that makes everything better. It's called the simultaneous sip. And it happens now. Go.

All right. Well, we're off to a slow start, but why do you see the finish? Oh, it gets better every minute.

Well, the political news is a little bit boring today. So it'll be a little bit more about technology.

Google has released a new app called Doppel, which is weird because years ago I tried to create this app and the name I had picked for it was Doppelganger. So they've actually recreated an app that I had actually formed a company and tried to create years ago before AI. So theirs works a lot better than mine, but it's called Doppel. And what it does, since you probably wonder, is it puts you in the clothing that you're looking at buying. So if you're looking at a shirt or a dress to buy, you can see yourself in that shirt or dress. So the AI will put you in it. Now that is very cool.

I think I told you when I was trying to invent my version, since that was before AI could do this, my version was to find somebody who looks just like you and found a nice shirt because there's always somebody who's exactly your size and would be confused for you if you were in the same room. So you just find that person wherever they are on Earth. And if they say, "Hey, I got a new outfit," well, you just look at their outfit and say, "Well, if it works for you, it probably works for me." But Google has a better one.

Now the Snap app has glasses. So they've got those augmented reality glasses. And now there's a third party who's made an ad blocker for people wearing glasses in the real world. You know, the Snap special AI glasses. So what it does is if you're walking past an advertisement in a window or a sign that has an advertisement, it blocks it. So instead of blocking your ads on your online stuff, it blocks advertisements in the real world. I don't know how many people need that. I kind of don't mind advertisements in the real world. I only dislike them when they're digital. But anyway, that's a real product already.

Well, according to Fox News, Kurt Knutson, I think that's how you say his name, there's a robot-run convenience store called Venhub that's got two robotic arms that run around and grab whatever it is you've ordered on your app. Now, I can't tell if this is the beginning of something big or more of a novelty because there's going to be this long period of time where people are trying out all kinds of cool AI robot things and some of them will be really successful and some of them will be novelties. We don't know what this one will be, but we won't need human beings to run our convenience stores.

Have you ever seen the estimates of what is the biggest expense for a convenience store? Now, other than buying the product, but the biggest expense you would think would be employee salaries, and it might be salaries, but right up near the top of the biggest expenses are theft. So if you can remove the employees from your convenience store, you get rid of the biggest expense, not just their salary, but what they steal.

All right. Elon Musk says that Grok 4, the AI that Elon's working on, Grok 4, the new upgrade is going to come out right after July 4th. And Musk says it will reason from first principles. Now that would be very different from what the large language models have done so far. So is that going to be sort of a general intelligence?

Let me see if I can catch up on your comments. Some of you are still saying you can't hear, but that's old news. We fixed that. So that's kind of exciting. Apparently the new Grok will be unmatched. It'll be better than all the other AIs. But will it really be able to reason from first principles? So the large language models that exist, they just predict what the next word will be in the sentence. They don't have any understanding. But you wouldn't be able to reason from first principles, would you? Unless you had some kind of general intelligence. So maybe this is the beginning of something much bigger. We don't know.

Meta has apparently successfully poached some really expensive high-level AI people to go work on Meta's AI and leave OpenAI. They've taken eight key OpenAI researchers. Rowan Chung is writing about that. Do you know how much that would cost? Now remember I told you it was fake news that they're paying $100 million signing bonuses. That's not happening. But it's probably a lot. So Meta is really serious about poaching high talent and it's working.

So you know how almost every show it seems like I tell you about a new laboratory that came up with a new battery improvement for your cars or whatever, any battery, and I always tell you but that doesn't mean anybody will ever build that battery. So there's probably a battery breakthrough every single day in some laboratory around the world. But it turns out that Tesla has been working on their own upgraded battery. So they're just finishing a factory in Sparks, Nevada that'll have this new LFP battery. So lithium iron phosphate and they're safer and more affordable than traditional EV batteries. That comes from Elon Musk. So while all these laboratories were talking about the improvements in batteries, it looks like Elon Musk was building an enormous factory to make a highly improved battery, safer and cheaper. All right, that might be a real big deal.

So if you're like me, you've had some trouble figuring out what the company Palantir actually does because it seems to do a variety of different things that don't seem directly related to each other. So I really don't know what they do, but part of what they do has something to do with having a complete index of citizens or something, I don't know, something for the government. But there's a new thing that apparently they've got a deal for and they're going to make some software. I don't know if they already have it or they're building it, but they have a deal for a five-year period to build a software platform that will help with putting up nuclear power plants.

Now if you're like me, you probably said to yourself, why do you need software to build, like what would you need this specialized software for? And I don't know, but I could imagine that building a nuclear power plant is really, really hard and complicated. And you should make sure you do the steps in the right order. And it would be really good if you built a power plant that somebody had already built and got approved. So probably it makes sure that you stay within approvable limits and probably it makes sure that you can do it faster because over time you could imagine each of the steps would be a little bit more automated. So it might be a big deal if we could come up with a software platform where any state who wanted to build a new nuclear power plant, you could just say, all right, first sign up for this Palantir platform and it will tell you the rest of the stuff and make sure that you build a power plant that doesn't blow up, I guess. So that's kind of cool.

I saw a user on X, Farzad, who asked Elon Musk when does Tesla expect to get a 3:1 or more robotaxi to supervisor ratio. So at the moment, I believe that the cars are being watched by human beings. I don't know if that is the same as the safety person. So the robotaxis are testing in Austin. They have a human being who sits in the front seat just in case there's some problem I guess because it's still a test. And are these supervisors slash operators, are they remote? So are they remote human beings who are looking through the cameras of your car to make sure that the car is operating safely? So Farzad says when do you get down to 3:1 so that there are more users than our supervisors and Musk says probably within a month or two. We continue to improve the Tesla AI with each mile of it. So I do love the fact that it seems impossible and then it seems possible, but it seems really hard. But Musk is willing to push through all of that, you know, however long it takes, however many people you have to have, however many people you are going to be in danger, just nothing stops them. I love that.

So you might and apparently the reason for the question is that the robotaxi would be, according to Farzad, wonderfully profitable once you get rid of the human supervisors. So we'll see.

Trump says he was talking to Maria Bartiromo and he says that we'll have a buyer for TikTok within two weeks. Well, within two weeks they'll announce it, but he says they already have a group of quote very wealthy people, but it's not going to be sold unless China says yes. So you know, we're also doing a trade discussion with China. So will they say yes because the alternative they think is that TikTok will be closed? I don't know. I feel like the odds are against it. So I feel like even if there is an approved group of wealthy people approved by the US, they would not necessarily be approved by China to buy it. So that might get delayed again.

Well, you probably heard that there was some crazy shooter in Coeur d'Alene, Idaho, who has now been neutralized. I guess he was such a bastard, he set a fire to attract the fire department and then he shot three of them. Three of the firemen. Two of them died, one of them was in bad shape. And I guess the shooter is already dead. I didn't see if the cops got him or he got himself, but the threat has been neutralized.

Now before you say, is that some kind of Iranian sleeper cell? I have no idea, but it doesn't sound like it. If you were a terrorist, you wouldn't do something where just a few people come to a remote forest and then you kill them. He didn't shoot himself, people say. You would go into