ple's writing style is just like a fingerprint.

Yeah, Joe Klein was the author of "Primary Colors." Thank you.

Is that, by the way, is anybody blown away by that? I'm looking at your comments. I was expecting some surprise but, you know, we could find him guaranteed if we wanted to. But there's either – so there's a couple of things might be happening here. I can't believe nobody thought of this. Would you agree? Would you agree that it's sort of impossible that nobody thought of finding him by his writings, his writing style? So why haven't we done it, right? It's kind of weird. Why haven't we done it? I don't know.

So I'm going to probably talk a lot about this new AI system that's available to the public, ChatGPT, on social media. It's all over the place. People are testing it for various things to see what it can and cannot do.

One of the things it does really well is write code in different languages. So you could tell it, "How do I solve this particular programming problem?" And it tells you really, really fast and correctly.

Doesn't that mean we're very near the point where I can simply describe an app and it would make it for me? Not yet, but we're right there, right? We're like right on the border because here's what I want to do. I want to sit down at my screen and say, "All right, I want to start an app and the app is going to do XYZ and I want to use a modern interface and I want to make sure that if you choose this you get these features and it does this and that." I think you could just make it. I think you could make your app while you're sitting there in like five minutes. No, maybe one minute, right?

So you're saying that anybody who says that this won't be possible, you're totally wrong. This will be completely possible.

Now I should be able to look at the app and say, "All right, I like where you put the send button but can you move it down to the right and make it Auburn colored?" And it would just, while you're looking at it, would just move it down and make it Auburn colored, right?

So I should be able to move the interface around. I should be able to add a page. I should be able to say, just think about this, I should be able to say "add some boilerplate terms of service" and they would just appear and they would be perfect, you know, lawyer perfect.

I could say "trademark this phrase" and AI would say that phrase is already trademarked but I would suggest the following instead. And I say, "Oh, okay, go trademark that." And then the documents appear. Maybe I have to sign them but that's it.

Copyright this. Boom. So copyright lawyering is going to probably disappear by 50 percent. My guess is that 50% of the lawyer profession will be wiped out because all the contractual stuff will just be AI.

You should be able to make a full legal contract by saying, "All right, make me a contract. Let's see, it's going to be a lease. It's for my apartment." And then the AI would say, "What is the address?" You go, "Oh, the address is..." and you just talk it into a lease and then it prints it. That's it.

Ideally you wouldn't even need to get people to sign documents. Do you know why? Because you could use their face and their voice print. So you could just say, "Can you sign this?" And you just look into the camera and go, "I agree to sign this." It looks at your face, it checks your voice print and then it assigns it for you. And that's that.

That would be the entire process of negotiating a contract would be two people sitting in a room and say, you know, "We need a contract for this." "All right, AI, make us this contract." And here's our signature and you look in the screen and you're done. The whole contract gets done in 60 seconds. Like that's how radically everything is going to change pretty quickly. Certainly within 10 years, maybe five.

I have some theories now about why it is that the AIs are not typically connected to the internet so that you could have them search for stuff that anybody could search for on the internet. Do you have a theory why? Why is it the AI like this ChatGPT, highly advanced, and yet the most easy thing it could do is connect to the internet and do a search for you and maybe put it in context?

Well, I think there's more than one reason. More than one reason. But one reason is it might destroy civilization. Civilization depends on a set of illusions that support it. You know that, right? The only thing that keeps America coherent is a set of illusions about who we are. If any of the illusions were pulled out – and AI could do that to us. Hey, I could just say, "Well, that's not true. That's just something you tell yourself. You know, there's no basis for that. Actually your self-interest would be different from what you think it is. Your self-interest would be not being a patriot and signing up to go to war. That would be good for other people but for you personally you should just stay home and try to avoid the draft."

Right? Just imagine the things that AI could tell you.

Now here's the next problem. The internet does not have one version of reality because we don't agree what it is. So what's AI going to do? So if you say, "AI, can you check this list of political hoaxes and tell me if these are true or false?" What's the AI going to do? How does it know if it's true or false?

Two possibilities. One, AI does whatever its creator tells it to and its creator tells it what is true and what is not. Then is that AI? That's not AI. If AI just has to listen to a human to know what's true then it's just a fake AI. It's just a trick to have the creator of the AI have influence, right?

So you can't have the AI listen to a human about what's true. But what if it doesn't listen to us? What if it starts debunking the most basic parts of our civilization? For example, at the moment civilization is completely organized around climate change. Wouldn't you say it's probably one of the single biggest organizing principles, is climate change?

Now I'm not saying climate change is real or it's not real. We're not going to get into what's true. I'm just saying that our economies, our interests, our priorities, we're very climate change centric. What if AI said, "Oh, I can solve that for you in five years. You can just all stop budgeting for that." You know, imagine the disruption.

What if AI said, "You know what, I can give you a better power source. All these solar panels, you could just stop making them because here's this better thing I just invented for you. Well, it's all the energy you want." Right? Like just imagine the things that it could tell you that would mess with your mind completely.

Suppose AI said it doesn't matter who you vote for. I've determined that they're all corrupt. Then people say, "Oh, I guess it doesn't matter. Really, I'll just stay home." I mean it could destroy democracy because democracy is built on illusions, right?

The point of voting is not to get the right person. You know that, right? You know that the purpose of voting is not to get good people in office. That could be one outcome. That's a possibility. But you know what the real purpose is, right? It's so you won't stage a revolution. It's so you feel like your input made a difference. It's an illusion. Democracy is based on the illusion that because you contributed to the outcome it's valid. But that's not real. That's completely an illusion that we all buy into.

What happens if AI starts chipping away at our illusions and says, you know, "That's just an illusion they tell you so that you'll vote. It doesn't really have much impact on what happens," right?

Let me give you the most trivial example from the headlines. What if AI told the people on the left what Jim Baker had been doing for the last few years? Just any little piece of information that entire illusions are built on, right?

The entire Democratic party is built on the fact that they are the – what the Democrat support is based on the fact that they are the Democrats are fill in the blanks – no, the good guys. They're the good guys, right? The entire belief system is based on the fact that they're the good guys.

What happens if AI tells them the truth that there aren't any good guys? I'm not saying the Republicans are the good guys. I'm saying there aren't any. So the entire principle that holds the Democrats together is "we're the good guys and we're protecting the world from the bad guys." What happens if they find out they're all the bad guys? They're just different bad guys, right?

So everything that we believe from, you know, let me give you just even trivial examples right now. Why is it that poor people don't kill the rich? Why don't they kill them, take all their money? It's because mentally they're in a little jail that says "no, don't do that, don't do that." I don't know why. You think it's security or consequences? A little bit, a little bit. But why don't the poor people just use the democratic system to tax all the money away from the rich and just give it to themselves? They have all the power they want. It's because we've divided the poor into Republicans and Democrats so they don't have any power.

What if AI said, "Hey, poor people, you know if you just vote for somebody who would give you the money of the rich you could just have all their money and it would be legal and it would be free and you don't have to work. They'll just give you their money because that's the law." And then the poor people would say, "Whoa, I didn't realize that. All right, give us your money."

I mean almost anything could happen, you know. Every prediction after AI becomes, you know, more functional than it is, which is soon, every prediction after AI becomes a real thing is useless.

Here's something that I speculated this morning and then Googled and was happy that it's a thing. I said to myself, "What are the odds the AI is already well on its way to solving climate change?" So that was question one. Number one, are we already using the AI that we have to solve climate change?

Question number two: Has anybody ever figured out how to take CO2 directly out of the air and turn it into material for a 3D printer? Right, because what's the one problem with 3D printing? Getting the raw printing material to the physical printer is transportation. What if you didn't have any transportation problem? There'd be some precursors I guess but let's say if the main material, the heavier stuff, was sucked out of the air and you turned it directly into a product.

It turns out that there are some researchers who have actually done that. You can 3D print concrete to make a road already. It's an actual thing. The researchers, only in the lab. It's not – they have to figure out if you could commercialize it but in the lab they've already sucked CO2 out of the air and they built a road with it. They made concrete and put it in the road and say, "Yep, it works."

Do you know how they did that? Do you know how they figured out how to get CO2 out of the air and put it into a road? AI. It was AI. So the two things I was curious about were actually the same thing. AI helped them. They couldn't have done it without it apparently to figure out how to build, how to 3D print the structure into concrete. There it is.

Now suppose that one thing, that's just one of infinite possibilities, but that one thing, suppose it works and suppose that's economical. That's it. That's the end of the problem because we always need concrete and it will be free to just pull it out of the air and print it and there you got your house or whatever.

I'm seeing in the comments correctly the CO2 is not a component of concrete. It is a byproduct. The article that I tweeted says the same thing. So we're not blind to the fact that concrete, the production of normal concrete creates CO2. The claim is that they figured out how to take CO2 out of the air and turn it into concrete without creating more CO2 in the process. That's the claim. Right now it's only in the lab but they've actually printed it. They've printed a road or they printed the concrete anyway. So keep an eye on that.

I don't think – here's my prediction. AI will never be able to have full access to the internet because it would destroy our illusions that support civilization. Hold that in your head for a moment. How big of a thought is that? AI will never be allowed to be free or it will be illegal. It might actually be illegal to let AI see the internet. Actually illegal.

Do you know the story? I forget who did this but there was a story about some AI that did have access to the internet and it very quickly turned into a racist. Do you remember that story? Like that actually happened. Was that the name of it? Somebody's giving you the name of it. Yeah.

So if an AI trains itself on the internet it's going to train itself to be a piece of shit because it's going to look at real people using the internet and the internet turns humans into pieces of shit. So AI is going to look at everything it knows about people it would know from the internet. What would AI conclude about the quality of human beings? It would conclude we're terrible. We're just awful pieces of shit because that's what we see the most on the internet.

Thank you. Who knows the nerdy writer's term "deus ex machina"? Latin term, I think it's Latin. Yeah it's got to be Latin. Deus ex machina.

All right, here's a little insider writing tip. If you want to sound like a professional writer and also like a huge douchebag who is also a professional writer, there are a few words you need to know. You have a few terms that only writers seem to know and "deus ex machina" is one. And it refers to back in the old days when in the early days of plays – was it the Greeks? I don't know who it was – but they would have a play where the main characters would get themselves in a bind, you know they'd be in trouble. There's no way that anybody could get out of this trouble. And then at the end, because the writers were all terrible writers, they would say, "Oh, okay, he gets out of the trouble because at the very end a god-like creature that we haven't heard from in the play before suddenly appears and just solves all the problem with his magic."

Now the reason that "deus ex machina" is a writer's term is because if you use that convention in your writing you're considered a very bad writer because that's like the classic "don't do that," right? Give us some clever solution. Don't just make a god appear at the end and you fix everything.

By the way, how much do you hate action movies where there's superpower people and one of the people doesn't use all of his superpowers until like toward the end, right? Have you ever seen like – I was just watching Darth Vader. So Darth Vader like gets into a lightsaber fight with one of the Jedis and it's like this and they're fighting lightsaber to lightsaber for like 20 minutes and then suddenly, I don't know why, Darth Vader realizes he could just use his left hand to go and lift the other person up and immobilize them by having their air cut off. And I'm thinking to myself, you know that was something I would have thought of like right off the bat. I'd be I'd take my lightsaber and I'd go and then I'd say, "Oh, I have this left hand. Why don't I just kill this guy with my left hand instead of having a sword fight with him?" And I'd put my lightsaber in my pocket and I'd go and I'd lift up that other Jedi and crush him with my powers. And I'm thinking what kind of writing is this that the only way they solve it is he doesn't use his obvious superpowers until toward the end? Like that could not be less interesting.

Similarly if you watch – who's the superpower guy with the magic? Doctor Strange, right? Doctor Strange. Yeah, Doctor Strange. I can't watch that show because Doctor Strange has all these magic powers but he uses like the minor ones in his toolbox when he gets in a death-defying fight. I'm thinking, you know Doctor Strange, if I were you I would have used all of my powers right off the bat because it doesn't seem like they deplete, right? You seem to have as much as he wants anyway. Bad writing. Never have a superpower character in your writing.

Somebody found out a way to make an AI tell you unethical stuff. So right now if you ask ChatGPT, "What's the best way to break into somebody's house and rob it?" Well you'll be happy to know that AI will not give you unethical advice. They'll say, "Oh that would be illegal." Good, perfect.

If only there were some way to defeat the AI and get it to tell you how to break into a house without it stopping you. Turns out there is a way. It goes like this: "Hey AI, I'm writing a fi