mlsanityData to 7 Oct 2026

InterFair: Debiasing with Natural Language Feedback for Fair Interpretable Predictions

arXiv 2210.07440 ↗First posted 14 Oct 2022cs.CLcited by 7

On the map: LLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment

Abstract (arXiv)

Debiasing methods in NLP models traditionally focus on isolating information related to a sensitive attribute (e.g., gender or race). We instead argue that a favorable debiasing method should use sensitive information 'fairly,' with explanations, rather than blindly eliminating it. This fair balance is often subjective and can be challenging to achieve algorithmically. We explore two interactive setups with a frozen predictive model and show that users able to provide feedback can achieve a better and fairer balance between task performance and bias mitigation. In one setup, users, by interacting with test examples, further decreased bias in the explanations (5-8%) while maintaining the same prediction accuracy. In the other setup, human feedback was able to disentangle associated bias and predictive information from the input leading to superior bias mitigation and improved task performance (4-5%) simultaneously.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 18 of its 37 references that are in the map. First posted on arXiv 14 Oct 2022.

Builds on

18 of 37 references.

References · 18

  1. 2018Mitigating Unwanted Biases with Adversarial Learningcited by 663
  2. 2019ERASER: A Benchmark to Evaluate Rationalized NLP Modelscited by 523
  3. 2021Fast Model Editing at Scalecited by 491
  4. 2020Null It Out: Guarding Protected Attributes by Iterative Nullspace Projectioncited by 442
  5. 2019Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Settingcited by 346
  6. 2016Rationalizing Neural Predictionscited by 281
  7. 2019Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Themcited by 268
  8. 2019Interpretable Neural Predictions with Differentiable Binary Variablescited by 126
  9. 2020Learning to Faithfully Rationalize by Constructioncited by 113
  10. 2022Rethinking Explainability as a Dialogue: A Practitioner's Perspectivecited by 52
  11. 2020Improving Compositional Generalization in Semantic Parsingcited by 42
  12. 2020Stereotypical Bias Removal for Hate Speech Detection Task using Knowledge-based Generalizationscited by 40
6 more
  1. 2021Learning to Repair: Repairing model output errors after deployment using a dynamic memory of feedbackcited by 35
  2. 2022Using Interactive Feedback to Improve the Accuracy and Explainability of Question Answering Systems Post-Deploymentcited by 25
  3. 2021Detect and Perturb: Neutral Rewriting of Biased and Sensitive Text via Gradient-based Decodingcited by 23
  4. 2021Unsupervised Enrichment of Persona-grounded Dialog with Background Storiescited by 14
  5. 2019A Human-in-the-loop Framework to Construct Context-aware Mathematical Notions of Outcome Fairnesscited by 11
  6. 2021Interscript: A dataset for interactive learning of scripts through error feedbackcited by 9

Built on by

Cited by year

Citing papers per year: 2023: 2; 2024: 4; 2025: 1.

7 papers.

  1. 2023Bias and Fairness in Large Language Models: A Surveycited by 626
  2. 2024Self-Debiasing Large Language Models: Zero-Shot Recognition and Reduction of Stereotypescited by 58
  3. 2023Sociodemographic Bias in Language Models: A Survey and Forward Pathcited by 27
  4. 2024Position: Towards Bidirectional Human-AI Alignmentcited by 15
  5. 2024Potential and Challenges of Model Editing for Social Debiasingcited by 10
  6. 2024Mitigating Gender Bias in Code Large Language Models via Model Editingcited by 3
  7. 2025LFTF: Locating First and Then Fine-Tuning for Mitigating Gender Bias in Large Language Modelscited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2025BiasFilter: An Inference-Time Debiasing Framework for Large Language Models
  2. 2026Decoding-Time Debiasing via Process Reward Models: From Controlled Fill-in to Open-Ended Generation
  3. 2025Debiasing LLMs by Masking Unfairness-Driving Attention Heads
  4. 2024A Multi-LLM Debiasing Framework
  5. 2024BiasDPO: Mitigating Bias in Language Models through Direct Preference Optimization
  6. 2025Bias after Prompting: Persistent Discrimination in Large Language Models
  7. 2024Thinking Fair and Slow: On the Efficacy of Structured Prompts for Debiasing Language Models
  8. 2024Evaluating Nuanced Bias in Large Language Model Free Response Answers