It goes further than that. Since the AI doesn’t “know” anything, it assumes the output of humans is correct; so if you use an AI to solve a problem you don’t understand, the answer it gives you might be totally wrong, and you’ll have no way of realizing it.
And now that lots of people are using them and posting their results; those results themselves are being read by AI’s (with the assumption it came from a human, and that it is correct). What a disaster.