I think that's really the crux of it. We currently require people to express the full extent of their intentions. Not writing that you feel so sorry for your c

I think that's really the crux of it. We currently require people to express the full extent of their intentions. Not writing that you feel so sorry for your colleagues your are leaving in the dust is a faux pas. ChatGPT fills that gap of generating random acceptable feelings, where you might either not care, or do not want to express your actual ones. I'd argue adding people to thank or sneaky attacks to the ones you hate wouldn't add much to the prompt when requesting the letter to the bot. The joke of course will be that progressively, I'd expect we'll have people sending the bot a prompt, have it converted into verbiage, and the receiving end tl;dr it in the same bot to get basically the initial prompt.