Whose Opinions Do Language Models Reflect?
Whose Opinions Do Language Models Reflect?
复制标题
DOI：
10.48550/arxiv.2303.17548
复制
发表时间：
2023-03
期刊：
影响因子：
--
通讯作者：
Shibani Santurkar;Esin Durmus;Faisal Ladhak;Cinoo Lee;Percy Liang;Tatsunori Hashimoto
中科院分区：
文献类型：
--
作者：
Shibani Santurkar;Esin Durmus;Faisal Ladhak;Cinoo Lee;Percy Liang;Tatsunori Hashimoto
语言模型（LM）越来越多地被用于开放式环境中，其中LM响应主观查询所反映的意见可以对用户满意度以及塑造整个社会的观点产生深远的影响。在这项工作中，我们提出了一个定量框架来调查LM所反映的意见-通过利用高质量的民意调查及其相关的人类反应。使用这个框架，我们创建了OpinionsQA，这是一个新的数据集，用于评估LM的意见与60个美国人口群体的意见在从堕胎到自动化等主题上的一致性。在各个主题中，我们发现当前LM所反映的观点与美国人口群体的观点之间存在实质性的不一致：与民主党-共和党在气候变化问题上的分歧不相上下。值得注意的是，即使在明确地将LM导向特定人口群体之后，这种错位仍然存在。我们的分析不仅证实了先前关于一些人类反馈调整的LM的左倾倾向的观察，而且还揭示了当前LM反映不佳的群体（例如，65+和丧偶的人）。我们的代码和数据可在https://github.com/tatsu-lab/opinions_qa上获得。
Language models (LMs) are increasingly being used in open-ended contexts, where the opinions reflected by LMs in response to subjective queries can have a profound impact, both on user satisfaction, as well as shaping the views of society at large. In this work, we put forth a quantitative framework to investigate the opinions reflected by LMs -- by leveraging high-quality public opinion polls and their associated human responses. Using this framework, we create OpinionsQA, a new dataset for evaluating the alignment of LM opinions with those of 60 US demographic groups over topics ranging from abortion to automation. Across topics, we find substantial misalignment between the views reflected by current LMs and those of US demographic groups: on par with the Democrat-Republican divide on climate change. Notably, this misalignment persists even after explicitly steering the LMs towards particular demographic groups. Our analysis not only confirms prior observations about the left-leaning tendencies of some human feedback-tuned LMs, but also surfaces groups whose opinions are poorly reflected by current LMs (e.g., 65+ and widowed individuals). Our code and data are available at https://github.com/tatsu-lab/opinions_qa.