GKnow: Measuring the Entanglement of Gender Bias and Factual Gender

• URL: http://arxiv.org/abs/2605.12299v1
• Date: Tue, 12 May 2026 15:52:30 GMT
• Title: GKnow: Measuring the Entanglement of Gender Bias and Factual Gender
• Abstract summary: gknow is a benchmark to assess gender knowledge and gender bias in language models across different types of gender-related predictions.<n>We test the impact of neuron ablation on benchmarks for disentangling stereotypical and factual gender.
• Score: 47.39675194552945
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent works have analyzed the impact of individual components of neural networks on gendered predictions, often with a focus on mitigating gender bias. However, mechanistic interpretations of gender tend to (i) focus on a very specific gender-related task, such as gendered pronoun prediction, or (ii) fail to distinguish between the production of factually gendered outputs (the correct assumption of gender given a word that carries gender as a semantic property) and gender biased outputs (based on a stereotype). To address these issues, we curate \gknow, a benchmark to assess gender knowledge and gender bias in language models across different types of gender-related predictions. \gknow allows us to identify and analyze circuits and individual neurons responsible for gendered predictions. We test the impact of neuron ablation on benchmarks for disentangling stereotypical and factual gender (DiFair and the test set of GKnow), as well as StereoSet. Results show that gender bias and factual gender are severely entangled on the level of both circuits and neurons, entailing that ablation is an unreliable debiasing method. Furthermore, we show that benchmarks for evaluating gender bias can hide the decrease in factual gender knowledge that accompanies neuron ablation. We curate GKnow as a contribution to the continuous development of robust gender bias benchmarks.

Related papers

• Neuron-Level Interventions for Gendered and Gender-Neutral Generation in Language Models [4.464224430863637]
  Language models (LMs) can produce gendered language and stereotypes even when given neutral prompts.<n>We study gender-specific neurons in LMs across three categories: feminine, masculine, and gender-neutral.<n>We propose a neuron-level intervention method to identify neurons that are strongly tied to each gender category.
  arXiv  Detail & Related papers  (2026-05-29T01:21:40Z)
• Gender Trouble in Language Models: An Empirical Audit Guided by Gender Performativity Theory [0.19116784879310028]
  Language models encode and perpetuate harmful gendered stereotypes.<n>Gendered terms that do not neatly fall into one of these binary categories are erased and pathologized.<n>Our findings lead us to call for a re-evaluation of how gendered harms in language models are defined and addressed.
  arXiv  Detail & Related papers  (2025-05-20T08:36:47Z)
• Beyond Binary Gender: Evaluating Gender-Inclusive Machine Translation with Ambiguous Attitude Words [85.48043537327258]
  Existing machine translation gender bias evaluations are primarily focused on male and female genders. This study presents a benchmark AmbGIMT (Gender-Inclusive Machine Translation with Ambiguous attitude words) We propose a novel process to evaluate gender bias based on the Emotional Attitude Score (EAS), which is used to quantify ambiguous attitude words.
  arXiv  Detail & Related papers  (2024-07-23T08:13:51Z)
• DiFair: A Benchmark for Disentangled Assessment of Gender Knowledge and Bias [13.928591341824248]
  Debiasing techniques have been proposed to mitigate the gender bias that is prevalent in pretrained language models. These are often evaluated on datasets that check the extent to which the model is gender-neutral in its predictions. This evaluation protocol overlooks the possible adverse impact of bias mitigation on useful gender knowledge.
  arXiv  Detail & Related papers  (2023-10-22T15:27:16Z)
• VisoGender: A dataset for benchmarking gender bias in image-text pronoun resolution [80.57383975987676]
  VisoGender is a novel dataset for benchmarking gender bias in vision-language models. We focus on occupation-related biases within a hegemonic system of binary gender, inspired by Winograd and Winogender schemas. We benchmark several state-of-the-art vision-language models and find that they demonstrate bias in resolving binary gender in complex scenes.
  arXiv  Detail & Related papers  (2023-06-21T17:59:51Z)
• Towards Understanding Gender-Seniority Compound Bias in Natural Language Generation [64.65911758042914]
  We investigate how seniority impacts the degree of gender bias exhibited in pretrained neural generation models. Our results show that GPT-2 amplifies bias by considering women as junior and men as senior more often than the ground truth in both domains. These results suggest that NLP applications built using GPT-2 may harm women in professional capacities.
  arXiv  Detail & Related papers  (2022-05-19T20:05:02Z)
• Theories of "Gender" in NLP Bias Research [0.0]
  We survey nearly 200 articles concerning gender bias in NLP. We find that the majority of the articles do not make their theorization of gender explicit. Many conflate sex characteristics, social gender, and linguistic gender in ways that disregard the existence and experience of trans, nonbinary, and intersex people.
  arXiv  Detail & Related papers  (2022-05-05T09:20:53Z)
• Evaluating Gender Bias in Natural Language Inference [5.034017602990175]
  We propose an evaluation methodology to measure gender bias in natural language understanding through inference. We use our challenge task to investigate state-of-the-art NLI models on the presence of gender stereotypes using occupations. Our findings suggest that three models trained on MNLI and SNLI datasets are significantly prone to gender-induced prediction errors.
  arXiv  Detail & Related papers  (2021-05-12T09:41:51Z)
• Gender Stereotype Reinforcement: Measuring the Gender Bias Conveyed by Ranking Algorithms [68.85295025020942]
  We propose the Gender Stereotype Reinforcement (GSR) measure, which quantifies the tendency of a Search Engines to support gender stereotypes. GSR is the first specifically tailored measure for Information Retrieval, capable of quantifying representational harms.
  arXiv  Detail & Related papers  (2020-09-02T20:45:04Z)
• Multi-Dimensional Gender Bias Classification [67.65551687580552]
  Machine learning models can inadvertently learn socially undesirable patterns when training on gender biased text. We propose a general framework that decomposes gender bias in text along several pragmatic and semantic dimensions. Using this fine-grained framework, we automatically annotate eight large scale datasets with gender information.
  arXiv  Detail & Related papers  (2020-05-01T21:23:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.