[2/19/2025] LLM that replace human participants can harmfully misportray and flatt

Large language models that replace human participants can harmfully misportray and flatten identity groups

https://www.nature.com/articles/s42256-025-00986-zwww.nature.com