arXiv cs.CY· Bryan Chen Zhengyu Tan, Zhengyuan Liu, Xiaoyuan Yi, Jing Yao, Xing Xie, Nancy F. Chen, Roy Ka-Wei Lee·· 12 小时前AI 评分34
Persona-Prompted LLM 能否模拟子群体价值观？文化对齐的泛化性与公平性实证分析
Can Persona-Prompted LLMs Emulate Subgroup Values? An Empirical Analysis of Generalisability and Fairness in Cultural Alignment
AI 导读
研究以新加坡为案例、基于 World Values Survey 数据检验 LLM 能否模拟人口子群体的文化价值观，发现 GPT-4.1 预测子群体偏好众数的准确率仅 57.4%。
正文 · 原文
Abstract:Despite their global prevalence, many Large Language Models (LLMs) are aligned to a monolithic, often Western-centric set of values. This paper investigates the more challenging task of fine-grained value alignment: examining whether LLMs can emulate the distinct cultural values of demographic subgroups. Using Singapore as a case study and the World Values Survey (WVS), we examine the value landscape and show that even state-of-the-art models like GPT-4.1 achieve only 57.4% accuracy in predicting subgroup modal preferences. We construct a dataset of over 20,000 samples to train and evaluate a range of models. We demonstrate that simple fine-tuning on structured numerical preferences yields substantial gains, improving accuracy on unseen, out-of-distribution subgroups by an average of 17.4%. These gains partially transfer to open-ended generation. However, we find significant pre-existing performance biases, where models better emulate young, male, Chinese, and Christian personas. Furthermore, while fine-tuning improves average performance, it widens the disparity between subgroups when measured by distance-aware metrics. Our work offers insights into the limits and fairness implications of subgroup-level cultural alignment.
| Comments: | ACL 2026. Code: this https URL |
| Subjects: | Computers and Society (cs.CY) |
| Cite as: | arXiv:2604.12851 [cs.CY] |
| (or arXiv:2604.12851v2 [cs.CY] for this version) | |
| https://doi.org/10.48550/arXiv.2604.12851 arXiv-issued DOI via DataCite |
Submission history
From: Bryan Tan [view email]
[v1]
Tue, 14 Apr 2026 15:06:13 UTC (330 KB)
[v2]
Thu, 15 Oct 2026 23:24:37 UTC (330 KB)
来源：arXiv cs.CY · arxiv.org