mlsanityData to 14 Oct 2026

DICES Dataset: Diversity in Conversational AI Evaluation for Safety

arXiv 2306.11247 ↗First posted 20 Jun 2023cs.HCcited by 86

Not on the map. Why?

Abstract (arXiv)

Machine learning approaches often require training and evaluation datasets with a clear separation between positive and negative examples. This risks simplifying and even obscuring the inherent subjectivity present in many tasks. Preserving such variance in content and diversity in datasets is often expensive and laborious. This is especially troubling when building safety datasets for conversational AI systems, as safety is both socially and culturally situated. To demonstrate this crucial aspect of conversational AI safety, and to facilitate in-depth model performance analyses, we introduce the DICES (Diversity In Conversational AI Evaluation for Safety) dataset that contains fine-grained demographic information about raters, high replication of ratings per item to ensure statistical power for analyses, and encodes rater votes as distributions across different demographics to allow for in-depth explorations of different aggregation strategies. In short, the DICES dataset enables the observation and measurement of variance, ambiguity, and diversity in the context of conversational AI safety. We also illustrate how the dataset offers a basis for establishing metrics to show how raters' ratings can intersects with demographic categories such as racial/ethnic groups, age groups, and genders. The goal of DICES is to be used as a shared resource and benchmark that respects diverse perspectives during safety evaluation of conversational AI systems.

Key facts. Cited by 86 papers in the map (data through 14 Oct 2026). First posted on arXiv 20 Jun 2023.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 6; 2024: 23; 2025: 24; 2026: 33.

86 papers.

Sort:

By area:

  1. 2024LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Taskscited by 243
  2. 2023Sociotechnical Safety Evaluation of Generative AI Systemscited by 157
  3. 2024The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Modelscited by 144
  4. 2024Red-Teaming for Generative AI: Silver Bullet or Security Theater?cited by 83
  5. 2024SafetyPrompts: a Systematic Review of Open Datasets for Evaluating and Improving Large Language Model Safetycited by 77
  6. 2024Introducing v0.5 of the AI Safety Benchmark from MLCommonscited by 62
  7. 2025Values in the Wild: Discovering and Analyzing Values in Real-World Language Model Interactionscited by 52
  8. 2025SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviorscited by 52
  9. 2025The Homogenizing Effect of Large Language Models on Human Expression and Thoughtcited by 35
  10. 2025Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptionscited by 33
  11. 2024Beyond correlation: The Impact of Human Uncertainty in Measuring the Effectiveness of Automatic Evaluation and LLM-as-a-Judgecited by 31
  12. 2023Capturing Perspectives of Crowdsourced Annotators in Subjective Learning Taskscited by 29
38 more
  1. 2025Value Profiles for Encoding Human Variationcited by 29
  2. 2024Risks from Language Models for Automated Mental Healthcare: Ethics and Structure for Implementationcited by 27
  3. 2023How Far Can We Extract Diverse Perspectives from Large Language Models?cited by 22
  4. 2025Understanding Model Calibration -- A gentle introduction and visual exploration of calibration and the expected calibration error (ECE)cited by 20
  5. 2024Holistic Safety and Responsibility Evaluations of Advanced AI Modelscited by 19
  6. 2024Annotator-Centric Active Learning for Subjective NLP Taskscited by 18
  7. 2025Pairwise Calibrated Rewards for Pluralistic Alignmentcited by 14
  8. 2023Scaling Laws Do Not Scalecited by 13
  9. 2024Near to Mid-term Risks and Opportunities of Open-Source Generative AIcited by 13
  10. 2024Risks and Opportunities of Open-Source Generative AIcited by 13
  11. 2024STAR: SocioTechnical Approach to Red Teaming Language Modelscited by 13
  12. 2024D3CODE: Disentangling Disagreements in Data across Cultures on Offensiveness Detection and Evaluationcited by 11
  13. 2024Beyond the Binary: Capturing Diverse Preferences With Reward Regularizationcited by 11
  14. 2025Dehumanizing Machines: Mitigating Anthropomorphic Behaviors in Text Generation Systemscited by 11
  15. 2025Whose View of Safety? A Deep DIVE Dataset for Pluralistic Alignment of Text-to-Image Modelscited by 11
  16. 2025K2-Think: A Parameter-Efficient Reasoning Systemcited by 11
  17. 2024Detectors for Safe and Reliable LLMs: Implementations, Uses, and Limitationscited by 10
  18. 2024The Future of Open Human Feedbackcited by 10
  19. 2024SafetyAnalyst: Interpretable, Transparent, and Steerable Safety Moderation for AI Behaviorcited by 10
  20. 2025Modeling Annotator Disagreement with Demographic-Aware Experts and Synthetic Perspectivescited by 10
  21. 2025Operationalizing Pluralistic Values in Large Language Model Alignment Reveals Trade-offs in Safety, Inclusivity, and Model Behaviorcited by 9
  22. 2024Discipline and Label: A WEIRD Genealogy and Social Theory of Data Annotationcited by 8
  23. 2025Doubly-Robust LLM-as-a-Judge: Externally Valid Estimation with Imperfect Personascited by 8
  24. 2023The Empty Signifier Problem: Towards Clearer Paradigms for Operationalising "Alignment" in Large Language Modelscited by 7
  25. 2024A Standardized Machine-readable Dataset Documentation Format for Responsible AIcited by 7
  26. 2024Whose Preferences? Differences in Fairness Preferences and Their Impact on the Fairness of AI Utilizing Human Feedbackcited by 6
  27. 2026PluriHarms: Benchmarking the Full Spectrum of Human Judgments on AI Harmcited by 5
  28. 2023Thinking Assistants: LLM-Based Conversational Assistants that Help Users Think By Asking rather than Answeringcited by 4
  29. 2024Insights on Disagreement Patterns in Multimodal Safety Perception across Diverse Rater Groupscited by 4
  30. 2024GIEBench: Towards Holistic Evaluation of Group Identity-based Empathy for Large Language Modelscited by 3
  31. 2025From Noise to Signal to Selbstzweck: Reframing Human Label Variation in the Era of Post-training in NLPcited by 3
  32. 2026PERSPECTRA: A Scalable and Configurable Pluralist Benchmark of Perspectives from Argumentscited by 3
  33. 2024Libra-Leaderboard: Towards Responsible AI through a Balanced Leaderboard of Safety and Capabilitycited by 2
  34. 2025Meta-PerSER: Few-Shot Listener Personalized Speech Emotion Recognition via Meta-learningcited by 2
  35. 2025Argument-Based Consistency in Toxicity Explanations of LLMscited by 2
  36. 2025Lost in Localization: Building RabakBench with Human-in-the-Loop Validation to Measure Multilingual Safety Gapscited by 2
  37. 2026Expert Evaluation and the Limits of Human Feedback in Mental Health AI Safety Testingcited by 2
  38. 2026The Consensus Trap: Dissecting Subjectivity and the "Ground Truth" Illusion in Data Annotationcited by 2

50 most cited of 86 All 86 (JSON)