mlsanityData to 7 Oct 2026

The Homogenization Problem in LLMs: Towards Meaningful Diversity in AI Safety

arXiv 2601.06116 ↗First posted 3 Jan 2026cs.AI cs.CL cs.CYcited by 1

Not on the map. Why?

Abstract (arXiv)

Generative AI models reproduce the human biases in their training data and further amplify them through mechanisms such as mode collapse. The loss of diversity produces homogenization, which not only harms the minoritized but impoverishes everyone. We argue homogenization should be a central concern in AI safety. To meaningfully characterize homogenization in Large Language Models (LLMs), we introduce a framework that allows stakeholders to encode their context and value system. We illustrate our approach with an experiment that surfaces gender bias in an LLM (Claude 3.5 Haiku) on an open-ended story prompt. Building from queer theory, we formalize homogenization in terms of normativity. Borrowing language from feminist theory, we introduce the concept of xeno-reproduction as a class of tasks for mitigating homogenization by promoting diversity. Our work opens a collaborative line of research that seeks to understand and advance diversity in AI.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 95 of its 244 references that are in the map. First posted on arXiv 3 Jan 2026.

Builds on

95 of 244 references.

References · 95

  1. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
  2. 2015A Diversity-Promoting Objective Function for Neural Conversation Modelscited by 1,068
  3. 2018Texygen: A Benchmarking Platform for Text Generation Modelscited by 484
  4. 2020Artificial Intelligence, Values and Alignmentcited by 443
  5. 2023JudgeLM: Fine-tuned Large Language Models are Scalable Judgescited by 351
  6. 2022The Vendi Score: A Diversity Evaluation Metric for Machine Learningcited by 299
  7. 2025Why Language Models Hallucinatecited by 205
  8. 2019Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practicescited by 170
  9. 2023Cultural Bias and Cultural Alignment of Large Language Modelscited by 149
  10. 2022Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reductioncited by 137
  11. 2025Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond)cited by 137
  12. 2020Evaluating the Evaluation of Diversity in Natural Language Generationcited by 127
38 more
  1. 2024MaxMin-RLHF: Alignment with Diverse Human Preferencescited by 121
  2. 2024A Note on Shumailov et al. (2024): `AI Models Collapse When Trained on Recursively Generated Data'cited by 114
  3. 2020Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligencecited by 100
  4. 2023The Curious Decline of Linguistic Diversity: Training Language Models on Synthetic Textcited by 97
  5. 2020Trading Off Diversity and Quality in Natural Language Generationcited by 94
  6. 2022Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?cited by 89
  7. 2024Understanding Hallucinations in Diffusion Models through Mode Interpolationcited by 87
  8. 2020Why Fairness Cannot Be Automated: Bridging the Gap Between EU Non-Discrimination Law and AIcited by 81
  9. 2024Open-Endedness is Essential for Artificial Superhuman Intelligencecited by 79
  10. 2025Outcome-based Exploration for LLM Reasoningcited by 79
  11. 2024Bias in Large Language Models: Origin, Evaluation, and Mitigationcited by 77
  12. 2018Diversity-Driven Exploration Strategy for Deep Reinforcement Learningcited by 65
  13. 2025NoveltyBench: Evaluating Language Models for Humanlike Diversitycited by 64
  14. 2022"Curse of rarity" for autonomous vehiclescited by 62
  15. 2024One fish, two fish, but not the whole sea: Alignment reduces language models' conceptual diversitycited by 57
  16. 2025Evaluating the Diversity and Quality of LLM Generated Contentcited by 51
  17. 2024Forcing Diffuse Distributions out of Language Modelscited by 48
  18. 2025Base Models Beat Aligned Models at Randomness and Creativitycited by 48
  19. 2024Detection and Measurement of Syntactic Templates in Generated Textcited by 44
  20. 2024Confabulation: The Surprising Value of Large Language Model Hallucinationscited by 42
  21. 2025Gradual Disempowerment: Systemic Existential Risks from Incremental AI Developmentcited by 42
  22. 2024Benchmarking Linguistic Diversity of Large Language Modelscited by 37
  23. 2023The Unfairness of Fair Machine Learning: Levelling down and strict egalitarianism by defaultcited by 36
  24. 2025The Homogenizing Effect of Large Language Models on Human Expression and Thoughtcited by 35
  25. 2024AI and the Problem of Knowledge Collapsecited by 34
  26. 2025Position: Model Collapse Does Not Mean What You Thinkcited by 32
  27. 2023On the Importance of Exploration for Generalization in Reinforcement Learningcited by 30
  28. 2023Closing the Curious Case of Neural Text Degenerationcited by 30
  29. 2024Generation through the lens of learning theorycited by 30
  30. 2024Forking Paths in Neural Text Generationcited by 30
  31. 2025The Shrinking Landscape of Linguistic Diversity in the Age of Large Language Modelscited by 28
  32. 2024Exploring Facets of Language Generation in the Limitcited by 27
  33. 2025LLM Probability Concentration: How Alignment Shrinks the Generative Horizoncited by 26
  34. 2021Measuring Model Biases in the Absence of Ground Truthcited by 24
  35. 2023Taxonomizing and Measuring Representational Harms: A Look at Image Taggingcited by 24
  36. 2024Detecting Mode Collapse in Language Models via Narrationcited by 24
  37. 2024Learned feature representations are biased by complexity, learning order, position, and morecited by 22
  38. 2019The Diversity-Innovation Paradox in Sciencecited by 21

50 most cited of 95 All 95 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Magic, Madness, Heaven, Sin: LLM Output Diversity is Everything, Everywhere, All at Oncecited by 0