Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Classification bias in GPT-4o mini, Mistral NeMo and Gemini 1.5 Flash [pdf] (diva-portal.org)
1 point by Bittermann on Nov 21, 2024 | hide | past | favorite | 1 comment


A University of Washington study found significant racial and gender biases in LLMs for resume screening, favoring white-associated names 85% of the time and female-associated names only 11%. In contrast, Black male-associated names were never preferred.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: