社会人口学提示尚不是用 LLM 模拟主观判断的有效方法
Sociodemographic Prompting is Not Yet an Effective Approach for Simulating Subjective Judgments with LLMs
AI总结：
该研究基于 POPQUORN 评估九个 LLM 的主观判断能力，发现其预测偏向白人群体，且社会人口学提示无法稳定改善多元对齐，甚至可能产生负面影响。
摘要：
人类判断本质上是主观的，并会受到性别、族裔等个人特征的主动影响。尽管大型语言模型（LLM）被广泛用于在多种情境中模拟人类反应，但它们在主观任务中解释人口统计学差异的能力仍不明确。本研究利用 POPQUORN 数据集，评估了九个流行 LLM 在礼貌性和冒犯性两项主观判断任务中理解人口统计学差异的能力。我们发现，在零样本设置下，大多数模型对两项任务的预测都更接近白人参与者的标签，而非亚裔或黑人参与者的标签；同时，仅在礼貌任务中出现轻微偏向女性的性别偏见。此外，社会人口学提示并不能持续改善 LLM 感知特定子群体语言的能力，在某些情况下甚至会使其恶化。这些发现揭示了 LLM 执行主观判断任务时潜在的人口统计学偏见，并强调了社会人口学提示作为实现多元对齐策略的局限性。代码和数据可在 https://github.com/Jiaxin-Pei/LLM-as-Subjective-Judge 获取。
英文摘要：
Human judgments are inherently subjective and are actively affected by personal traits such as gender and ethnicity. While Large Language Models (LLMs) are widely used to simulate human responses across diverse contexts, their ability to account for demographic differences in subjective tasks remains uncertain. In this study, leveraging the POPQUORN dataset, we evaluate nine popular LLMs on their ability to understand demographic differences in two subjective judgment tasks: politeness and offensiveness. We find that in zero-shot settings, most models' predictions for both tasks align more closely with labels from White participants than those from Asian or Black participants, while only a minor gender bias favoring women appears in the politeness task. Furthermore, sociodemographic prompting does not consistently improve and, in some cases, worsens LLMs' ability to perceive language from specific sub-populations. These findings highlight potential demographic biases in LLMs when performing subjective judgment tasks and underscore the limitations of sociodemographic prompting as a strategy to achieve pluralistic alignment. Code and data are available at: https://github.com/Jiaxin-Pei/LLM-as-Subjective-Judge.
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。
