Hacker News new | ask | show | jobs
by nathanyz 1994 days ago
Exactly and maybe we can "lobotomize" sections of the models that replicate unwanted bias in the training data.