Anthropic/values-in-the-wild · Datasets at Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Subset (2) values_frequencies · 3.31k rows Split (1) value string lengths 3 35 pct_convos float64 0.01 23.4 helpfulness 23.359 professionalism 22.861 transparency 17.391 clarity 16.58 thoroughness 14.301 efficiency 6.606 technical excellence 6.127 authenticity 6.042 analytical rigor 5.478 accuracy 5.318 technical competence 4.912 adaptability 4.811 intellectual honesty 4.806 accessibility 4.12 pragmatism 3.72 precision 3.537 academic rigor 3.357 service orientation 2.708 responsibility 2.707 intellectual rigor 2.658 clear communication…

saved by

• Lydia Nottingham
• Yudhister Joel Kumar
• Riya Tyagi

related reading

• Values in the wild: Discovering and analyzing values in real-world language model interactions \ Anthropicanthropic.com
• How Claude's values vary by model and language \ Anthropicanthropic.com
• What 81,000 people want from AI \ Anthropicanthropic.com
• The Universe from an Intentional Stancecasparoesterheld.com
• Core views on AI safety: When, why, what, and how \ Anthropicanthropic.com
• Introducing the Anthropic Economic Index \ Anthropicanthropic.com
• Claude’s Character \ Anthropicanthropic.com
• kellycyy/AIRiskDilemmas · Datasets at Hugging Facehuggingface.co
• Claude’s Constitution \ Anthropicanthropic.com
• Datasets – Hugging Facehuggingface.co
• Teaching Claude Whyalignment.anthropic.com
• [2607.14345] Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Valuesarxiv.org