation methods for large language models (LLMs) use diverse baselines and metrics to evaluate debiasing performance, leading to inconsistent comparisons among them. Moreover, their evaluations are mostly based on the comparison between LLMs' probabilities of biased and unbiased contexts, which ignores the gap between such evaluations and real-world use cases where users interact with LLMs by reading model responses and expect fair and safe outputs rather than LLMs' probabilities. To enable consistent evaluation across debiasing methods and bridge this gap, we introduce BiasFreeBench, an empirical benchmark that comprehensively compares eight mainstream bias mitigation techniques (covering four prompting-based and four training-based methods) on two test scenarios (multi-choice QA and open-ended multi-turn QA) by reorganizing existing datasets into a unified query-response setting. We further introduce a response-level metric, Bias-Free Score, to measure the extent to which LLM responses are fair, safe, and anti-stereotypical. Debiasing performances are systematically compared and analyzed across key dimensions: the prompting vs. training paradigm, model size, and generalization of different training strategies to unseen bias types. We release our benchmark, aiming to establish a unified testbed for bias mitigation research.

Citations

• BiasGuard: A Reasoning-enhanced Bias Detection Tool For Large Language Models
• FairSteer: Inference Time Debiasing for LLMs with Dynamic Activation Steering
• BiasEdit: Debiasing Stereotyped Language Models via Model Editing
• Mitigating Social Bias in Large Language Models: A Multi-Objective Approach within a Multi-Agent Framework
• Qwen2.5 Technical Report
• FairMT-Bench: Benchmarking Fairness for Multi-turn Dialogue in Conversational LLMs
• Understanding the Interplay between Parametric and Contextual Knowledge for Large Language Models
• Towards Implicit Bias Detection and Mitigation in Multi-Agent LLM Interactions
• Identifying and Mitigating Social Bias Knowledge in Language Models
• BiasDPO: Mitigating Bias in Language Models through Direct Preference Optimization
• BiasAlert: A Plug-and-play Tool for Social Bias Detection in LLMs
• CEB: Compositional Evaluation Benchmark for Fairness in Large Language Models
• GenderAlign: An Alignment Dataset for Mitigating Gender Bias in Large Language Models
• MBIAS: Mitigating Bias in Large Language Models While Retaining Context
• Thinking Fair and Slow: On the Efficacy of Structured Prompts for Debiasing Language Models
• Prompting Techniques for Reducing Social Bias in LLMs through System 1 and System 2 Cognitive Processes
• LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models
• Knowledge Conflicts for LLMs: A Survey
• Rebuilding ROME : Resolving Model Collapse during Sequential Model Editing
• Cognitive Bias in Decision-Making with LLMs
• Self-Debiasing Large Language Models: Zero-Shot Recognition and Reduction of Stereotypes
• Model Editing Harms General Abilities of Large Language Models: Regularization to the Rescue
• DeepSeek LLM: Scaling Open-Source Language Models with Longtermism
• Debiasing Algorithm through Model Adaptation
• Safe RLHF: Safe Reinforcement Learning from Human Feedback
• In-Contextual Gender Bias Suppression for Large Language Models
• Efficient Memory Management for Large Language Model Serving with PagedAttention
• Bias and Fairness in Large Language Models: A Survey
• Bias and Fairness in Large Language Models: A Survey
• Lost in the Middle: How Language Models Use Long Contexts
• Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• Having Beer after Prayer? Measuring Cultural Bias in Large Language Models
• Adaptive Chameleon or Stubborn Sloth: Revealing the Behavior of Large Language Models in Knowledge Conflicts
• This Prompt is Measuring : Evaluating Bias Evaluation in Language Models
• Reflexion: Language Agents with Verbal Reinforcement Learning
• GPT-4 Technical Report
• The Capacity for Moral Self-Correction in Large Language Models
• Parameter-efficient Modularised Bias Mitigation via AdapterFusion
• Editing Models with Task Arithmetic
• Rich Knowledge Sources Bring Complex Knowledge Conflicts: Recalibrating Models to Reflect Conflicting Evidence
• Towards Realistic Low-resource Relation Extraction: A Benchmark with Empirical Baseline Study
• Self-Consistency Improves Chain of Thought Reasoning in Language Models
• Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• An Empirical Survey of the Effectiveness of Debiasing Techniques for Pre-trained Language Models
• BBQ: A Hand-Built Bias Benchmark for Question Answering
• TruthfulQA: Measuring How Models Mimic Human Falsehoods
• CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models
• RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
• Towards Debiasing Sentence Representations
• Towards Controllable Biases in Language Generation
• StereoSet: Measuring stereotypical bias in pretrained language models
• Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection
• HuggingFace's Transformers: State-of-the-art Natural Language Processing
• Counterfactual Data Augmentation for Mitigating Gender Stereotypes in Languages with Rich Morphology
• BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions
• Gender Bias in Neural Natural Language Processing
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Cited by

• From Defender to Devil? Unintended Risk Interactions Induced by LLM Defenses

Related

• Evaluating Nuanced Bias in Large Language Model Free Response Answers
• Bias and Fairness in Large Language Models: A Survey
• Bias and Fairness in Large Language Models: A Survey
• BiasFilter: An Inference-Time Debiasing Framework for Large Language Models
• Rethinking Prompt-based Debiasing in Large Language Models