 Hello everyone, welcome back to another chat GPT tutorial. In this one, I'm going to show you the two what I believe are the best ways to detect plagiarism. If you're an educator or a teacher and you want to see if somebody might be using some extra help, here's a way to do it. Here's two ways, in fact. And if you're a student, do your homework. You don't need to use this to cheat. Anyways, let's get on with this morality aside. Here we go. The first method is by hugging face. This is a GPT2 output detector demo. Please note that that says GPT2, and that matters. So while I hit the button here, you're going to see that it gives me a real percentage of 3.44 and a fake probability of 96.56. There's no absolutes in this. So busting somebody or getting busted is still, there's still some wiggle room in there. But it does a very good job. That said, it's using the GPT2 version. And right now, chat GPT is up to 3.5. So keep that in the back of your head. The second version or the second tool I like to use is GPT0. This is a more recent tool. And this one was created by Edward Tian in Princeton. He's a student at Princeton University. And yeah, he created this. And it's pretty damn wild. It's going to run here at the top. What does it do? It checks for perplexity. And it checks for burstiness. These are the two things that it looks for. Perplexity, as you can read here, is the randomness of the text. It's a measurement of how well a language model, like chat GPT, can predict. And what did I say? Yeah, predict a sample text. And yeah, and burstiness is basically it's just checking, well, anyways, you guys can read it, but it exhibits non-common items that appear in random clusters. That's the gist of it. That said, I clicked on chat GPT, get the result. It's going to take a few seconds, and I will come back when it's ready. All right, welcome back. So the result is GPT has finished analyzing. Your text is most likely to be AI generated, busted. All right, guys, again, there's no absolutes in this. And you can argue either way. But we did a full test on it. It gave us a score of 19.009, which doesn't mean a lot because who knows what that means on any X and Y graph. But note that it gives us a score right up to the end of this little line here. So it's about 96%. Same as the hugging face GPT2 detector. Thanks for watching.