I was just thinking about this earlier today. The fact that, setting aside all of the moral issues, these AI models just... don't do the things that they are supposed to do. Like we're a couple years into this and that's just pretty obvious yet people are still using them and that baffles me.
