Assessing GPT's Bias Towards Stigmatized Social Groups: An Intersectional Case Study on Nationality Prejudice and Psychophobia
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912389226561536 |
|---|---|
| author | Kashif, Afifah Patel, Heer |
| author_facet | Kashif, Afifah Patel, Heer |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_17045 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Assessing GPT's Bias Towards Stigmatized Social Groups: An Intersectional Case Study on Nationality Prejudice and Psychophobia Kashif, Afifah Patel, Heer Computation and Language 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. |
| title | Assessing GPT's Bias Towards Stigmatized Social Groups: An Intersectional Case Study on Nationality Prejudice and Psychophobia |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2505.17045 |