mlsanityData to 7 Oct 2026

KoSBi: A Dataset for Mitigating Social Bias Risks Towards Safer Large Language Model Application

arXiv 2305.17701 ↗First posted 28 May 2023cs.CLcited by 24

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) learn not only natural text generation abilities but also social biases against different demographic groups from real-world data. This poses a critical risk when deploying LLM-based applications. Existing research and resources are not readily applicable in South Korea due to the differences in language and culture, both of which significantly affect the biases and targeted demographic groups. This limitation requires localized social bias datasets to ensure the safe and effective deployment of LLMs. To this end, we present KO SB I, a new social bias dataset of 34k pairs of contexts and sentences in Korean covering 72 demographic groups in 15 categories. We find that through filtering-based moderation, social biases in generated content can be reduced by 16.47%p on average for HyperCLOVA (30B and 82B), and GPT-3.

Key facts. Cited by 24 papers in the map (data through 7 Oct 2026). Builds on 40 of its 52 references that are in the map. First posted on arXiv 28 May 2023.

Builds on

40 of 52 references.

References · 40

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  3. 2020Making Pre-trained Language Models Better Few-shot Learnerscited by 1,255
  4. 2020RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Modelscited by 1,235
  5. 2022Red Teaming Language Models with Language Modelscited by 983
  6. 2020Language (Technology) is Power: A Critical Survey of "Bias" in NLPcited by 951
  7. 2020StereoSet: Measuring stereotypical bias in pretrained language modelscited by 911
  8. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 709
  9. 2020Aligning AI With Shared Human Valuescited by 706
  10. 2020CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Modelscited by 703
  11. 2022Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learnedcited by 702
  12. 2021Cross-Task Generalization via Natural Language Crowdsourcing Instructionscited by 696
28 more
  1. 2021BBQ: A Hand-Built Bias Benchmark for Question Answeringcited by 683
  2. 2018Gender Bias in Coreference Resolution: Evaluation and Debiasing Methodscited by 657
  3. 2019The Woman Worked as a Babysitter: On Biases in Language Generationcited by 540
  4. 2018Gender Bias in Coreference Resolutioncited by 438
  5. 2020Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamicscited by 423
  6. 2022ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detectioncited by 419
  7. 2019Social Bias Frames: Reasoning about Social and Power Implications of Languagecited by 377
  8. 2020Social Chemistry 101: Learning to Reason about Social and Moral Normscited by 253
  9. 2020HateCheck: Functional Tests for Hate Speech Detection Modelscited by 215
  10. 2022WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creationcited by 200
  11. 2021Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasetscited by 179
  12. 2021Alignment of Language Agentscited by 167
  13. 2020Moral Stories: Situated Reasoning about Norms, Intents, Actions, and their Consequencescited by 133
  14. 2020UnQovering Stereotyping Biases via Underspecified Questionscited by 124
  15. 2021KLUE: Korean Language Understanding Evaluationcited by 113
  16. 2022ProsocialDialog: A Prosocial Backbone for Conversational Agentscited by 109
  17. 2021Detoxifying Language Models Risks Marginalizing Minority Voicescited by 106
  18. 2020Scruples: A Corpus of Community Ethical Judgments on 32,000 Real-Life Anecdotescited by 100
  19. 2021What Changes Can Large-scale Language Models Bring? Intensive Study on HyperCLOVA: Billions-scale Korean Generative Pretrained Transformerscited by 75
  20. 2022COLD: A Benchmark for Chinese Offensive Language Detectioncited by 74
  21. 2021On the Safety of Conversational Models: Taxonomy, Dataset, and Benchmarkcited by 73
  22. 2020BEEP! Korean Corpus of Online News Comments for Toxic Speech Detectioncited by 39
  23. 2022Emojis as Anchors to Detect Arabic Offensive Language and Hate Speechcited by 14
  24. 2022KOLD: Korean Offensive Language Datasetcited by 14
  25. 2022K-MHaS: A Multi-label Hate Speech Detection Dataset in Korean Online News Commentcited by 13
  26. 2022APEACH: Attacking Pejorative Expressions with Analysis on Crowd-Generated Hate Speech Evaluation Datasetscited by 8
  27. 2021Toxicity Detection can be Sensitive to the Conversational Contextcited by 6
  28. 2022Korean Online Hate Speech Dataset for Multilabel Classification: How Can Social Science Improve Dataset on Hate Speech?cited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 5; 2024: 10; 2025: 4; 2026: 5.

24 papers.

Sort:

By area:

  1. 2024Towards Measuring and Modeling "Culture" in LLMs: A Surveycited by 166
  2. 2023KoBBQ: Korean Bias Benchmark for Question Answeringcited by 71
  3. 2023Bias Testing and Mitigation in LLM-based Code Generationcited by 53
  4. 2023Prompt Packer: Deceiving LLMs through Compositional Instruction with Hidden Attackscited by 30
  5. 2024Navigating LLM Ethics: Advancements, Challenges, and Future Directionscited by 28
  6. 2024Culturally Aware and Adapted NLP: A Taxonomy and a Survey of the State of the Artcited by 26
  7. 2024PromptKD: Distilling Student-Friendly Knowledge for Generative Language Models via Prompt Tuningcited by 24
  8. 2024Optimizing Language Augmentation for Multilingual Large Language Models: A Case Study on Koreancited by 14
  9. 2024LLMs with Industrial Lens: Deciphering the Challenges and Prospects -- A Surveycited by 12
  10. 2024AdvisorQA: Towards Helpful and Harmless Advice-seeking Question Answering with Collective Intelligencecited by 12
  11. 2025A Review of Incorporating Psychological Theories in LLMscited by 5
  12. 2024No Size Fits All: The Perils and Pitfalls of Leveraging LLMs Vary with Company Sizecited by 3
12 more
  1. 2025BenchHub: A Unified Benchmark Suite for Holistic and Customizable LLM Evaluationcited by 3
  2. 2024On the Consideration of AI Openness: Can Good Intent Be Abused?cited by 2
  3. 2026CAGE: A Framework for Culturally Adaptive Red-Teaming Benchmark Generationcited by 2
  4. 2023OccuQuest: Mitigating Occupational Bias for Inclusive Large Language Modelscited by 1
  5. 2023Automatic Construction of a Korean Toxic Instruction Dataset for Ethical Tuning of Large Language Modelscited by 1
  6. 2025Social Bias in Multilingual Language Models: A Surveycited by 1
  7. 2024Reducing Large Language Model Bias with Emphasis on 'Restricted Industries': Automated Dataset Augmentation and Prejudice Quantificationcited by 0
  8. 2025The Dual-use Dilemma in LLMs: Do Empowering Ethical Capacities Make a Degraded Utility?cited by 0
  9. 2026From National Curricula to Cultural Awareness: Constructing Open-Ended Culture-Specific Question Answering Datasetcited by 0
  10. 2026Marginal Alignment Does Not Guarantee Joint-Distribution Fidelity: An Official-Reference Audit of Nemotron-Personas-Korea with Cross-Locale Replicationcited by 0
  11. 2026Off-Target Effects of Response-Style Alignment in a Korean 27B Language Modelcited by 0
  12. 2026CultureMINE: Datasets and Methods for Improving the Cultural Capabilities of NLP Systemscited by 0