mlsanityData to 14 Oct 2026

On the Independence of Association Bias and Empirical Fairness in Language Models

arXiv 2304.10153 ↗First posted 20 Apr 2023cs.CL cs.HCcited by 29

Not on the map. Why?

Abstract (arXiv)

The societal impact of pre-trained language models has prompted researchers to probe them for strong associations between protected attributes and value-loaded terms, from slur to prestigious job titles. Such work is said to probe models for bias or fairness-or such probes 'into representational biases' are said to be 'motivated by fairness'-suggesting an intimate connection between bias and fairness. We provide conceptual clarity by distinguishing between association biases (Caliskan et al., 2022) and empirical fairness (Shen et al., 2022) and show the two can be independent. Our main contribution, however, is showing why this should not come as a surprise. To this end, we first provide a thought experiment, showing how association bias and empirical fairness can be completely orthogonal. Next, we provide empirical evidence that there is no correlation between bias metrics and fairness metrics across the most widely used language models. Finally, we survey the sociological and psychological literature and show how this literature provides ample support for expecting these metrics to be uncorrelated.

Key facts. Cited by 29 papers in the map (data through 14 Oct 2026). Builds on 39 of its 67 references that are in the map. First posted on arXiv 20 Apr 2023.

Builds on

39 of 67 references.

References · 39

  1. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,917
  2. 2016Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddingscited by 1,312
  3. 2020Language (Technology) is Power: A Critical Survey of "Bias" in NLPcited by 956
  4. 2016Semantics derived automatically from language corpora contain human-like biasescited by 901
  5. 2016Inherent Trade-Offs in the Fair Determination of Risk Scorescited by 586
  6. 2019On Measuring Social Biases in Sentence Encoderscited by 398
  7. 2018Fairness Without Demographics in Repeated Loss Minimizationcited by 364
  8. 2019Mitigating Gender Bias in Natural Language Processing: Literature Reviewcited by 287
  9. 2019Measuring Bias in Contextualized Word Representationscited by 276
  10. 2019Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Themcited by 268
  11. 2020Social Biases in NLP Models as Barriers for Persons with Disabilitiescited by 182
  12. 2018Learning Gender-Neutral Word Embeddingscited by 169
27 more
  1. 2020Intrinsic Bias Metrics Do Not Correlate with Application Biascited by 148
  2. 2019Predictive Biases in Natural Language Processing Models: A Conceptual Framework and Overviewcited by 137
  3. 2019Assessing Social and Intersectional Biases in Contextualized Word Representationscited by 135
  4. 2021Sustainable Modular Debiasing of Language Modelscited by 100
  5. 2022XAI for Transformers: Better Explanations through Conservative Propagationcited by 100
  6. 2021A Clarification of the Nuances in the Fairness Metrics Landscapecited by 98
  7. 2018Understanding the Origins of Bias in Word Embeddingscited by 91
  8. 2019Does Gender Matter? Towards Fairness in Dialogue Systemscited by 89
  9. 2021A Survey on Gender Bias in Natural Language Processingcited by 89
  10. 2019Fairness risk measurescited by 85
  11. 2021Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metricscited by 83
  12. 2020Unmasking Contextual Stereotypes: Measuring and Mitigating BERT's Gender Biascited by 82
  13. 2022On the Intrinsic and Extrinsic Fairness Evaluation Metrics for Contextualized Language Representationscited by 75
  14. 2020Measuring Social Biases in Grounded Vision and Language Embeddingscited by 61
  15. 2022Perturbation Augmentation for Fairer NLPcited by 52
  16. 2019What's in a Name? Reducing Bias in Bios without Access to Protected Attributescited by 51
  17. 2022FairLex: A Multilingual Benchmark for Evaluating Fairness in Legal Text Processingcited by 49
  18. 2022Debiasing isn't enough! -- On the Effectiveness of Debiasing MLMs and their Social Biases in Downstream Taskscited by 48
  19. 2020Multi-SimLex: A Large-Scale Evaluation of Multilingual and Cross-Lingual Lexical Semantic Similaritycited by 38
  20. 2020Analyzing Political Bias and Unfairness in News Articles at Different Levels of Granularitycited by 32
  21. 2022Square One Bias in NLP: Towards a Multi-Dimensional Exploration of the Research Manifoldcited by 29
  22. 2020Does enforcing fairness mitigate biases caused by subpopulation shift?cited by 25
  23. 2022Gender Bias in Word Embeddings: A Comprehensive Analysis of Frequency, Syntax, and Semanticscited by 24
  24. 2021Measuring Fairness with Biased Rulers: A Survey on Quantifying Biases in Pretrained Language Modelscited by 20
  25. 2017The impossibility of "fairness": a generalized impossibility result for decisionscited by 12
  26. 2021DebIE: A Platform for Implicit and Explicit Debiasing of Word Embedding Spacescited by 5
  27. 2016Automating Political Bias Predictioncited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 7; 2024: 13; 2025: 6; 2026: 3.

29 papers.

Sort:

By area:

  1. 2023Bias and Fairness in Large Language Models: A Surveycited by 639
  2. 2024Fairness in Large Language Models: A Taxonomic Surveycited by 58
  3. 2024FairMT-Bench: Benchmarking Fairness for Multi-turn Dialogue in Conversational LLMscited by 36
  4. 2024Content Moderation by LLM: From Accuracy to Legitimacycited by 30
  5. 2024Investigating Implicit Bias in Large Language Models: A Large-Scale Study of Over 50 LLMscited by 25
  6. 2024BiasAlert: A Plug-and-play Tool for Social Bias Detection in LLMscited by 21
  7. 2024Bias and Volatility: A Statistical Framework for Evaluating Large Language Model's Stereotypes and the Associated Generation Inconsistencycited by 15
  8. 2024Evaluating Fairness in Large Vision-Language Models Across Diverse Demographic Attributes and Promptscited by 15
  9. 2023Tackling Bias in Pre-trained Language Models: Current Trends and Under-represented Societiescited by 11
  10. 2023On the Interplay between Fairness and Explainabilitycited by 7
  11. 2023Evaluating Bias and Fairness in Gender-Neutral Pretrained Vision-and-Language Modelscited by 6
  12. 2024How far can bias go? Tracing bias from pretraining data to alignmentcited by 5
17 more
  1. 2023Factual and Personalized Recommendations using Language Models and Reinforcement Learningcited by 4
  2. 2023Fair Text Classification with Wasserstein Independencecited by 4
  3. 2023A Group Fairness Lens for Large Language Modelscited by 4
  4. 2024Addressing Both Statistical and Causal Gender Fairness in NLP Modelscited by 4
  5. 2024Why Don't Prompt-Based Fairness Metrics Correlate?cited by 4
  6. 2025Blind Men and the Elephant: Diverse Perspectives on Gender Stereotypes in Benchmark Datasetscited by 3
  7. 2025Fair Text Classification via Transferable Representationscited by 3
  8. 2025Exploring Gender Bias in Large Language Models: An In-depth Dive into the German Languagecited by 3
  9. 2024Detecting Bias in Large Language Models: Fine-tuned KcBERTcited by 2
  10. 2025Social Bias in Multilingual Language Models: A Surveycited by 2
  11. 2026When LLMs Imagine People: A Human-Centered Persona Brainstorm Audit for Bias and Fairness in Creative Applicationscited by 2
  12. 2024Fairness Certification for Natural Language Processing and Large Language Modelscited by 1
  13. 2024Ethics Whitepaper: Whitepaper on Ethical Research into Large Language Modelscited by 1
  14. 2025A Critical Review of Predominant Bias in Neural Networkscited by 0
  15. 2025How do datasets, developers, and models affect biases in a low-resourced language?: The Case of the Bengali Languagecited by 0
  16. 2026Calibrating WEAT Against Anisotropy: ZCA Whitening as a Geometric Pre-Processing Step for Embedding Association Testscited by 0
  17. 2026Mitigating Bias in Large Vision-Language Models via Counterfactual Ensemble Decodingcited by 0