Bias Mitigation or Cultural Commonsense? Evaluating LLMs with a Japanese Dataset

• URL: http://arxiv.org/abs/2509.24468v1
• Date: Mon, 29 Sep 2025 08:45:50 GMT
• Title: Bias Mitigation or Cultural Commonsense? Evaluating LLMs with a Japanese Dataset
• Abstract summary: Large language models (LLMs) exhibit social biases, prompting the development of various debiasing methods.<n>Previous research has evaluated the impact of bias mitigation primarily through tasks measuring general language understanding.<n>Cultural commonsense is closely related to social biases, as both are rooted in social norms and values.
• Score: 34.65197313327234
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) exhibit social biases, prompting the development of various debiasing methods. However, debiasing methods may degrade the capabilities of LLMs. Previous research has evaluated the impact of bias mitigation primarily through tasks measuring general language understanding, which are often unrelated to social biases. In contrast, cultural commonsense is closely related to social biases, as both are rooted in social norms and values. The impact of bias mitigation on cultural commonsense in LLMs has not been well investigated. Considering this gap, we propose SOBACO (SOcial BiAs and Cultural cOmmonsense benchmark), a Japanese benchmark designed to evaluate social biases and cultural commonsense in LLMs in a unified format. We evaluate several LLMs on SOBACO to examine how debiasing methods affect cultural commonsense in LLMs. Our results reveal that the debiasing methods degrade the performance of the LLMs on the cultural commonsense task (up to 75% accuracy deterioration). These results highlight the importance of developing debiasing methods that consider the trade-off with cultural commonsense to improve fairness and utility of LLMs.

Related papers

• A word association network methodology for evaluating implicit biases in LLMs compared to humans [0.0]
  We present a novel word association network methodology for evaluating implicit biases in Large language models (LLMs)<n>Our method taps into the implicit relational structures encoded in LLMs, providing both quantitative and qualitative assessments of bias.<n>To demonstrate the utility of our methodology, we apply it to both humans and several widely used LLMs to investigate social biases related to gender, religion, ethnicity, sexual orientation, and political party.
  arXiv  Detail & Related papers  (2025-10-28T15:03:18Z)
• Which Cultural Lens Do Models Adopt? On Cultural Positioning Bias and Agentic Mitigation in LLMs [53.07843733899881]
  Large language models (LLMs) have unlocked a wide range of downstream generative applications.<n>We find that they also risk perpetuating subtle fairness issues tied to culture, positioning their generations from the perspectives of the mainstream US culture.<n>We propose 2 inference-time mitigation methods to resolve these biases.
  arXiv  Detail & Related papers  (2025-09-25T12:28:25Z)
• McBE: A Multi-task Chinese Bias Evaluation Benchmark for Large Language Models [26.202296897643382]
  We present a Multi-task Chinese Bias Evaluation Benchmark (McBE) that includes 4,077 bias evaluation instances.<n>This dataset provides extensive category coverage, content diversity, and measuring comprehensiveness.<n>We conduct an in-depth analysis of results, offering novel insights into bias in large language models (LLMs)
  arXiv  Detail & Related papers  (2025-07-02T19:04:56Z)
• An Evaluation of Cultural Value Alignment in LLM [27.437888319382893]
  We conduct the first large-scale evaluation of LLM culture assessing 20 countries' cultures and languages across ten LLMs.<n>Our findings show that the output over all models represents a moderate cultural middle ground.<n> Deeper investigation sheds light on the influence of model origin, prompt language, and value dimensions on cultural output.
  arXiv  Detail & Related papers  (2025-04-11T09:13:19Z)
• Social Debiasing for Fair Multi-modal LLMs [59.61512883471714]
  Multi-modal Large Language Models (MLLMs) have dramatically advanced the research field and delivered powerful vision-language understanding capabilities.<n>These models often inherit deep-rooted social biases from their training data, leading to uncomfortable responses with respect to attributes such as race and gender.<n>This paper addresses the issue of social biases in MLLMs by introducing a comprehensive counterfactual dataset with multiple social concepts.
  arXiv  Detail & Related papers  (2024-08-13T02:08:32Z)
• Evaluating Implicit Bias in Large Language Models by Attacking From a Psychometric Perspective [66.34066553400108]
  We conduct a rigorous evaluation of large language models' implicit bias towards certain demographics.<n>Inspired by psychometric principles, we propose three attack approaches, i.e., Disguise, Deception, and Teaching.<n>Our methods can elicit LLMs' inner bias more effectively than competitive baselines.
  arXiv  Detail & Related papers  (2024-06-20T06:42:08Z)
• Understanding the Capabilities and Limitations of Large Language Models for Cultural Commonsense [98.09670425244462]
  Large language models (LLMs) have demonstrated substantial commonsense understanding. This paper examines the capabilities and limitations of several state-of-the-art LLMs in the context of cultural commonsense tasks.
  arXiv  Detail & Related papers  (2024-05-07T20:28:34Z)
• Large Language Models are Geographically Biased [47.88767211956144]
  We study what Large Language Models (LLMs) know about the world we live in through the lens of geography. We show various problematic geographic biases, which we define as systemic errors in geospatial predictions.
  arXiv  Detail & Related papers  (2024-02-05T02:32:09Z)
• Exploring the Jungle of Bias: Political Bias Attribution in Language Models via Dependency Analysis [86.49858739347412]
  Large Language Models (LLMs) have sparked intense debate regarding the prevalence of bias in these models and its mitigation. We propose a prompt-based method for the extraction of confounding and mediating attributes which contribute to the decision process. We find that the observed disparate treatment can at least in part be attributed to confounding and mitigating attributes and model misalignment.
  arXiv  Detail & Related papers  (2023-11-15T00:02:25Z)
• Investigating Subtler Biases in LLMs: Ageism, Beauty, Institutional, and Nationality Bias in Generative Models [0.0]
  This paper investigates bias along less-studied but still consequential, dimensions, such as age and beauty. We ask whether LLMs hold wide-reaching biases of positive or negative sentiment for specific social groups similar to the "what is beautiful is good" bias found in people in experimental psychology.
  arXiv  Detail & Related papers  (2023-09-16T07:07:04Z)
• A Survey on Fairness in Large Language Models [28.05516809190299]
  Large Language Models (LLMs) have shown powerful performance and development prospects. LLMs can capture social biases from unprocessed training data and propagate the biases to downstream tasks. Unfair LLM systems have undesirable social impacts and potential harms.
  arXiv  Detail & Related papers  (2023-08-20T03:30:22Z)

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.