I see a lot of ways that people get in trouble with AI. In fact, here's an example. I was at someone's house and they pulled out their phone, brought up ChatGPT. I was trying to do a complex calculation of different, it was just different fractions of, of ratios of mixing some chemicals together.
And he went ahead and posed a question to ChatGPT. He asked the question wrong and ChatGPT gave him an answer. And what he did is he turned around and said, "Dave, the answer is this." Which is completely wrong. I knew it was wrong viscerally. The math did not make sense. It was wrong, it was impossible even.
And yet he was absolutely sure it was right because he asked ChatGPT and ChatGPT can't lie. The biggest problem is that when we don't know how to ask the prompt, right? The answer we get back oftentimes will be completely fallacious, wrong. You've got to be smart in how you ask your questions, and you have to be able to have a sniffer that can check and go, whoa, that's right, or that's not, you at least have to be able to check it.