> can you prove to me in a verifiable way that no matter what prompt I put into ChatGPT, it won't give me pornography back?
No, but if it's turning you away even when you're explicitly asking for it, it's probably doing good enough. Nobody held Yahoo, Lycos or Altavista to this standard.
If accidental erotica is the worst outcome you can imagine for the shortcomings of AI teaching, please leave worrying about this to the professionals. Consider flawed chemistry lessons, where it tells some kid to mix two things they shouldn't. That will actually cause material harm to everyone around them.
It is the teacher's responsibility to evaluate any materials they present to students. If they are given an output they interpret to be pornographic, they decide whether to provide it or not to students. I imagine it is possible that you might determine something to be pornographic that a given teacher may not. Pornography is an interpretation, which varies culturally and politically. Regardless, it is definitely not my responsibility to prove what ChatGPT will provide whatsoever, I don't work for OpenAI.
