PLURAL: A Global Dataset for Value Alignment

arXiv cs.CL Papers

Summary

Introduces PLURAL, a large-scale preference dataset grounded in the Integrated Values Survey across 92 countries, with ~500,000 preference triplets from 20 diverse countries, aimed at improving cultural value alignment in LLMs.

arXiv:2607.08034v1 Announce Type: new Abstract: Large language models (LLMs) are used worldwide, yet disproportionately reflect Western values, limiting their ability to represent diverse value systems. We introduce PLURAL, a large-scale, value-focused preference dataset grounded in the Integrated Values Survey (IVS), a nationally representative survey spanning 92 countries. Using a two-stage generation pipeline, we transform survey responses into synthetic preference triplets that preserve normative value signals while producing realistic scenarios. We release an initial version of PLURAL containing ~500,000 preference triplets representing people in 20 diverse countries. We evaluate PLURAL in three ways: (i) dataset-level validation showing that it preserves both cross-country value differences and within-country diversity from the original survey; (ii) automated evaluation showing that training on PLURAL improves alignment with target countries' cultural profiles, reducing mean absolute error by up to 27.7% relative to strong baselines; and (iii) blind human evaluation with 176 evaluators in India, Brazil, and Japan, who judge PLURAL-aligned responses as more representative of their national values. Together, these results show that PLURAL contains learnable signal for value steering, offering a scalable resource for pluralistic alignment. Dataset: https://huggingface.co/datasets/agdhruv/plural-alignment
Original Article
View Cached Full Text

Cached at: 07/10/26, 06:12 AM

# PLURAL: A Global Dataset for Value Alignment
Source: [https://arxiv.org/abs/2607.08034](https://arxiv.org/abs/2607.08034)
[View PDF](https://arxiv.org/pdf/2607.08034)

> Abstract:Large language models \(LLMs\) are used worldwide, yet disproportionately reflect Western values, limiting their ability to represent diverse value systems\. We introduce PLURAL, a large\-scale, value\-focused preference dataset grounded in the Integrated Values Survey \(IVS\), a nationally representative survey spanning 92 countries\. Using a two\-stage generation pipeline, we transform survey responses into synthetic preference triplets that preserve normative value signals while producing realistic scenarios\. We release an initial version of PLURAL containing ~500,000 preference triplets representing people in 20 diverse countries\. We evaluate PLURAL in three ways: \(i\) dataset\-level validation showing that it preserves both cross\-country value differences and within\-country diversity from the original survey; \(ii\) automated evaluation showing that training on PLURAL improves alignment with target countries' cultural profiles, reducing mean absolute error by up to 27\.7% relative to strong baselines; and \(iii\) blind human evaluation with 176 evaluators in India, Brazil, and Japan, who judge PLURAL\-aligned responses as more representative of their national values\. Together, these results show that PLURAL contains learnable signal for value steering, offering a scalable resource for pluralistic alignment\. Dataset:[this https URL](https://huggingface.co/datasets/agdhruv/plural-alignment)

## Submission history

From: Dhruv Agarwal \[[view email](https://arxiv.org/show-email/e34725fc/2607.08034)\] **\[v1\]**Thu, 9 Jul 2026 01:18:17 UTC \(2,273 KB\)

Similar Articles

Personalization, Personas, and Forecasting in Value Alignment

arXiv cs.AI

This paper tests whether different prompt framings (personalization, role-play, third-person forecasting) are interchangeable in eliciting cultural values from LLMs, using the World Values Survey. Results show that prompt framing significantly affects model responses and measured cultural alignment, with third-person forecasting yielding the strongest directional alignment.

Rushes: A Human Preference Dataset for Pluralistic Alignment

arXiv cs.CL

Introduces Rushes, a large-scale dataset of human engagement preferences in AI-generated branching narratives, revealing that current LLMs like GPT-5 fail to outperform simple baselines in predicting user choices, highlighting the need for personalized alignment.

Evaluating Pluralism in LLMs through Latent Perspectives

arXiv cs.CL

This paper introduces a domain-agnostic multi-layered framework for unsupervised extraction of perspectives to evaluate pluralism in LLM-generated text, finding that rare perspectives are disproportionately underrepresented.