CULTURE-GEN：通过自然语言提示揭示语言模型中的全球文化感知
CULTURE-GEN: Revealing Global Cultural Perception in Language Models through Natural Language Prompting
AI总结：
本研究通过文化条件生成揭示三个SOTA模型对110个国家和地区在8个文化主题上的感知，发现文化符号多样性不均及地理差异，促进全球文化感知的公平性研究。
摘要：
随着大型语言模型（LLMs）在全球范围内的使用激增，它们对多样化的全球文化拥有充分的知识和公平的呈现变得至关重要。在这项工作中，我们通过文化条件生成，揭示了三个最先进的（SOTA）模型在110个国家和地区、8个文化相关主题上的文化感知，并从这些生成中提取了与每种文化相关联的符号。我们发现，文化条件生成包含语言“标记”，这些标记将边缘化文化与默认文化区分开来。我们还发现，LLMs在文化符号方面具有不均匀的多样性程度，并且来自不同地理区域的文化在LLMs的文化无关生成中具有不同的存在感。我们的发现促进了进一步研究LLMs中全球文化感知的知识和公平性。代码和数据可在以下网址找到：https://github.com/huihanlhh/Culture-Gen/
英文摘要：
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/
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
