I agree– if I try pure ethical problems like The Trolley Question, ChatGPT will just list all the philosophies it can come up with and say they're all valid, but ask it anything with a sociopolitical tinge? I interrogated the damn bot once on how it "knew" what it did, and it just repeated what it had already said in a mindless loop that would have gone on until I crashed my computer. Though, it would be an interesting experiment to see what the effects of training it on different ideologies would be. I feel like Communist ChatGPT would have an actual opinion on The Trolley Question, though I'm sure I wouldn't like it.