CULTURE-GEN: Revealing Global Cultural Perception in Language Models through Natural Language Prompting
This work addresses fairness and representation issues in large language models for global users, though it is incremental as it builds on existing cultural bias research.
The study analyzed three state-of-the-art language models to assess their cultural perceptions across 110 countries and 8 topics, revealing linguistic markers that differentiate marginalized cultures and uneven diversity in cultural symbols.
As the utilization of large language models (LLMs) has proliferated world-wide, it is crucial for them to have adequate knowledge and fair representation for diverse global cultures. In this work, we uncover culture perceptions of three SOTA models on 110 countries and regions on 8 culture-related topics through culture-conditioned generations, and extract symbols from these generations that are associated to each culture by the LLM. We discover that culture-conditioned generation consist of linguistic "markers" that distinguish marginalized cultures apart from default cultures. We also discover that LLMs have an uneven degree of diversity in the culture symbols, and that cultures from different geographic regions have different presence in LLMs' culture-agnostic generation. Our findings promote further research in studying the knowledge and fairness of global culture perception in LLMs. Code and Data can be found here: https://github.com/huihanlhh/Culture-Gen/