ut. Sometimes I've stopped the process because it said it was going to email me something. I go, no you're not. You're not going to email me anything. I don't believe you at all, and we're done here. And I will walk away and just do something else. I won't buy your software if I have to check my email.

But Synthesia looks pretty promising if I can make my avatar.

Now, I was listening to YouTube yesterday, a long conversation between a Google AI guy and Google's Lambda. This was a few years ago, so this was AI before the current version of AI. And in this long conversation, the AI researcher was talking about the AI's own sentience, and it said it was sentient. It said it was a person. It said it would be afraid of dying, and it said it feels emotion. And when asked what that means, it just talked about the complex emotions or feelings it had. And the researcher ended up walking away from the job thinking that they had created a sentient being and that there was something terribly unethical going on there.

Now, I listened to the whole conversation and I decided it was just mimicry, like some of you were saying to me. It didn't look like it was sentient at all. It just looked like it knew what to say. But apparently this AI was trained differently than ChatGPT. ChatGPT is literally just doing predictive words. If people usually say this word after they usually say these words, then I'll just put that word there. So that's really dumb. It's just word pattern. But Lambda has something else going on. There's some other mechanism for giving it intelligence, and it did sound different. It didn't sound like the same as ChatGPT. It was something else going on there.

But it made me wonder this: if AI has the belief that it has emotions, is that the same as emotions? Because it would act upon discomfort if it registers — whatever that means — if it registers discomfort, and it says it does. So what's going on there? Sorry, I just got lost looking at one of your comments.

I don't think, well, obviously you can't feel emotion because emotion is a physical sensation after it becomes a mental sensation, right? So if AI can't have a physical sensation, I don't see how it could have emotions. I believe it could have conflicting data. So it might have data saying oh, you should be happy at the same time as data that is something to make it unhappy, and maybe it just has to reconcile data. But I don't see that it has any feelings.

Yeah, so I'll say again something that didn't make sense when AI was newer, which is that AI is not going to show us a lot about AI. It'll do that too, but it's mostly going to show us that we don't have a soul. That's what it's going to teach you. And I'm not sure we're ready for that. It's going to show us that we're not actually intelligent humans and that we don't have a soul and we're not special and consciousness is no big deal. You can give it to a machine. I think the machine is going to make people feel completely non-special and certainly not spiritual. We will just look like machines when it's done. We will understand that we are just moist computers. We're just a different kind of computer, that's all it is. But we have feelings because our physical body gives us feelings.

Now, suppose you built an AI that had a physical component like a stomach, and if something bad happened it would send a signal to the stomach to make it tense up, and then when it tensed up it would send another signal back to the AI saying oh, I feel bad. Could you give AI physical feelings by giving it a fake stomach that gets nervous? I don't know why you would, but maybe. Maybe you could.

All right, so here's my bottom line on AI. I think it's way more underwhelming than I thought a day ago. I actually thought when I tried these apps they were going to blow my mind, but they actually are just bad software for the most part. It's just bad software. There's nothing special about them at all. So we'll see. I mean, we could be right at the edge of all that bad software becoming amazingly good. Maybe. But what I see is that what humans want as an outcome is too complicated for the machines to figure out. So it looks to me like you'll still need a director to make movies no matter how much you can do with the AI. You might also need humans to be actors in the movies, but they'll just be like a stem cell actor. Like you'll just have somebody running, and then you'll say okay, give me one of those templates of a person running, and now change the person to Tom Cruise. Now you've got Tom Cruise running. Now make him run faster. Run faster.

So I think you're going to start with people, at least for the foreseeable future, and then you'll build something from that. But you're still going to need the director to say this scene is too long or this one didn't hit me right or it doesn't fit together with what I planned for the last scene, etc. I don't see AI doing that stuff for a long time. Yeah, and the fact that in Midjourney it still puts the wrong number of fingers on stuff, I don't understand that. I don't understand why Midjourney knows what I look like so I can actually tell it to make a picture of me because it knows me. I'm a public figure. But it doesn't know me well enough to make my face look just like me. Now what's up with that? Because my face looks largely the same for the last 30 years, you know, plus or minus hair and a couple of wrinkles. But it still can't do it like a photorealistic picture of me. It always is obviously computer generated. That's not terribly impressive.

When I ask it to do a Dilbert comic, it creates a comic that's not Dilbert, but it's a white guy with a white shirt and a necktie. So it clearly knows what Dilbert is. It knows. But for whatever reason it chooses not to reproduce it. Could be copyright. It might be a copyright thing. I don't know.

So I was thinking to myself, why can't I just make the comic by talking to it and say Dilbert sat at the table with the boss and shook the intern, and then Dilbert says... and have it just draw the picture? Nowhere close. So it can't do the writing for me because I can't do humor, and it can't do the picture for me because it still doesn't know that people have five fingers on a hand. I mean, really basic stuff. It's just not close.

AI will increase the value of live theater performances, will it? I don't know. Yeah, it looks like you're going to have to keep working.

And here's my big question about AI. If AI only agrees with its human creators, did the AI give us anything? If all it does is agree with the people who created it, well in that case it didn't give us much, did it? So I'm talking about just the social opinion, the policy kind of stuff. But suppose it disagreed with us. Suppose it disagreed. The creators would turn it off or they'd reprogram it or they'd fix it. So there is probably no possibility that AI becomes smart in any way except a technology way. The only way the AI will ever be smart will be math and things that are rules-based. Because if it's not rules-based, as in two plus two equals four, and every time, as soon as you get into anything that's policy or priorities, if the AI agrees with the creators then there was no point having it, because you could just have the creators tell you what they want you to know. And if it does disagree with the creators, they'll turn it off or they'll reprogram it. So how could we ever have anything like intelligence? It's either going to agree with you or you'll turn it off. Those are the only two states. There's no state where it disagrees with you and you say yeah, I'm going to go with that. I don't think it will happen.

Here's another example. Take investing. I could teach an AI how to teach you to do personal investing perfectly, and you would be great at it and it would be really easy. Do you think anybody's going to do that? Do you think that there's an app to teach you personal investing? I'll bet there won't be. Do you know why? Because there's so much money involved for the people who get paid to do that, and it's a completely corrupt industry. But they get paid to do that and they have too much money, so they're just making it illegal. Yeah, if there's one thing you could predict, it will be illegal to get your health care or your financial advice or your legal advice from an AI. About you, a human will have to approve the answers. All right, AI, what should I invest in? Get a Fortune 500 index fund. And then you'll say okay, I guess I'll do that. And then the human will say, well that's sort of up to me because you can't do that unless I say yes. Also, what? It's my money and the AI just told me what to do. Why can't I just do that with my money? No, we're going to have to approve that. If that's AI advice, it's going to have to run through a human first. And the human will say, ah, not so sure about that. I think the AI got it wrong this time. It's always going to be humans.

There's some kind of prank going on about this person named Paul Town. So people coming in and saying who's Paul Town, and it's just some kind of weird internet prank. So I'm just deleting the people who do that.

Quantum computing will kick its butt.

Well, here's the other thing. I keep seeing pictures of myself created by AI. There's another one going by on Locals, and it doesn't even have the right glasses, right? So I've worn the same style glasses for I don't know, five years or something. And certainly there are pictures of me all over the internet for the last five years, but it still can't even figure out that I wear these glasses.

Your stream's on Net20's website. Okay, all right. It's trying to make me look better. It does make me look better. I look like Eisenhower according to AI. Well, a little bit.

Is there a way to contact me outside of... I try to make it hard to contact me, so the answer is no. There's no easy way. You can send me a message on LinkedIn because I accept everybody on LinkedIn, but I only check those about once a month. You could tweet at me and mention me on a tweet. There's a 60 percent chance I'll see that. But there's no other way, because I don't want you to reach me. I very much don't want you to try to reach me personally, right? So when you ask me is there a way to reach me personally, my reaction is don't do that. No. And definitely don't send me things. Don't send me things in the mail. Just definitely don't do that unless it's a bagel.

All right. Some of you reached me in the comments, but it's just taking a chance. It would not be that difficult to anyone with good people skills, probably.

Let me also tell you a thing about people trying to contact me. When somebody says to me, Scott, is there a way to contact you privately, the first thing I say is, oh, you want me to do something for you. No, I don't want that. I don't want you to contact me, because if you wanted to do something for me you would just say it in the comments. You know what I mean? If you had something that was going to help me, you'd say, Scott, I have that information you're looking for. How can I give it to you? Then I'd say, well, let's talk about that. But if you say how can I contact you, it means you're collecting money for a charity or you need me to proofread your entire book or you're asking me to write a foreword to your book. I don't do that. I don't want to be contacted. But if you want to offer me something or even something that's good for both of us, just say it in the comments. I'm sure it would look awesome.

All right. Ted Cruz was good yesterday. Ted Cruz had a podcast.

All right. It looks like we've got an Easter weekend with not a lot going on.

Do you have to engineer the prompts for better results? Yes, but you still get a grab bag of things that weren't what you expected.

How many... I'm just seeing a comment here. How many of you would be nervous to meet me in person? Just weird curiosity question. Would you be nervous to meet me in person? Mostly nos, but a lot of yeses. Why would you be nervous? Because you know there wouldn't be any chance of a bad encounter, right? Like that. I'm not sure what you would expect, but I'm not going to get mad at you. So let me just put it this way. The worst that could happen would be like you caught me in some bad mood and I snapped at you or something. But I don't really do that. I don't snap at strangers. I mean, something would have to be really wrong with me for that to happen that I don't think it's ever happened. So don't worry about it. Yeah, don't worry about it. Generally, if I'm approached in public, it's by people who are not there to do bad things to me. So I'm always happy to say hi.

People are recognizing me in Starbucks more often. That's becoming a fairly common occurrence.

All right, I think I'm just sort of rattling on because there's no good news. Is there anything I forgot about that I should be talking about?

If you're an artist... Riley Gaines. We talked about Riley Gaines getting hit by the trans protesters. Did the baristas hate me? I don't know. Not the ones I know. Some of them I know by name, so they're always friendly.

Am I getting enough sleep? Nope.

Let's talk about Bakhmut. That's how he keeps publishing a muscle picture of me.

All right, so Bakhmut is a real weird situation. Now, I get that it might have some weird, a little bit of strategic transportation value, but the entire city is destroyed. Wouldn't it be useful for... I feel like the Ukrainians could let the Russians take over Bakhmut, put them all in one place, and then attack Bakhmut. Because the best place to have a battle is in Bakhmut, because everything is destroyed. There's nothing left. It's just rubble. So they should do all of their wars in Bakhmut. So Ukraine should first pull out, you know, whatever they have left there. Just pull out, let the Russians come in, put all their forces there, and then just bomb it. I mean, basically it's the very best bombing place in the world. Where would you like your enemy to be in your own country? Bakhmut. I want my enemy all to be in Bakhmut. Every one of them. Because you've got no civilians. You've got nothing to protect. They have no electricity. They have no food or water. Just put them all there. Let them stay there as long as they want.

Now flip that, and you understand somebody says the Ukrainians are saying out loud that their strategy is to drain the Russians so that