OpenAI, code name Sweet Pea, sweet peas. Siri will work. Anthropic launched cowork. AI on every corner. Technology is a lot, there's little point. Unauthorized sexy photos of women in Grok. The growing criticism of Amazon because of AI, a tool that houses goods without vendors. Conflict around the AI Book Club. China is checking Meta's deal for two billion dollars in buying AI start-up. The "You're dead" app went off at ten times. And in China, medical AI BaiChuan M3 was reported by a stronger GPT 5.2 and any doctor. Hello, everybody! We're on ToTheMoon. Technological news and sites from the Silicon Valley and around the world. Our e-weekly graduation, uh, Sunday. Oh, write off the audience, and it's interesting who saw the AI culture on Wednesday and how much you liked? There's some comment, I see, a lot of it. If you didn't, look at it, too. Ah, um, s-- Maxim had a new guest at our place, Maxim Grigoriev, and, uh, he's planning to do a safety issue with him. We'll see when he's here. In the next month, I guess it's theoretical in the next month.
Let's get started. So, uh, I think it's very interesting. We talked about the last time that OpenAI is focusing very much on creating audio-based devices. And a big rate this year is being made. And a leak in Chinese news shows, uh, a leak in the first, uh, the iron development of the OpenAI under the leadership of Johnny Ava and showed the Chinese right how it looked. Well, some kind of maquito, not what it looks like. And the point is, it's not, well, it's a device that's put in the ear. And it seemed like I started reading the article, uh, I thought I was gonna go with that thing. You know, when I was young, I remember I'm here--
The servants are like that.
Hands free, yes, it was. There was free hands in youth. Or, uh, here, uh, a capsule showing up in the ear, and a little podium like a capsules. It is expected that, that is, a discussion device closer to smartphones than just a device that writes a voice. I mean, it's an idea, at least once again, we don't know anything about the discussion, see nothing, understand. Because I had the first question: would there be some kind of sym card, like that, digital? How will this work? So, in the descriptions, there's a conject device, there's super cool chips and there's gonna be a condom, including, on the phone, on the iPhone and, uh, on the other devices. A specific condom, a long time to work, and, and the whole interface will be approaching the work with a voice. Well, that's close to what we see in the artificial intelligence movies. Oh, and again, it's gonna be a long time and costly job. It's announced at the time that it says it's gonna cost a lot. Although...
Six hundred dollars, don't I think? I heard somewhere.
Well, I don't know. I saw it, I saw it, and there's a lot of stories in the media. It says it's very expensive. I thought, Ilnar, honestly, it's very expensive that it's, uh, a human base, that's supposed to be thousands and, you know, not a thousand, right? And while the plans are supposed to be in OpenAI, there's a plan that they'll sell tens of millions of things in the first year, right? And, uh, I'll see, by the way, that's a very interesting fact. Oh, we're, uh, gonna talk about this, but, uh, Apple. So Apple had a crazy failure in artificial intelligence. In 2025, they didn't release everything they promised. They couldn't do it. If you look at them, and, uh, the release, the sale of their iPhone 17, they're all over the market for the sale of cell phones. I mean, the mobile phone industry in the 25th year grew by 3 per cent, and Apple's 20s. So, sales. And it would seem they sold a phone that just gave a new color. It would seem like another or some new location of the cameras. So he didn't give it up on top of the idea. We'll talk about Apple. And you, Ilnar, saw six hundred dollars, right?
Yeah. But, uh, all we're discussing is leaking. Yeah? I mean, again, they might be right, they might be wrong. There's no official information. There are pictures, diagrams that went somewhere in China and went further, and that's why they got jealous. The people have. And, uh, the six hundred dollars I saw, that's, you know, that's a rough estimate. Look at the fix, like, these chips, there's a lot of a battery, that's a lot of pressure. Ah, compare it, there, with the top phones, and we'll read the screen. And we're doing a good-minus thing, just without a screen. And there's a plus or minus of $600 left. It looks expensive anyway. A parole guy. Well, again, why does it look expensive?
Because no one understands what function, what is the difference between this equipment, if it comes out of the headphones that are cranked up there, i, to MacBook, to something else, right? If the calculation is made right on the device itself. The device, there, the size of the egg, I think, yeah, usually in the leaks. It's also in itself, there, I don't know, Mac Studio, about which Max Grigoriev said, which could be a big model, it didn't work for that. It's still some sort of proxy that's gonna throw everything you're saying. She's only got some small calculations inside her. What's the difference between the phone program and the headphones that are in the middle of? I don't understand. So there's a lot of leaks. The whole Internet is burlit and wonders, and what can be done? I haven't seen any ideas on this yet. I mean, well, to be a straight-up, real difference, some new espionage compared to, uh, headphones. Here. But it looks interesting. We'll wait. Johnny Eve, don't let me down.
And none of these devices have a screen, right?
There's no leak. On this leak, you have a device, there, if I remember correctly, the size of the egg, inside which two capsules you either put in your ears or you're slapping behind. I never understood the picture, actually, how--
Yeah, it's like a picture of you as if you were in. But nothing is clear, of course. Not much of a Chinese character picture, and that's why you realize it's painted by Chinese people. Yeah, it's a , obviously, it's not a picture Johnny Eve painted, is it?
As the Soviet Union said, Engineer Madinchin was doing. Made in China was then read by Madinchin.
It did. Here. So, as a matter of fact, it's not clear, but it's all about the voice interface, that's, you say something, you hear the answer. These are weird flashbacks. Sas, you said about the garnitural of hands free, with which people go, look forward, say something right in front of you, and you don't understand that you were told or in your ear, in your phone, right? But that's how it's gonna be more and more. We'll talk out loud, but not the interviewers, but here's some device that listens to us.
Well, actually, it's for people who are more easily audible to take. I don't think they're most. More like people to visualize the text, something to jump on, accent, a scrimple. I don't know how much I can imagine without a screen. For me personally.
We're waiting for contact lenses.
Yeah, with the lens, maybe
it's fun.
There's an interesting aspect here, of course. We have a lot of experience, and this experience is kind of big, and it's not really big. Twenty years ago, I don't remember the first time I had a ninety-ninth, probably a big duct with antenna, right? And there, but Nokia's the first one in the beginning of the two-two, there's just a little more than twenty years ago. And obviously, we had no idea, in fact, we didn't want the text. We were just gonna need a number or a samescu, you know, esemescu's gonna send it. And how we've been getting her hard. In theory of this interface, it's shown there, I don't know, in the Black mirror, it's a lot, you know, or in different movies, but it still looks like a little more built-in chip. I mean, if you have a permanently built chip that goes around with you, it's all over you, plus some kind of external audit system in different spaces that's connected, synchronized, It's a stable job, independent of the Internet. Again, the question is independent of the Internet, right? We can't get a chance to write a book. What to talk about in terms of contact.
There's a problem with the internet in the Silicon Valley.
Sure. Well, I'm always bringing an example with Alexa. Amazon now started rolling all the new voice assistant, this AI is new, including not as a concrete but automatic. And, well, there's a question for him, yeah, when you have a alarm bell there for 30 seconds, sometimes, right? I mean, you know that's-- that's an interface, that's not necessary. At the same time, I have a friend for two weeks now, and we're recording a lot of very much of a content on the outside audio system like a pet, yes, which will be further identified in the annex. Well, you know, this thing is wearing in your hand, there, you forget to turn it on, you forget, uh, some marks there. So you realize that the soft is so simple, it's a little bit of information. Well, I mean, still in a big question. I, for example, would like to see ChatGPT show up something--
I wish there was something in ChatGPT. But Ilnar had the right thing to say, what's the difference really between the headphones? Well, the headphone's hanging, it's hanging. I probably wouldn't want to wear some kind of thing in my ear forever. I'm not sure it's healthy, am I? Well, she's sitting there, but she's probably in theory--
We always carry iPhones with us anyway. You'll be out for five minutes without an iPhone, already disturbing. If there's anything going to happen, there's no iPhone.
Who else is concerned, write to anyone, please, who is disturbed without an ipho-- I'll remember Tanya's husband again. So he and I were going out there on a trip like this, on a mission, on a mission and a little fun. We're going to town, and I'm saying, "Well, we're going home?" He says, "Yeah, yeah, yeah. Let's go. So we're going with him. I said, "What are you doing so fast?" He says, "I've had two hours, I've had my phone. I've already tried to buy powerbank somewhere, but I just didn't find it. That's why, of course, everyone has anxiety of dia, including two hours without a phone. So, um, interesting, if they could do it. Again, viewers, what do you think?
Oh, sweet peas. Sweet peas, of course, funny, uh, but go on. So, again. What's clear? We've talked about this last year, and that Apple chooses who to integrate with in terms of artificial intelligence. There was a discussion about what they did their own, their function. They already said he-- will now use the Google Gemini models, pay for them a billion or a hundred dollars, no matter, uh, Google. The point of what? They said they would use their models for Si-- for Siri, uh, temporarily until they were building their models. I mean, I think this phrase, these contract restrictions, and everything, suggests that, of course, Apple will come to their own models, maybe not. Yeah, here's a release with Max on Wednesday. Uh, can it be that they'll never do it, under any circumstances? Oh, but the point is, I'm just saying that, uh, that name at the beginning, Siri will work. I think it's, uh, well, for me as a user of iPhone, there, the last 15 years is personal, of a great, just incredible meaning, because the current Siri is a stupid interface is absolutely. And maybe someone has a healthy experience, of course, you can tell me about it, but I'd like to see a cool voice assistant inside Apple and I forgot about a bunch of extra sophth, absolutely. It's pointless and unnecessary. Another point would Apple ever develop any inside products from the AI perspective? Because they have, uh, the last and a half years or two years of trouble with this, with the integration of an extra one. So they've done a kind of convenient, more intelligent vetting interface, there's a text you're writing inside, or some microfiche, but a normal solution is great to get you to work with notes or with notes. dictophone records work. Or this Apple Intelligence button you use ChatGPT, and ChatGPT doesn't keep the data. I mean, it looks really weird. Again, who's the Apple users, I know someone's gonna sit here and say, "Apple Large, like we don't need it." But who's Apple's user, how much do you believe in this integration with Gemini, how long is it? I personally would like this to happen. And in a year, we were sitting there, recording the release, and I said, "Well, there's a bunch of Apple safts coming." Here, let's see what happens. Let's move on to make it work, there's a lot of things.
Yeah, yeah, yeah.
I'd be here, Sasha, just adding that there's some irony. Google is the same company Android keeps, and it's, like, for the iPhones, it's gonna be the Apple model. I'm sure Androids won't be split. Android's comment is gonna say that five years ago, it was all and it's already worked. Here, when ChatGPT wasn't, but somehow. So, in my view, a big victory for Google that they pushed the OpenAI out because before that Apple said that they would cooperate with the OpenAI and, in fact, took the niche. And I also think that a big win for Apple is because I like Gemini models more and more. Very cool models. And that they will be available from under Siri, it looks like a very good move forward.
Yeah, but it doesn't mean users say, too, that, for example, Apple will use VEO models. For example, there's a new model now, too, this week, on the generation of text to video, yeah, from the text in video or in the pictures. So we don't know, Apple will use its own further, continue to develop inside or not. Some things sometimes surprise me. I went to the mountain this week and accidentally, in the cave, I clicked on the iPhone. I got a video on my foot when I'm filming what I see. And plus Apple-- iPhone shoots me, and I show everybody, "Look what function." Then he sent a family to the chat room. My wife says, "How did you put it on the iPhone?" Well, we all know that this function is spread in other applications, but not on the iPhone. And it's sometimes a little bit of a foul. Like a pulp that has been present for thousands of years and Apple has no. However, um, the same might be a logic of someone being stupid or developing this annex, but at the same time there is a logic of what allows us to work side by side, to create side annexes. Yeah, we all know that Apple makes huge money to create side applications, and it's never clear that what really drives them. We're not inside. Not enough, Apple is a very hidden company, crazy, totally insanely hidden, yeah. And if Apple had heard a rumor about something, it's a thousand, I think. I mean, more of the OpenAI now, tell me something than Apple.
Because in Apple, it's hard to find out, and it's hard to understand. Let's go to the subject, to the theme Anthropic. Anthropic released Cowork, and-- A, now, well, we'll talk about it in a little more detail. And when he was released, I read it there, was in discussions with some group of people. And people said, "What a cool thing! If it's really working, it's cool and everything." And I thought, what did a great man get Anthropic from a daily job? Oh, and I'm not the Anthropic user, but, well, it's always there every day. I'm the most using ChatGPT and then Gemini, yes, or Grok can, but not Anthropic. And by reading and studying what really does this job in Cowork, yeah. I mean, you, uh, give access to the selected file on the computer. Claude can read it, edit it, create files in this folder and do different actions inside, and it can connect different modules, so there's some sort of presentation and so on. I read it all, I looked it up, and I thought, "What's the innovation? And I'm done, yeah. Who sees from our viewers, too, what do you think about this function, about Cowork, how cool is it? Again, I think Google has already done a huge amount of these things. Or I'm wrong again, right? And it's some supernatural function. Ilnar, you used it very hard. You see some super plus? Well, except for some, some bonus is. It's obvious that this is some bonus.
They just pulled into the standard operating room what programmers could use for a long time. If you remember when MCP discussed it was just less than a year ago, I was just recording a little special episode where, uh, Claude attached some extra models, respectively. There, uh, files were sorted on their computer, determined there was a cat on the picture or there's no cat on the picture. So all this could have been done, what they're doing in Cowork now, and they're doing on their own, with little programming skills. Now, just seeing how Anthropic himself in his note said they were surprised, and how they used everything they created, they decided to simplify a part of the scenarios. So they say that there are quite a lot of users with Claude on their computer, there, files edit, create or do something like that, but there's a need for some manipulation. Do it. They've cut that path now. You said right, you can access some daddy and in that daddy ask Claude, I don't know, there, create files, rename all the files in some way, sort them out or something. Something like that. So you can do some basic routine. If you think there's any more integration with a voice, well, there's a bridge to the previous one, you can just say that there's a file that makes these files, do it. - I'm not. Again, I'm saying absolutely some primitive, uh, scenarios.
Yeah. Well, the difference is, you have this on your computer, like your files, you don't have to move. But again, the Anthropic itself says, "We're just wondering, how you're gonna use it. So now we've opened it a little earlier than we planned, just to see in the bet and what scenarios you've been using in your head. And we'll actually keep them down somehow." Here. It's a fun move. I mean, it's getting closer to LLM, you know, not to the real world, but to the real files on your computer that you're interacting with. Less to copy, to put, from one window to another window. You can ask for a quick answer in the form of a file to keep, which you'll either be out there or print or something. I mean, there's some use. Yeah, not a revolution, but it's a move to the real world that LLM is getting closer and closer.
Yeah, well.
It's getting closer.
And Anthropic did not, however, write in the article that it was some kind of vau and super-invention. Aaah, I want to move on to the next topic, which I think is very important in the world. If you look at the-- in companies, how art is involved, very often they are about some innovation that they should be fired, there are hundreds of staff members automate all processes that were automate. Not automated, automated, automated, and saved a lot of money. So the way it's like artificial intelligence is something supernatural. Although, in fact, if most companies in the world are taken, their level of automation is real and the details are really being processed, he is very seriously affected. And then I-- now there was a pop-- a show in Las Vegas. I've been called it a lot. I don't like to go to these exhibitions, honestly, I love driving and I don't. That was CES 2026. And there, uh, there was a problem that said AI was on every corner. Technology is a lot, there's little point. What was the point? That every time in this exhibition, every gadget was a sticker of "coustic intelligence", even if it was practically not useful at all. It's like, uh, Elnar said that, you guys, you guys, you guys, you say it's an artificial intelligence, and that was on Android, there, ten years ago, yeah, it's done. And the machine training says the program there like that, so... What do we understand under artificial intelligence? Or like Maxim, Max was telling me this in the final edition, Wednesday, where Google started to do very serious engineering training 15 years ago, and all we're seeing now is in terms of the way we're working. Mode-- preparation for models like this. And with all the gadgets that showed up at the exhibition, there was anything, from the toothbrush to the toasters, right? And the impulse starts showing you that you have an artificial intelligence working in your toothbrush that gives you some super-information. But if we actually see it or it's in the toaster, right? It's like a toaster that works, a smart device that works in the kitchen, yeah, an automatic toaster. I remember the video was shown long enough, over ten years. Somehow, they showed how the house was automated at Zuckerberg's. And I remember there was a subject about what supertoster was automatically worth, which means by some time it's a toast for you. But I don't know any man, no one knows who's using a smart toaster. And I have a question that always comes up: I have to get some kind of closed package or bread, like, cut it off to dry it up, put it in the toaster. He's in there, so he's gonna spend the night working on him this morning, right? It's like any smart, I don't know, I've got a smart Wi-Fi refrigerator, I have a smart oven. I don't use it because I have to, like, in the fridge, put something in the oven beforehand? It's there, it's just going to ruin it. I mean, you're the one who puts things in the oven, you start cooking, you start cooking. I see there are species. I'm making a lot of cooking myself. There are some prescriptions that can be established. Write out, by the way, the audience who uses what smart devices in the kitchen. But the point is that the title of manipulation is this brand, which is artificial intelligence. We hear it from a huge number of companies. Many sell so many things that we have artificial intelligence. And remember the case of an educational company, uh, in the education world, for example, many companies have added AI thirds, and the clients say, "Look, if you add AI thirds, we're all We're not buying your products because we have a ChatGPT, are we? We'll use ChatGPT. Why would your AI-Trior?" That's the subject, which is really urgent and problematic. And all people, well, my personal recommendation, I'm trying to do this myself, to take very careful of any such statement, innovation and genuinely test and verify. I have this bracelet, whoop, he gave me a recommendation, for example, that he didn't give me a recommendation, just after he had a lot of data, that I, uh, that supposed coffee, did not affect me, for example, sleep or sleep, or whatever. We must first understand that we don't see their algorithms, but secondly, what is the artificial intelligence? Well, fair. Yes, Ilnar? What's the artificial intelligence? Well, they did, they've set up a thousand parameters between themselves, there. Well, you've been wearing it for a lot of days. So the volume of these data has nothing to do with the notion of artificial intelligence as such. This is a programmable code defined, right? It's not that the language model has turned to something, so I'm in a super-result.
And if it is, it can be hallucinating, yes.
All these AI-shoulders, they remind marketing these gluten fries on sunflower, cholesterol fries.
He's on it too. I mean, obviously, there's no one there, but somehow...
And vegan! The vegan champoo.
Yeah, yeah, yeah, yeah, yeah, yeah. It's just a better sell.
Coscher. The champoo and the vegan on the champoo.
There's no such thing as a colder in the sunflower.
The most important thing is who loves kosher shampoo. Who likes the kosher shampoo and who uses the kosher shampoo.
And you know, it might be like you're getting better from the idea that you're using kosher shampoo. The effect of the flamebo was not overturned. He's more powerful than many other effects.
You'll write it. What kind of vegan is he? Well. So, uh, there's a very interesting case, actually, that's the trend that's going on and on with Guglom, that's what we're looking at now and now, uh, more at Amazon. So now there's been a surge of criticism of Amazon about, uh, having AI tools that house the goods without the seller's knowledge. Look, the case is very interesting in general, in its very essence. What's the logic? That Amazon is testing two experiments. The first is to show in search of goods that are not available on Amazon, to pull data from the brand pages and to keep the buyer on the outside shopping site. And there's a problem that the brands complained they didn't ask them in advance. Actually, it's kind of a way to do it or not. And the second regime that's called Buy for me is buying for me. And that's what Google declared their agent regime so big. The point is, this is a more agent mode. Amazon's inside is a almost ordinary product card, and when you press this Amazon button, it's the one that makes a book on the website on the client's behalf. We remember the OpenAI doing this research, too, yes, I think he stayed there, he stayed there. This is the external purchase on behalf of the client and, well, it's supposed to be encrypted, so it's transferable. But Amazon's move is really fun, by the way, right? I mean, Amazon basically said, not through the GPT chat, you're gonna buy the Amazon stuff, but you're all sitting on Amazon. Well, I see, every country has its services. But we'll see it as one of the leaders who, believe me, has so much money to test, that, well, that's just unreal. Yeah, it's the richest marker of the world. And then they said, "Look, you need to buy something, we'll buy it for you, even if we don't have an Amazon." That's enough, it's a good story. I'm like a very big user of Amazon, and I can say that there's a advantage that if you see the Amazon product, it is, you'll always see the lowest price. The probability of getting a lower price on other sites is very small. She's just connected to the fact that if there's any kind of intrusion, why is Amazon running all the sites? And if the goods see him below his place, he, he, he, he, he, he, he, too, is downgrading his price. Yeah, so that's where the algorithm works. These two functions, of course, will clearly be developed by different companies in the world now. And once again, I remember that Google, even Google had a case, a lawsuit, yes, Elnar, on this subject. Not a lawsuit, but there's a...
It's a story like that.
There's
a dispute.
Yeah, it's a story, but it's a little on the other side. Yeah. I can tell you a little about her.
Yeah, yeah, yeah, sure, sure. It's just that, it's still a little bit of a topic. They're different, of course.
Their plastic cards, yes, Tanya. Well, you use them, you use their office, you use their infrastruct-- their support. So they're tied to themselves. Plus, if you bought them a merchandise, they have an automatic system that might be connected to that seller and offer him to host himself. They're reducing the likelihood of buying. You and I know, for example, that in America there is Instacart, a very large delivery system. It was copied once by a company called Sbermarkt in Russia, right? It was called instamart there in Russia, and then renamed it at Sbermarkt. The point of what? What is this, a system that allows me to deliver food from different stores. But what's interesting is, in America, there's a company called Trader Joe's. It's a very big operator, it's been working for decades, selling food. So, Trader.
Selling food. So, Trader Joe's can't be ordered through external delivery systems, and they don't have their delivery system. Look, they don't have their delivery system. Yeah, by the way, Trader Joe's a fun trend. I don't know, you saw, you didn't see. We have a pair of bags here at Trader Joe's. In London, they're being designed, so the trend and bags are worth up to $1,000. And here you can see in our valley how people get out of the bus and buy these bags. There are Chinese people sitting right out there, running and buying bags. And we have, and Polina--
I was thinking about buying it this week, you know? Turns out it's a trend.
So Polina said that the restrictions were now in place. You come to the store, you can't buy a bag of Trader Joe's, you know, more than two, I think, right? But the point is, it's not that store when you--
We've got plastic bags removed, so how everyone goes to these bags. And there's this trend.
Yeah.
That's cool. Yeah.
Yeah. But Trader Joe's was, yes, Trader Joe's was, like...
Yeah, but there's got to be something to do for those who don't know what a Trader Joe's is. They have a lot of fun, unusual products that no longer exist in any store. They make them themselves. I don't know how, but it's interesting, yeah, that you're still going to this trader Joe's because he's not gonna get anywhere, and the products are unusual.
I mean, their model is pretty cool, too.
Yeah. And here I answer your question before turning to the topic of the Google seminar that why they don't online or why they can't be ordered to Instacart? Because they understand that if I start ordering them through the food aggregator, I'll get used to the food aggregator and start forgetting this brand. And gradually, I'm gonna be offered to order the same product, the same product, because in the Trader Joe's, there's still a lot of unique things. Suddenly, they still have eggs under their own trademark, but eggs are still under the number of other stamps. Yeah? Or there's a packed, I don't know, a chicken or a salmon, or something, there, I don't know, a handwriter, whatever, right? Well, you think I'm not buying a hand, buying under another brand. What's the difference, huh? And you're in it through the aggregator, and you're in a choice. And once you have a choice, you're missing this point of the brand application. Which, in fact, was what Amazon did. You don't care what you bought there sometimes, I don't know, a line or a glue tape, or I don't know, a tournament is there, no matter what. So you're a lot of things that are brandy that stop being taken at all. I mean, if I've known what brands exist that produce a sports product, like I don't know, a tournament, I'm actually in Amazon now, if I order, it doesn't matter to me, right? And that's why Amazon is trying to pull himself over, taking as much leadership as he can, not losing. Because, however, companies like Google and Google first show up-- they tried, Google wanted to, through a search for a way to get people into their goods, he couldn't do it in the last few days. Twenty years, and their models understand new logic of a new user path. They can do that, right? And that's why they're interested in it. I remember ChatGPT once again presented it three or four months ago to your own basket when you had the opportunity to get some stuff through the integration and spend money through their basket, yeah, right, here.
Yeah, and here, in fact, the Consumer Protection Organization in the US has issued a warning that this Google purchase protocol, through Google agents, is not very good for consumers. There's a story with the pseils, for example. And we're doing this here? What Google, if you use Gemini and use Google infrastructure, you have Gmail, photographs and stuff, your activity is all screaming, he knows a lot about you, and on that basis he can offer you the goods, You know, you don't really need it, but at the right moment, if you light them up, you'll probably buy them. Moreover, he knows what purchases you're making, what price category and so on. And there's a fear that, for example, he might start offering you at a high price, knowing that you might still buy it. And here's the consumer protection organization on this subject making noise. A little scandal, H's got these posts. Google had to even answer that story. Of course, they all refute everything, and so on. But somehow, if you do not buy yourself, as it is now, although it is clear that all labels do, in one way or another, crack your behavior and put you in the goods they want. Google has a lot more influence on you in this regard. And on that subject, the Google dispute is about to begin, and it's about to turn.
I don't know what that would lead to. Probably won't lead to anything, but it's not good to think about it in one way or another. There's some news here that's lying there, too. Google in beta-version is starting personal intelligence. What does that mean? If you use Gemini, but you have pictures on Google clouds, you have mail, you have some Google mottos, too. You can let Gemini get access there if you need to, dig up there and give you a response. They give examples of this case. There's a man here, there, I don't know, buying new wheels for his car, and he doesn't remember the radius of the wheels. He's asking Gemini. Gemini's on Google photos of him. On the basis of this, it is in fact determined what wheels are needed. And he also sees that there was a family trip on some photograph, somewhere in the mountains where the snow is. And they recommend that he buy all the sazz tyres there. Or a man forgot his car number. How can that be? Not very clear, but Gemini's been digging in the pictures, finding a picture. And well, maybe that's understandable.
I remember this seminar all the time, because I have a picture of this license plate, yeah. And when I need to, I can't remember her for years.
Me too, I don't know either.
I'm just looking.
I know a couple of cars.
They're in America. Sasha, you have two hundred cars at all. I don't know how you just remember them. I can't remember my own. I'm really, when I need to put it in the parking lot, I'm just looking for a picture of it.
Tanya, you'll have a number soon. Tanya, you'll have a number that can be easily introduced.
He's a repeater, my husband is a repeater, and I made it up, and he did it to himself. Only one letter has changed.
Gemini can be accessed to the data, respectively. Yeah, he could dig into your mail if he showed that he needed it to answer your question. Maybe he'll dig up in the pictures, recognize the data from there or something. The calendar can be stacked. I mean, it's a good one, but on the other hand, if you put it on a shopping story, he'll know a lot about you. And if he knows a lot about you, he can offer you recommendations that are as useful as possible to you. Either the shadow side, knowing that you're buying expensive goods, can put a cheap merchandise down, but with a good brand on top, with a good price on top, yeah, there, or a little something you don't have to put down. You want to buy, but you can't help yourself. And you're buying it.
Well, it's a story. People often think that Google, like Google or companies like Amazon, helps buying, helps to make a proper personalized advertisement. In fact, these companies are only interested in buying more, spending more money, right? Not that it's better. And or is Apple there supposedly helping to choose the app or what? They're advantageous when you buy and don't write off the apps because they're taking 25 percent of the commission there, right? And they're generally benefiting from you just forgetting about it and you're connected to it. Yeah, they-- they're a law where they, for example, have to send you a notice, a letter. They're sending you this notice for one reason, because there's a law like that, not because they really want you to help.
Another story that is very much discussed this week is that, uh, that even name was the conflict around the AI Book Club, and the authors are not satisfied with the new AI functions again, I'm not sure what's going on with the books. And in-- inside Amazon Kindle and inside Eleven Labs. What's the logic? What can we do? For example, a book may have some brief description of the book and do it either in a voice or in a text. It is clear that many things have been there, but the logic is that the authors call and say that, first of all, there is no agreement, and you're not compensating any of the right holders. We know with you that the whole world is trying to read things fast. As fashionable as it was, I remember, even the payroll was software, the apps, they are now where the books were, I remember, a pay-as-you-you-you-you-you-you-you know, a pay-as-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you--you----you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you-you Like a squeeze, yeah, the basics. And I have a friend who I think they bought. And you could as well read a short--- now, there's a ChatGPT, you can come in and tell him, give me a brief account of any book, not even load.
But it's not that, you know?
Of course not!
Of course not.
I'm just saying that, too, and then I just dropped this idea because, first of all, well, there's no point in art literature at all. And in such a bad way, you don't get any general thoughts, you know that.
People don't watch TVs to get the point, do they? Or the book doesn't read to just get it sometimes. If we get the point, we don't read the articles and we're putting them in ChatGPT. Give me a short summit. Okay. But if a man saw a series of six seasons in the 15th series, he clearly didn't look at it to get the content. It may receive a content in the form of one article. That was born, killed, born, killed and everything, right? And you don't have to waste your time.
When I was in school, the writing kits, the pre-written or the short contents of some books were very helpful. So, in general, I understand the people who do it. But again, if there's a target, there, I don't know, to do something, to know the general content or write some sort of essay.
Elnar, you're not gonna be told you read the books in the summary. I never would have.
I read the math books. Here's the
Just read ten million books in a brief. So I have to ask how many times I read and how many times.
Tell me you used your books and everything.
I wrote the writings.
Did you use the books?
Can I not comment?
I can. The answer is clear.
Let's make some aldscull. There's a Demidovic book on the mechmat. It's a mathematical analysis assignment. That's so fat, it's only a task. And there's an anti-Demidovich. Same book, just decisions.
And I used it.
I knew it existed, but I didn't have it.
Well, that's it.
No, unfortunately not.
But I knew it existed.
I see.
I had to decide myself. If we remember a little bit about school. Here, Sas, you were pretending Gemini added, uh, learning opportunities inside. And I was very sceptical when, uh, GPT leaked this story, yeah, when you can learn with GPT. And here with Gemini, I think, let me try. They're doing good things anyway. And I was very happy in general. So I've been studying the topic of vector search with deep research, that is, writing articles, squeezing, finding first-hand vectors, digging something and so on. And the subject was pretty good at all. And here I think, let me try this same subject just with Gemini. And I'm honestly admitting it's very good at work. I'm probably gonna use this subject. We need to see how it works at ChatGPT. I may have been too prejudicial, and it's really working well to deal with some of the topics. I'm really recommending that we just stroll, see. Very funny. I liked it.
Especially for people, Elnar, who can't be super cool, like writing proms or knowing how to start, you know, using AI tools, getting information. And it's even good for those who can. Because we get used to, used to, created some sort of implation for ourselves, and we're going to, you know, create, write these proms, some headlines, everything. I've sent thousands of messages there, I have certain habits, and I can see how I need to change these habits, right? Because I created these habits, for example, when GPT 4 was still in or GPT 3.5, they started to form, and now the answers are completely different, different levels. And I'm thinking, you know, I should use other scenarios, yeah, and ask otherwise. And these systems, including that Google, I think it's a very useful tool to at least see what other scenarios are in the process of obtaining information.
I found a huge difference for myself. I've been making all the deep research, but you've got a story there at nine, twelve pages, and you're just trying it. She's a good, very meaningful, and it's a very useful story. But if you have any questions in the middle, either in the first half or somewhere, a little uncomfortable, there, interrupting reading, somewhere to ask, yeah, in another chat, to go, to ask or to ask, or... I'm not. And here comes a piece of this piece of story, and you're dealing with every step, something you don't understand, asking for examples of something, something else. I mean, it's a literary way of communicating. I tried to get away in math a little bit. He's very well in the way of talking, very qualified. I really liked it. So that's what I'm gonna use, maybe in a series, maybe in parallel. I recommend that our viewers also look at this Learn or the one that's Studying, as it is, the GPT and Gemini chat room, if you use it.
Very interesting experience.
Quick, quick. We, we talked about last year's graduation, last Sunday, if we didn't see, we told them that the OpenAI had launched a separate environment called Health, medical. Whoever used it, by the way, who had access, write like this environment. But now, just a few days ago, a medical AI BaiChuan M3 was introduced in China, who said he was more powerful than GPT 5.2 and doctors. Why would I say that? He's got us all the time, I remember our graduations when we started the models, we called the number of parameters in these models. So this system has two hundred and thirty-five billion parameters. I already think it doesn't matter in terms of the number of parameters. Well, the number is high, yes, two hundred and thirty-five billion parameters, and the optimistic diagnosis and treatment recommendations. And they say that key medical tests are diagnosed, diagnosed therapy. This model has been advanced, and, until-- it has been generally not just, but, uh, a statistically qualified doctor. She's out of the way, including, uh, five and two from OpenAI. Here. And, uh, it's open there to test the various specialists. Oh, and generally, just to keep that background. Oh, again, China has a lot of models. And by the way, head Microsoft, and, a few days ago, said that, uh, everything, well, China, again, is a very high risk to American AI and American-American artificial intelligence. There's a lot of super stuff going on. For example, Aliba showed a cancer diagnostic model. They're actually saying they're like number one on this subject with AI. Well, I'm guessing there's still a lot of things we don't see. But what you want to do is tell me quickly what we've declared.
Ah, the first topic is very important, it can be unmarked for many people in the world, but it's not visible in Ame-- in the States, although it's probably the same in the world. Oh, there's a X--- it's a plac-- a platform for-- it's a former Twitter, it's growing up quite. One of the main news platforms in the world now. It's inside Grok, and Grok allowed, and, uh, he had a new function, he lets the people get naked. You know, you know, you know, you know, you know, like, like, like, like, like, a little bit of a bluff, right? So you could give a picture, there, men or women. I haven't seen any headlines that Grok is smashing men. Everyone was worried that Grok was messing up women. So you're sending a picture, saying, "Take me a man naked." And he did it, didn't he? And he started doing the same thing with the kids. There's been a crazy number of claims around the world. They didn't want-- they didn't shut this function down, they put it in a pay-as-you-go. I think, well, some kind of marketing idiot mistake is absolute. We were told, "If you want to increase the number of subscribers, 'cause you're the one who's got this job, and you've shut it down from everyone." In Malaysia, for example, they were completely banished and brought in judicial investigations, where they were different. Well, I think they'll get a lot of it for that. They've shut this off. But in general, what happened? When this function was available, there was a large number of people dispersed online again. So people were put out with pictures. In fact, it's gonna take a while to find people on the Internet in a picture of us. Now, there's still a problem with that. Now, a lot of systems limit this to do, not looking for it completely and so on. So, there's gonna be time, we all have to understand, yes, that we can find, uh, a man, a lot, well, pictures of a lot more on the Internet and under all circumstances than that. Now. And, of course, the existence of these naked stories and the deepfake is a problem. I think it's a big problem. And she, uh, no government in the world, no system, no one to date, I don't think she can fight or can't. Yeah? We've heard about how Nana Banana could do all the left stuff, there, then in Chinese models, you could just do anything, some people around you.
And you're at some point in time, you're not gonna understand where, who's been dating, who's really been, who's a friend, not acquainted, and so on. I mean, checking old information is getting a little bit unreal. Yeah? Here. And, uh, continuing, it means China. China has an interesting history, so it's happening. There, uh, a huge blow, uh, an annex that had such a specific name means "You're dead." What's the essence of this app? The popularity has been completely stupid and logical. It's a...
He was a logic. It's, and they've just got crazy-- it's starting to cost a lot of money. Oh, uh, logic was that you should just click a button every day, a button that says you, you exist. If you didn't throw it, you're dead. And then there was some other person. It's just a laugh, of course. There was another man who created the same app. The programmer, uh, I'm alive, yeah. So he said he didn't have to talk about it. I'm alive. And that's the same thing that's going on. It's just parallel to the fact that, uh-oh, how much technology in today's social world, there, the world, um, affects lonely people, yeah. And in Ki-Kitai, it's assumed that 100,225 million people live, uh, alone.
It's like Tamagochi from childhood. The game was like that. You should have fed her, and if you forgot to feed her, yes, she died there or something happened to her.
And adults played it, and adults played it.
It's the same thing, actually. You just need to feed this app every day, say that there's another one of my life, yeah, there, and there's a few days there, you've got it all going on. Uh, about the size of the models. Here's two hundred and thirty-five billion. On the one hand, it's a lot, on the other hand, uh, GPT-4, I think it's been six hundred and a billion. DeepSeek R1 was about six hundred billion, too. And now, modern Gemini 3 and GPT-5 2, they've already been over the
trillion.
Yeah.
So, yes, in general, that the two hundred and thirty-five billion model hits large models, that's cool. Well, the only thing to understand is that she's in there specialized in this case.
Yeah, look, she's not that small. You'll agree that this isn't so small for a specialized model.
You're not gonna miss the phone, are you?
Yeah. And it's still a model that's specialized for medicine. I mean, you know, it's probably a very serious subject, and you can really get some volume there. Another moment. If the Chinese did it, what's the medicine, huh? I mean, well, I'm not saying she's bad. I'm saying, on the other hand, what's with the medical perspective? I mean, I like medical requests to date when I ask. I say, "Tell me, please, as it is in Japan, like in Korea, like it is in China, as I don't know, in France, as it is in Britain, like in the United States. So give me a wide range of rules, and he's giving me every country." That's interesting. I wonder if these models, uh-a, are they complicated, how does that work? So what are the data there, what are the studies actually uploaded? They're-- they're closed research, yeah, which is kind of unavailable. This is also a topic of interest, of course, in the examination and comparison of recommendations. We see a large number of cabs in politics when the system responds differently. There's a very strong struggle in the States. Comparing often like Grok and OpenAI say something, yes. When the President of Venezuela was taken, Grok asked. All X's been ripped. Grok asked if the president was captured or not. He said, "Yes." And I was describing why. OpenAI asked about Venezuela. OpenAI said, "Nothing's happened, it's okay." And so, here, uh, the community just went down. Clearly, OpenAI means that this supposed model does not update information on super important things in time. Iia, well, the question is, what regime did you ask, huh? If you asked the regime, uh, um, reasoning, with the reasoning to study these data in depth now, he would have studied everything, yes. If you ask for the author, well, in the author's room, you'll have some data on the model. So here, so in medicine, I think we'll still see, see, they're just not so many boxes when systems give completely different advice, or when the same system can give it to.
different recommendations, not to mention different systems. Here. Describe. Describe what your briefcases were like. It's interesting to hear from the audience, too. And, well, that's a very big subject. Yes, Ilya?
Yeah.
To keep it up. Well, I see they've done it before. Just the OpenAI released this story earlier. So big companies are coming this way. We wait for Google when, um, Mask and Grok are actually gonna let go, uh, too. I'm sure they're working on it too. So we're about to get some updates in this direction.
Yeah, the subject is very fun. And as you say, keep an eye on the news and be with us. Every Sunday we're on ToTheMoon. Technology news and sites from the Silicon Valley.