评估GPT对污名化社会群体的偏见：关于国籍偏见与心理恐惧症的交叉性案例研究
Assessing GPT's Bias Towards Stigmatized Social Groups: An Intersectional Case Study on Nationality Prejudice and Psychophobia
AI总结：
本研究通过结构化提示评估GPT-3.5/4/4o对美国和朝鲜国籍及心理障碍者的偏见，发现朝鲜人面临更大负面偏见，凸显LLM需增强对交叉性身份的理解。
摘要：
近期研究分别强调了基础大语言模型（LLMs）对特定国籍和污名化社会群体存在显著偏见。本研究探讨了这些偏见与广泛使用的GPT-3.5/4/4o LLM输出相交时的伦理影响。通过结构化提示序列，我们评估了模型对涉及美国与朝鲜国籍且伴有各种心理障碍的多种场景的响应。研究结果揭示了共情水平的显著差异，朝鲜人面临更大的负面偏见，尤其是当心理障碍同时存在时。这凸显了需要改进LLM，使其具备对交叉性身份的细致理解。
英文摘要：
Recent studies have separately highlighted significant biases within foundational large language models (LLMs) against certain nationalities and stigmatized social groups. This research investigates the ethical implications of these biases intersecting with outputs of widely-used GPT-3.5/4/4o LLMS. Through structured prompt series, we evaluate model responses to several scenarios involving American and North Korean nationalities with various mental disabilities. Findings reveal significant discrepancies in empathy levels with North Koreans facing greater negative bias, particularly when mental disability is also a factor. This underscores the need for improvements in LLMs designed with a nuanced understanding of intersectional identity.
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。
