From Surveys to Narratives: Rethinking Cultural Value Adaptation in LLMs

2025-05-22

AI Summary

Large language models (LLMs) face significant challenges in cultural value alignment, including entrenched biases, homogeneous cultural representations, and degradation of factual knowledge. Prior approaches rely predominantly on the World Values Survey (WVS), which risks cultural flattening and fails to support fine-grained, task-specific cross-cultural understanding. This work is the first to systematically expose the representational limitations of WVS as a sole cultural source. We propose a “narrative-augmented” paradigm that integrates encyclopedic knowledge from Wikipedia with situated cultural narratives from NormAd, enabling a multi-source alignment training framework. Experiments demonstrate substantial improvements in cross-cultural question answering and value judgment tasks: cultural discriminability increases by 37% (p < 0.01), while factual consistency is preserved. Our approach achieves synergistic optimization of cultural distinctiveness and task adaptability.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Adapting cultural values in Large Language Models (LLMs) presents significant challenges, particularly due to biases and limited training data. Prior work primarily aligns LLMs with different cultural values using World Values Survey (WVS) data. However, it remains unclear whether this approach effectively captures cultural nuances or produces distinct cultural representations for various downstream tasks. In this paper, we systematically investigate WVS-based training for cultural value adaptation and find that relying solely on survey data can homogenize cultural norms and interfere with factual knowledge. To investigate these issues, we augment WVS with encyclopedic and scenario-based cultural narratives from Wikipedia and NormAd. While these narratives may have variable effects on downstream tasks, they consistently improve cultural distinctiveness than survey data alone. Our work highlights the inherent complexity of aligning cultural values with the goal of guiding task-specific behavior.

Problem

• Challenges in adapting cultural values in LLMs due to biases and limited data
• Inadequacy of survey data for capturing cultural nuances in LLMs
• Need for enriched cultural narratives to improve cultural distinctiveness in LLMs

Innovation

• Augment WVS with encyclopedic cultural narratives
• Use scenario-based narratives from Wikipedia and NormAd
• Improve cultural distinctiveness beyond survey data