▲ 609 ▼ ethinically ambigaus (feddit.de) submitted 2 years ago by sexy_peach@feddit.de to c/technology@lemmy.ml 64 comments fedilink hide all child comments
[–] lars@programming.dev 29 points 2 years ago Yes. The training data has a bias, and they are using a cheap hack (prompt manipulation) to try to patch it. permalink fedilink source parent