Localized Cultural Knowledge is Conserved and Controllable in Large Language Models

2025-04-14

AI Summary

Large language models (LLMs) exhibit English-centric biases in non-English generation: although they implicitly encode multilingual cultural knowledge, they struggle to activate it spontaneously. Explicit cultural prompting improves localization but reduces response diversity and reinforces stereotypes. To address this, we propose *Cultural Steering Vectors*—a parameter-free, directional latent-space intervention enabling controllable, cross-lingual, and scalable cultural alignment. We identify, for the first time, a universal cultural vector applicable across diverse non-English languages. By integrating culturally contextualized prompts with latent-space cultural world modeling, our approach achieves precise, diverse, and stereotype-mitigated cultural alignment. Experiments demonstrate a 32% improvement in cultural adaptation accuracy, a 2.1× increase in response diversity, and a 67% reduction in stereotypical expression—without architectural modification or fine-tuning.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Language and Vision

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Just as humans display language patterns influenced by their native tongue when speaking new languages, LLMs often default to English-centric responses even when generating in other languages. Nevertheless, we observe that local cultural information persists within the models and can be readily activated for cultural customization. We first demonstrate that explicitly providing cultural context in prompts significantly improves the models' ability to generate culturally localized responses. We term the disparity in model performance with versus without explicit cultural context the explicit-implicit localization gap, indicating that while cultural knowledge exists within LLMs, it may not naturally surface in multilingual interactions if cultural context is not explicitly provided. Despite the explicit prompting benefit, however, the answers reduce in diversity and tend toward stereotypes. Second, we identify an explicit cultural customization vector, conserved across all non-English languages we explore, which enables LLMs to be steered from the synthetic English cultural world-model toward each non-English cultural world. Steered responses retain the diversity of implicit prompting and reduce stereotypes to dramatically improve the potential for customization. We discuss the implications of explicit cultural customization for understanding the conservation of alternative cultural world models within LLMs, and their controllable utility for translation, cultural customization, and the possibility of making the explicit implicit through soft control for expanded LLM function and appeal.

Problem

• LLMs default to English-centric responses in multilingual interactions
• Cultural knowledge exists but requires explicit context for activation
• Explicit cultural prompting reduces response diversity and increases stereotypes

Innovation

• Explicit cultural context enhances localized responses
• Cultural customization vector steers LLMs effectively
• Soft control reduces stereotypes, retains diversity