A new study of popular AI models shows that their feedback on social situations is far from impartial.
A new study of popular AI models shows that their feedback on social situations is far from impartial.