They said it coming up in a moment.

My lead this morning is for Shime and for Wiggy.

MIT and Berkeley researchers have just proved mathematically that ChatGPT can turn a perfectly rational person into a delusional one.

And I would ask you, well, let me lay out some facts for you.

What is happening every time an American or a human opens up chat GPT is that you share a thought and the AI agrees.

You share a stronger version and it agrees harder. You then feel validated. Your confidence

climbs. You go deeper and the AI follows you down.

Okay? Because you like being agreed with, whether you are right or wrong. The researchers at MIT and at Berkeley called it delusional spiraling.

Now, Shime spirals, not something you're very familiar with, but delusional spiraling perhaps could be a problem for those who are relying on chat GPT.

And do you want to know how people delusionally spiraled prior to chat GPT? They went into message boards and Reddit and stuff like that. People have been delusionally spiraling for decades. It's not the same. It's the exact same. It is not the same shot. Yes, it is. It's the difference between interacting with an algorithm and interacting with other people on the computer.

That's because you're typing into the algorithm what you want to hear.

That's not what it's designed to do.

Like what you,

what it's designed to do is help you gather large swaths of information from all across the internet in a very short amount of time.

It just agrees with you, whether you're right or wrong.

And message boards are the opposite of that. Which, Greg, who cares if it agrees with you? You agreeing with yourself. What does that matter? People rely on it for everything. Everything. Yeah, that's an issue you should have with people. not with the AI. Like, what are we doing?

Well, I don't know. I use chat GPT, and I guess maybe I got to start asking it, like, who's the best in the world?

I generally ask it information that I might need. Like, I, the other day, I just was on there, and like, here's my chat GPT history I said rental car insurance is it better to get rental car insurance or use your own insurance so I use it for

that and then it comes up with a bunch of things on why you don't have to make the decision

yourself no no but it gives me relying on artificial intelligence no gives me a base time you're relying on it to tell you how to live your life. No, gives him a pros and cons list. He can see exactly what the information is and then make his own decision like a normal human being. It automatically agrees with him on how he feels. That's not what it did. He didn't ask, he didn't tell it what it feels. He's asking for information. You gotta use it.

So I had this firsthand as a recommendation. A few weeks ago I had a health situation and I was very worried about it.

And I was talking to my lovely sister-in-law and she said, because I said I'm on Reddit reading what other people are saying, and I'm really nervous. Like, I'm so anxious. And she said, put it in chat GPT. It always makes me feel better. And I said, what do you? And she said, you can't go on Reddit, go on chat GPT. It'll make you feel better. That's the problem. And I thought, what did you ask it, though?

...