Moral stereotyping in large language models

• 成果类型: Article
• 署名作者: Zewail, Aliah; Figueroa, Alexandra; Graham, Jesse; Atari, Mohammad
• 署名单位: University of Massachusetts System; University of Massachusetts Amherst; University of California System; University of California Berkeley; Utah System of Higher Education; University of Utah
• 刊物名称: PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
• ISSN/ISSBN: 0027-8424; 1091-6490
• DOI: 10.1073/pnas.2519941123
• 发表日期: 2026-03-10
• 页码: e2519941123
• 关键词: morality; Large language models; culture; ai; FOUNDATIONS; competence; PSYCHOLOGY; accuracy; Covid-19; motives; VALUES

摘要

Can Large Language Models (LLMs) accurately estimate various societies' moral values? Here, we query the perceptions of LLMs regarding the moral norms of the average person from 48 nations and compare them to a large-scale (n = 90,802) survey of six moral values (Care, Equality, Proportionality, Loyalty, Authority, and Purity) from those populations. Our findings indicate that LLMs poorly capture the moral diversity around the globe, systematically overestimating some moral values (particularly Care) and underestimating others (especially Purity). Notably, examining various versions of Generative Pre-trained Transformer (GPT) shows that these LLMs may overestimate the overall moral concerns of some Western countries (e.g., the United States, Canada, and Australia) while underestimating those of non-Western countries (e.g., Nigeria, Morocco, and Indonesia). Our work demonstrates that LLMs are inaccurate generators of cross-cultural estimations in the moral domain; in other words, they stereotype the moral values of non-Western populations in predictable ways. Our results highlight the ethical and epistemic risks of relying on LLMs to estimate the endorsement of moral values around the globe.