Even after anti-racism training AI chatbots like ChatGPT still exhibit racial prejudice

[Redaktion]

AI chatbots like ChatGPT-4 can still produce racially prejudiced responses even after safety training, researchers have found. The study highlights the need for greater care and vetting for "covert prejudice" before LLMs are made publicly available.

https://www.notebookcheck.net/Even-after-anti-racism-training-AI-chatbots-like-ChatGPT-still-exhibit-racial-prejudice.811940.0.html