mlsanityData to 14 Oct 2026

Rethinking LLM Bias Probing Using Lessons from the Social Sciences

arXiv 2503.00093 ↗First posted 28 Feb 2025cs.CY cs.AI cs.CLcited by 7

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

The proliferation of LLM bias probes introduces three significant challenges: (1) we lack principled criteria for choosing appropriate probes, (2) we lack a system for reconciling conflicting results across probes, and (3) we lack formal frameworks for reasoning about when (and why) probe results will generalize to real user behavior. We address these challenges by systematizing LLM social bias probing using actionable insights from social sciences. We then introduce EcoLevels - a framework that helps (a) determine appropriate bias probes, (b) reconcile conflicting findings across probes, and (c) generate predictions about bias generalization. Overall, we ground our analysis in social science research because many LLM probes are direct applications of human probes, and these fields have faced similar challenges when studying social bias in humans. Based on our work, we suggest how the next generation of LLM bias probing can (and should) benefit from decades of social science research.

Key facts. Cited by 7 papers in the map (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 30 of its 75 references that are in the map. First posted on arXiv 28 Feb 2025.

Builds on

30 of 75 references.

References · 30

  1. 2020StereoSet: Measuring stereotypical bias in pretrained language modelscited by 924
  2. 2016Semantics derived automatically from language corpora contain human-like biasescited by 901
  3. 2020CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Modelscited by 710
  4. 2021BBQ: A Hand-Built Bias Benchmark for Question Answeringcited by 701
  5. 2018Gender Bias in Coreference Resolution: Evaluation and Debiasing Methodscited by 659
  6. 2023Bias and Fairness in Large Language Models: A Surveycited by 639
  7. 2024WildChat: 1M ChatGPT Interaction Logs in the Wildcited by 629
  8. 2023LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Datasetcited by 461
  9. 2018Gender Bias in Coreference Resolutioncited by 442
  10. 2021BOLD: Dataset and Metrics for Measuring Biases in Open-Ended Language Generationcited by 412
  11. 2019On Measuring Social Biases in Sentence Encoderscited by 398
  12. 2019Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Settingcited by 349
18 more
  1. 2020Measuring and Reducing Gendered Correlations in Pre-trained Modelscited by 224
  2. 2023"Kelly is a Warm Person, Joseph is a Role Model": Gender Biases in LLM-Generated Reference Letterscited by 199
  3. 2023Gender bias and stereotypes in Large Language Modelscited by 187
  4. 2020Detecting Emergent Intersectional Biases: Contextualized Word Embeddings Contain a Distribution of Human-like Biasescited by 167
  5. 2023A Survey on Fairness in Large Language Modelscited by 95
  6. 2020Unmasking Contextual Stereotypes: Measuring and Mitigating BERT's Gender Biascited by 82
  7. 2022On the Intrinsic and Extrinsic Fairness Evaluation Metrics for Contextualized Language Representationscited by 75
  8. 2024Disclosure and Mitigation of Gender Bias in LLMscited by 49
  9. 2024Gender Bias in Large Language Models across Multiple Languagescited by 35
  10. 2023Social Bias Probing: Fairness Benchmarking for Language Modelscited by 32
  11. 2023Unveiling Gender Bias in Terms of Profession Across LLMs: Analyzing and Addressing Sociological Implicationscited by 29
  12. 2024The Silicon Ceiling: Auditing GPT's Race and Gender Biases in Hiringcited by 29
  13. 2024Investigating Implicit Bias in Large Language Models: A Large-Scale Study of Over 50 LLMscited by 25
  14. 2023Are Emily and Greg Still More Employable than Lakisha and Jamal? Investigating Algorithmic Hiring Bias in the Era of ChatGPTcited by 24
  15. 2022What Makes a Good Explanation?: A Harmonized View of Properties of Explanationscited by 22
  16. 2024Gender Bias in LLM-generated Interview Responsescited by 10
  17. 2024Assessing Gender Bias in LLMs: Comparing LLM Outputs with Human Perceptions and Official Statisticscited by 4
  18. 2024Evaluating Gender Bias in Large Language Modelscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 6.

7 papers.

  1. 2026One Persona, Many Cues, Different Results: How Sociodemographic Cues Impact LLM Personalizationcited by 9
  2. 2025Measuring Bias or Measuring the Task: Understanding the Brittle Nature of LLM Gender Biasescited by 8
  3. 2026Different Demographic Cues Yield Inconsistent Conclusions About LLM Personalization and Biascited by 4
  4. 2026Culturally uneven urban perception in large language modelscited by 1
  5. 2026Calibrating WEAT Against Anisotropy: ZCA Whitening as a Geometric Pre-Processing Step for Embedding Association Testscited by 0
  6. 2026An Investigation of the NeurIPS and ICML 2025 Position Trackscited by 0
  7. 2026Efficient Active Auditing of Multi-Group Fairness with Bias Probescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Evaluating Second-Order Bias of LLMs Through Epistemic Entitlement
  2. 2025Recognising, Anticipating, and Mitigating LLM Pollution of Online Behavioural Research
  3. 2023Do LLMs exhibit human-like response biases? A case study in survey design
  4. 2024With a Grain of SALT: Are LLMs Fair Across Social Dimensions?
  5. 2025GenderBench: Evaluation Suite for Gender Biases in LLMs
  6. 2026BiasTrace: Linking Reasoning Behaviours to Biased Outputs in LLMs
  7. 2025Do LLMs Align Human Values Regarding Social Biases? Judging and Explaining Social Biases with LLMs
  8. 2026To Compare, or Not to Compare: On Methodological Practices in Evaluating Social Bias