人口统计多元主义：多元人类偏好分布的推理时建模
Demographic Pluralism: Inference-Time Modeling of Pluralistic Human Preference Distributions
- Amazon Science（亚马逊科学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
提出人口统计多元主义推理时框架，通过生成群体内多视角估计观点分布，在四个骨干模型上比模块化多元主义降低Jensen-Shannon距离8.4%-26.4%，并发现等权聚合最优。
AI中文摘要：
大型语言模型（LLM）越来越多地用于文化敏感场景，在这些场景中，对齐需要代表群体内的多样化偏好。然而，现有方法在粗粒度的人口统计或社区层面建模群体，忽视了群体内部的差异。我们提出人口统计多元主义（Demographic Pluralism），一种推理时框架，通过生成人口统计基础群体内的多个视角，在无需观点分布训练数据或任务特定微调的情况下估计群体层面的观点分布。在GlobalOpinionQA和VITAL上的四个骨干模型上，与模块化多元主义（Modular Pluralism）相比，它将Jensen-Shannon距离降低了8.4%-26.4%。在加权、等权和逆加权聚合中，等权聚合总体表现最佳；群体层面误差也随群体权重增加而增大，这有助于解释加权聚合表现较弱的原因。
英文摘要：
Large language models (LLMs) are increasingly used in culturally sensitive settings, where alignment requires representing diverse preferences within populations. Yet existing methods model populations at coarse demographic or community levels and overlook within-group variation. We introduce Demographic Pluralism, an inference-time framework that estimates population-level opinion distributions without opinion-distribution training data or task-specific fine-tuning by generating multiple perspectives within demographically grounded groups. Across four backbones on GlobalOpinionQA and VITAL, it reduces Jensen-Shannon distance by 8.4%-26.4% over Modular Pluralism. Among weighted, equal-weighted, and inverse-weighted aggregation, equal weighting performs best overall; group-level error also increases with group weight, helping explain weighted aggregation's weaker performance.
