arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2502.02696cs.CL
收藏

语言模型如何包容性地感知社会与道德规范?

How Inclusively do LMs Perceive Social and Moral Norms?

Georgia Institute of Technology(佐治亚理工学院)

Michael Galarnyk, Agam Shah, Dipanwita Guhathakurta, Poojitha Nandigam, Sudheer Chava

AI总结:

本研究调查语言模型(LMs)跨人口群体感知社会道德规范的包容性,通过向11个LMs提示经验法则并与100名人类标注者比较,引入ADA-Met量化对齐度,发现其对年轻高收入群体对齐度更高,需提升多样性包容。

摘要:

本文包含并讨论冒犯性内容。语言模型(LMs)被用于决策系统和交互式助手。然而,这些模型的判断与人类价值观的多样性(尤其是社会和道德规范方面)的契合度如何?本研究调查了LMs如何跨人口统计学群体(如性别、年龄和收入)包容性地感知规范。我们向11个LMs提示经验法则(RoTs),并将其输出与100名人类标注者的现有回应进行比较。我们引入绝对距离对齐指标(ADA-Met)来量化有序问题的对齐度。研究发现LMs回应存在显著差异,年轻、高收入群体的对齐度更高,这引发了对边缘化视角代表性的担忧。我们的研究结果强调了进一步努力使LMs更包容多样化人类价值观的重要性。代码和提示可在GitHub上获取,遵循CC BY-NC 4.0许可。

英文摘要:

This paper discusses and contains offensive content. Language models (LMs) are used in decision-making systems and as interactive assistants. However, how well do these models making judgements align with the diversity of human values, particularly regarding social and moral norms? In this work, we investigate how inclusively LMs perceive norms across demographic groups (e.g., gender, age, and income). We prompt 11 LMs on rules-of-thumb (RoTs) and compare their outputs with the existing responses of 100 human annotators. We introduce the Absolute Distance Alignment Metric (ADA-Met) to quantify alignment on ordinal questions. We find notable disparities in LM responses, with younger, higher-income groups showing closer alignment, raising concerns about the representation of marginalized perspectives. Our findings highlight the importance of further efforts to make LMs more inclusive of diverse human values. The code and prompts are available on GitHub under the CC BY-NC 4.0 license.

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。

↑