Basically, if a cop can say “we asked AI to decide if we should shoot this person” and people are set up to think “if AI said it, it’s probably true,” then we’re in big trouble.