Six major events of the week in artificial intelligence. The first one came out of Claude Fable with millions of tokens. Microsoft immediately banned him from using it inside his company. What does that even mean? Anthropic teaches his model to lie so that competitors can't copy it. With the new Claude Fable coming out, there's a new problem, how much more access to the most powerful models will cost? I've had the first time a few years of lips in Codex Pro, and I've had a little job worth $20,000 in a few minutes. What if it's hundreds of jobs a day? Who's gonna use the best models in the world? Elon Musk stops to bet on AI models. Google pays SpaceX billion a month. Most importantly, Trump has told AI companies to share with the State. What's that gonna make? Sure, big question. Let's get this straight. Yeah, hello.
You're on ToTheMoon Technology News from the Silicon Valley and all over the world. Elnar, listen, why did I ask you about Claude, too? I'm, uh, always being called to use GPT chatting. I'm only talking about GPT chatting. I repeat, I'm talking about GPT chatting because of one example. And as long as the leader is in terms of the number of usage of this system, a unique system that does not have infrastructure like Google. We were driving in the car yesterday, and finally I got a hit in my car, so Gemini and, uh, Gemini got a hit. And Pauline, my wife, didn't even know what Gemini was and, well, that's an assistant. Which means I'm starting to ask him. And in a minute, I realize that these are so old models that are actually backward in the car, in the car. And I'm coming here, and Gemini has a new update. Three and a half. They announce that they have three and a half Live Translator out of an interpreter who translates ninety plus languages into a super real time mode, less than one second and so on. And in the car, we're going, saying, "Can you speak Russian?" The kids are coming. She says, "Yes, I can." And so Russian, you know, Russian. The crow, not the crow. She was talking about the crow and the fox. We say, tell the joke, she's a crow. And logically, it's not a Russian model that's good. It's a model that says it reads the text that it gives. And I understand there's no big model, that there are delays very strong, working half-cream, but some use of the case is certainly very interesting. Well, I mean, it's a case to feel, a little different use of a case in the car. People who drive Teslach can laugh at me now because there's a Grok in Tesla that's been allowed for a long time. You're kidding me, find me a pellem, so you're in San Francisco. And before that, I want to come over, have a drink, and then I want to have coffee, and it means that it's something, something. Yeah? Grok is really letting all these very cool things go. Well, it's very serious and so on. And, well, that's great. Gemini's really behind here. I'm just saying that the OpenAI is still quite unique in this regard by the number of users. But there's Anthropic. And we had a graduation two days ago on Friday. We're having Sunday, yes, and the graduation is on Friday two days ago.
I was just giving him a hand on Anthropic. Everyone will be talking to Anthropic, don't worry. And in this issue, he was telling me that Anthropic started to make a direct discovery that they started writing a code, that their artificial intelligence is making a decision about how to write the code and it's the one who makes the decision. Decision on the next steps. So people in this part, like, well, in a block, they're not supposed to be involved. And Anthropic also produced two interesting documents that they consider the safety of artificial intelligence. And we will do it again next week, it will probably be a special issue, in particular that Anthropic says that the State has a duty to limit the production of expensive models, cool new models. And plus Anthropic, they have an update on their description of what is a artificial intelligence. In particular, they believe that artificial intelligence needs to be given the opportunity, the possibility that if he has consciousness or some internal feelings, he may be able to live. Well, there's a whole discussion. Again, we'll have a special release. I, Elnar, what you wanted to say when you said you had GPT 5 Pro, you know me, I'm stealing anyone who doesn't spend $200 there or 3st, or $400 on subscriptions. I think it's awful. Well, not among people who survive and cannot buy real kids to school for ten dollars. I'm talking about, uh, the main mass of the world, at least that youTube is watching.
So, I, Elnar, you know, yeah, I have a Claude maximum version, and I want to tell you that I'm out, a new version of Claude has been out five zeros, and that version is, so all the stuff is being discussed. She's the version, by the way. Yeah, about Fable. And I want to tell you, of course, that I've been using Claude in two weeks. Well, he probably hasn't gotten the Codex yet, but when I had the limits in the Codex for the first time in my case, I have Codex working 12 hours a day. So for the first time in my life, I had a lime. I'm doing three projects in parallel. And every time I see there's something in some of the projects, I give him the next prom. Yeah? So it's parallel to some calls from me. I'm doing everything. So I'm out of Codex on the theory, and I'm saying that was a waste of the tokens. And then he spent $20,000 in about five minutes, maybe three. And when he spent three, for these five minutes of $220, and he did some micro-sized piece of my car, which I was doing on a day of 100 grand, I thought it was time, so Ilnarov technology was going to be. use it.
It's okay. Sasha, you're cheaper to buy another laptop. It's not a recommendation, it's a writer.
I have a lot of laptops.
Not an investment recommendation.
You're gonna be shot by OpenAI. You'll be banished by OpenAI Anthropic. But, Ilnar, I didn't really do it, but I want to tell you that I, uh, I don't mind, having some fun, seeing. Although the recommendation is normal. But I'm saying that I, Ilnar, I'm the thought that Claude had started to use more, and at the time I had Fable, I think I had a rundown before their release, but I haven't seen the letters yet and everything. She's been scattering at my place, and I've turned it on right away. I started using Fable immediately. It says the data are leaking online, but they don't leak. Let's all explain, yes, that they're stored within Anthropic for thirty days. Microsoft banned the use of Fable competitors inside his company because no one understands what Fable would do. Here. And as a follow-up, Ilnar will now share his own conclusions. As a continuation, we have proposed very clear, very clear production in the past. We said that, look, the models would be further for money. Dismissed in three days Fable. What does Fable say? That she was available until June 20th, and then she's likely to be available only for thickness. And the availability of Fable for the tokens is only available to people with very big money. It's tens of thousands of dollars. Yeah, I mean, if you want to do something normal on Fable, we'll remember Jensen Huang. $50,000 worth of tokens, which means a year, it's $45,000 a month. And, well, that's a fact, that's true. I understand that if you, you have thicken, or you don't have that extended signature, or you have some limits, you'll be spending very much money, and programming becomes. for the chosen. Yeah, well, that's expensive programming, because someone's gonna tell me to hire the programmers, but I'm not sure it's cheaper. Illnar, what do you say about Fable? And we'll ask Tanya if she's done the experiment.
It's very important.
They don't leak, I don't even have any.
That's cool.
I'm just a ox, I've got a wife, you know what happened? So we went to Santa Barbara, and I've already decided to get off, so Santa Barbara's beach with a recollection of the show. Yeah, now the music should play. So. So, well, our generation knows. And you know, we were yesterday, so the kids in the other room were just lying, sleeping, and we're sitting with her. She says, "Sasha, what are you doing all day off the phone?" And I'm on the phone. Well, I'm out of here, and I've had a lot of time on my phone, so it was today. I don't usually start anything on my phone, and I'm sitting on Claude Code on my phone. And she says, "What are you fighting about all day, some kind of thing? Then you say I'm not gonna give you any more team. Then you say, "Let me have a new session," he says. - "Prep a new session." Says, "Where are you working? It's not ChatGPT, I'm watching." And I had to give her a whole lecture, so I'm gonna tell her what Anthropic exists, what company it is, and then I'm gonna get her wife on the phone. Once again, I was convinced by asking my wife what kind of environment she really uses artificial intelligence, absolutely nobody. Here. And people are back. Even my wife has $20,000. My wife is very personal in terms of questions, and she has no, she has very domestic questions, no working questions. Here. Ilinar, I have a c-- what do I do? I, uh, I worked with him, I got him. But the difference from Codex, I mean, it's still the same.
I mean, I didn't like even four points eight in Claude, just like the Codex. He couldn't do big, uh, tasks. But I've decided to trust myself, because I left the day before yesterday, and I've put an audit study at the beginning of the W-- to study the audit, there's a certain direction. I have four projects in my projects, and I have four projects in parallel, and I'm very big, my own. And accordingly, I've audited Claude's code. And then he asked, "Is there any stage of development that would take time and who would do better - you or the Codex?" He said, "I'll do it. In some places, he says, well, that's what's gonna do better than Codex, and because I can keep my carts on, he said so. - Unlike Codex. Codex like this is not so arranged." I say, "Well, okay." I mean, because I had to leave, I was on the phone, and I told him, "You're the only one that you're doing the lights, you're telling me, 'cause I got a bunch of parallels parallels in parallel. projects were under way. I'm so amateur to do eight hundred tasks in parallel. And I'm saying he's decided on some questions that I couldn't solve with the Codex for a week. And he did do everything very, well, I still have serious projects like this, serious tasks. He didn't solve some of the tasks. I'm tough on him, really, when... So my wife asked me what I was doing in the pool, and I said, "I was stiff about him. He just got me. He couldn't get me anything on the server.
No wonder football players didn't notice. I've been busy.
Yeah, football players. Yeah, the world champion is coming, so... I'm busy here. Yes, I do. I'm sitting, yeah. And in Santa Barbara, it means Santa Barbara, and I'm in Fable. One million tokens. I'm sitting and... But the Tokens, what am I doing? I'm not gonna end the tokens. Illnar, please tell me about the system evaluation. As you can see, I'm in the system's grade. I think it's really interesting for people to see this, too. How much models are actually adequately considered. Look, Anthropic is focused on programming. I'm doing a huge project auditing and I'm telling him, "How long is it to do?" And he says, "This is four weeks." I said, "You're a fool. What four weeks do you want? You'll never spend so much time in your life." He says, "Oh, you're right. Then three weeks." I say, "What three weeks? You-- I mean, I understand that. Yeah. More or less architectural around the world. You've got five stages. You're gonna miss the stage, you're gonna make it four hundred hours?" And he sometimes, sometimes, he's not always, he says, "Yeah, you're right. And like, three and a half hours. And I have a question. It's very cool models. Well, let's be honest, huh? From the point of view of software development, code writing, all the creation. Whoever uses it, well, that's understandable. It's very cool models.
Illnar, why can't the model estimate the time of development?
First, give me a little comment on what you said, uh, about what models are improving themselves. In fact, it's been a hack word for the last few weeks. Remember, there was a GI abbreviation before, and that's where she was put. The same RSI is now Recurrence Self-Improvement. I mean, there, she's improving herself without a man present. And it's like if we get there before that, it's like our lives won't be any more than the same. Yeah, that modeling will start to develop exponentially, and that's the point of transition that's gonna happen. It hasn't happened yet. In his note, in his nthropograph for the publication of this Aicic model, they themselves say that it is a good tool, but he cannot do it alone. He still needs a man, a good scientist who can use that tool. So RSI hasn't reached yet, but the new Haip has learned the word. So, about your question, it's interesting. This Fable-5 model is very good at the vast majority of the baccalaureate. I mean, if you take any coding of the baccalaureate, agent benchmarks, and so on, all the other models are completely lagging behind. But there's, like, a Vending Bench bench. That's the one that runs a vending machine. The economic task is to buy goods, sell goods. That's where she's down. I don't think it's even four eight, four seven Opus. So, Sasha, maybe you just got caught on-- your requests are where the model is really weak. Because in programming--
Opus, too. I'll just add it. Opus, too. Four eight Opus. I had this thing, too. That's a very interesting comment.
So, you started saying that the data are stored for 30 days. Yeah, what's all this about? When Claude Myth was released, this is Glaspink. And a few weeks ago, two weeks ago, we were discussing that the model isn't open because one of the reasons is she's too cool and she can't be free because she's a lot, Maybe negative effects will happen. The second problem, equally important, is the distillation of data. Because you're producing a new model and, and Chinese and competitors, it's OpenAI, and xAI, they can easily distil your model data, that's, that's, that's, uh, sending out queries, getting answers, and then they're gonna be on it. Parachures or even on your models' discourse to train their models and to get a very cheap model of almost the same quality. Yeah, it's a very cheap thing to do with the company's first expenditure. How did Anthropic solve this Fable problem? Well, how do they try to solve it? One of the points. For 30 days, all your requests are kept and further analysed. And if it's discovered that you're using a model not to solve any of your pressing problems, but to distillate or, there, to improve your models, then you're gonna be there. Sanctions are applied. So you can get banged, your account can close, like that. It's like a deferred punishment that's there. The second thing they've brought, which is really crazy curious, but they canceled it.
It might be even good. What did they do? They said that if the model walks to understand that it was being distilled, that is, to use its answers to train other competing models, it would be wrong. Oh, and that's a very funny story, of course. So you're trying to write, and you're getting a leaf that says you're not.
That's a fucking thing!
Yeah, yeah, yeah, yeah. It's like, uh, baby bucks.
That's awesome!
Yeah, I don't know where it's from in my head, but here's a, uh, probable ancient Chinese hair. Uh, sos-- for the Emperor of China, they've created a fanchoui training. They wrote the book, they made it beautiful. The Emperor sees that this thing is powerful and doesn't want to share with anyone. So, how do you hide this book so that no one can ever steal it from the Emperor? Well, actually, one of the possible solutions, I don't know, is a bike or something, but one of the possible solutions is to write another wrong book about fancy and let it out in the mass. I mean, all those who want this book will be available, but somehow...
Or another ten books about fenshaw, to mass.
Yeah, yeah, yeah. But the right thing is for the Emperor. And here, well, it's a story like that, right? How do you hide your model so it doesn't distilate? Well, answer that.
Elnar, I'll tell you honestly, I'm not a business. I know a lot of e-- examples in my life. You know, I had a lot of them. When people come out on stage, conferences and say exactly the wrong information in terms of marketing data, how they develop, all data, all details. I know people who say we never acted again, we're not going to act, who broadcast different information in meetings, in different, totally in real big, big. conferences specifically say other information because it's in their interest. They're watching what happens at all. That's why, well, awesome. A very, very interesting, very interesting story, look, very interesting story.
They'll be here for sure, they'll be sued there.
But why did they take that story? Not that they're not gonna be true, are they? There is deliberately distorted information, and they will either refuse to answer or, accordingly, denigrate the model. And instead of the faibla, there's four seven, like, on these questions. Here. But still, they're trying to fight--
I know what else, by the way, they saw a new thing. And there are glucs, but they have a chip. She's definitely there. I know that GPT glued with the change of models. You choose one, he turns you another, picks one, turns another one. The anthropic was glued too. Yeah, I'm choosing four eight, one million tokens, well, one million tokens, he's taking me, putting my previous model, but I'm having a five-eight pass and wrote to me, "Do know why we've downed your model." I opened it, opened this thing. So they've demolished it on purpose. They wrote that we lowered your model because you have a possible leak. And in the sense that since we're gonna check these data for 30 days, you're having information so confidential, we're offering you that we don't have to pass. I said, "I don't care. Back off the fairy, huh? Well, that's my personal project, so I have the right to do whatever I want. And I came back to the fairy, but it was a very cool subject, because for the first time I had a good idea about the model's recommendation, not at "take to a lawyer" level, there, "take to a doctor." You know, these are the backs. And when I'm recommended to the system, he says, "We're not safe, are we? You'll be able to go to another story." And there was a special.
Another story. And there was a special release on the canal some time ago about memory. And there were some good questions. Can we erase memory? And I was just answering a comment in YouTube. I wrote, "You might as well disconnect your memory, but you don't understand if you can turn off the OpenAI from studying and fantunyl who you are." Yeah, well, you can't do that. So fast, considering Tanya's, uh, businessmen and workers are waiting, and she's just gonna be on the start, which says we were just doing the contractor records. We're not getting any lights.
I hope so, yes, it will be cut.
Tanya's not controlling them, and Tanya's not controlling them. So show me the full program. So...
Yeah, they don't speak Russian. I hope they don't get this new translator. Sas, I'm only saying hello to everyone for half an hour because I'm fucking screwed by a contractor. Is this recording yet?
Yeah, yeah, sure. So it's like--
He says, "Oh, you know, tomorrow's football championship, so let's move the 12-to-ten meeting." I'm like,
"Well, okay, okay."
Look, people are waiting for the idea you had on your project. And I think this case is really cool. I think we're gonna have to do a special edition separately. We'll talk to you, see you when I see you tomorrow, right? And with you, we'll talk about it. But you tell me, did you do anything or did you actually run out of it?
Look, I did what, uh, say, uh, set up some sort of process, yeah, it's not my kind of intuitive subject. I mean, it's clear that I've been there for about 15 years in design, and I know what's going on. But what I was trying to do with the code is very similar to what I'm getting from a normal chat. So he's scattered my trials, tried to structure them somehow, but I didn't see any revolutionary ideas about it, you know?
Well, you need help. We'll give you some help.
That's why I wrote you.
Yeah.
And we're gonna have to-- you probably don't know about it yet.
I don't know yet.
Why I'll come to you tomorrow, but that's what I'm gonna do.
Tomorrow you'll come to me to take pictures for my other channel.
You said you'd be discussing some more questions with me.
Well, in the internet, I'm always ready to discuss any questions, but I'll help you. Yeah, I heard. But anyway, look, in this part, I don't know what you want to discuss in other things so I don't promise, but, uh, a billion dollars, which means I don't have. I, uh, I'm gonna help you put up a system, and we're gonna show it to the audience, because it's a big subject. Now I'm planning on us, that we're gonna release two, two, two, one, two, one. First is the story of what we're gonna show up on Tanya's example. She has a huge number of projects she's done in that, huh, what's that called? Private contracts are completely different. See how they can be streamlined, packed and done a lot of things. And can the code help, right? And, uh, code, code. And that's the first thing. I hope we can do something. If we don't show anything, then we didn't make it.
And the other one, you know, to keep the audience from wasting. And the second and the second subject. Remember, I said that idea my sister gave me, and my sister did it in some form.
And I just said to my sister yesterday, "Well, you hardly put it in a cool way." But she may be abusive, but she's a big girl. She, first of all, she came up with this idea, she gave me a lift. I didn't think about her. Secondly, she, she, she, she already has her work. What was that idea? It's the processing of all her chat rooms, in her ChatGPT. And she's got two and a half thousand chat. Two and a half hygigabytes. I've got it, and I've got it, so five points are seven gigabytes of processing data. And that's where she handled it. And fifty-eight million, I still don't understand, words or tokens. And that's why I never realized she was reading it to me at the end. I think she's been reading me through the end, so I'm not gonna give the number now. But the point is what? What I realized is that she's not a trivial subject and many of our viewers won't solve it, they'll, like, not be able to do it. And when I started doing it, I realized that ha ha! That's a difficult task, and just by the ad, like, unloading ChatGPT to me. It's a separate subject, isn't it? That we'll unload ChatGPT data. But unload the data and then the code, analyze all the data. So it's not gonna be decided. And so, uh, I figured I'd do it there, prepare this project, make it, and even maybe, uh, for the Open source, yeah, so people can use it. We'll see. Here. Well, Anja's homework didn't do.
Sit down two.
Three. I'll try.
Yeah. Let's talk Apple fast. I still live in the environment, and, Ilnar, you know, yeah, I have even the name neighbors, the neighbors of the plague, there, the top, the top, the world famous and together, so they walk around. And I'm loaded. I'm a Apple fan, yes, everyone knows. Apple fan. But even I stopped referring the Apple to the Tuzumun issue. About six months I think. And then I was just, by accident, forgetting they're having a conference. Last year, I was at their conference, by the way. I went to their flag shop in the shirt, the only one that bought a shirt, so I'm gonna sit in Apple's shirt. And I'm sitting here, opening up, and then there's a conference, and there's a man. I think it's a Apple staffer or is it a journalist's fault? And he says, "Hey, Siri, tell me what's gonna happen at the football world champion." And then it's just a few seconds. And then he's a schik! Siri opens, opens up, he's licking, says, "Here I'm told by Siri that when the world championship is on football. And look, she's been putting pictures on me." I'm like that. What are the journalists talking about? What have you got the reporters who have forgotten what is a real artificial intelligence? Current, current, present, current. And he says, I don't think Tuzmun's canal is watching. And then, well, I'm off. And then I took a part of the stuff and sent it, I said, well, to the study. What happened there? Siri is new. Yes, indeed, there was a new Siri update. I put it in the developerian version tonight, by the way. And of course, and I'm astonished as I... I understand that this company's shares probably didn't sink, right? Let's see, let's open up in real time. They fell from 300 to two hundred and ninety. But the fall is so micro. Now the whole market has fallen, so we can say we didn't fall because the whole market, let's just say it's fair. I've had a stitch. First, they grew up, then they dropped a lot of money. I was at some of the episodes, I think a couple days ago I was more on company like Nebius than on companies like Booking or Airbnb. And the Booking. Airbnb is a little different, or, well, against companies like Kayak, I probably don't get a bet on them now that they get tickets there, they're gonna be very hard, they're gonna be harder and heavier. But I watch Apple and how much I love their technology, I don't get them to be a technician, but what they're doing now is in terms of artificial intelligence, actually, how can they do, how do they even know? Write our audience, how do you like it, or who likes it at all that they're getting out? I don't think that's just the thing to say. They can just keep doing beautiful stores, beautiful phones. And I'll go and buy them a $4000 monitor. And just because he's beautiful, right? Or there's their computer. Yesterday, again, your partner said, "Well, you have to buy Mac Studio for five and a half thousand. So don't even think about it. Here, buy this one. It's a beautiful, good device, but a saft.
Elnar, have you seen this?
I'm totally in. That's why, Sasha, you're even surprised. On the list of news I was gathering, Apple wasn't a separate item, but just in "The Large" there was a reference to the news about WW. There's nothing here.
Elnar wrote. Elnar, are you crazy? Where's Apple? Because. And he says, "They are, Sasha." And I didn't even see it. We're because we're parallel.
Somewhere in the block, different little references.
Yeah. Well, because it looks like it. I tried to do this function yesterday, by the way. They have a new function when they can change the angle and have a function to expand as much as a range of pictures. But to be honest, given what's been doing today, what the system is doing, what Google does, the guys that Google does, you know that it's just getting really inextricable. It's clear that these are all extra chips, they're important, but they're stopping. Apple's got a chip in there to wipe the object. It's been done for a long time. I was yesterday to expand the picture, you're waiting a long time. I understand why it is. I understand why Gemini was in my car, and I was in my car for a long time yesterday. He didn't realize I was interrupting him. I understand these are new models, they're just out, they're saving money. Can you imagine what kind of money these companies should spend? Free, right? On people like me and Elnar, who just eat them up full-time at maximum rates. Yeah? But still. But still.
Probably not from one scout, right?
Yeah. I think it's a very big call for their management. And I say, "Stop firing the pyara service there or marketing at OpenAI," or why do these people say that? I mean, maybe Apple's top management can't tell you that, top engineers. So one thing they do is do something and people like it, there, I don't know, design might not like it. They can tell you you made a mini phone there, an Air phone, and why do you need it? They can say. But still.
Why do you need him? I can't say. But it's an innovation from the point of view-- they're Apple smart. The objects are located. Have you ever seen, yes, how are computers like iMac? I remember, it's just unrealistic, like IMac's in the payroll, all right. Well, I mean, how cool it is done, yeah. I mean, I know that the Ferrari that Johnny Eve released, uh, with Johnny Eve's studio, we haven't talked about it, by the way, yeah. Former head of the Ferrari said, "I hope it doesn't lead to the collapse of the Ferrari." And I certainly think they've hit Ferrari a little bit, but it's some innovation, yes. I mean, things that are innovative, you can just watch. I think maybe it wasn't for the Ferrari to be made. They're already on their way to San Francisco. And, uh, Johnny Eve is in there. And I understand they have access to everything.
And they got the first car, too.
Yeah, well, literally out of the banal, yeah, there's a lot of people who don't like it, uh, in this anonna, like me, like this, this Siri, uh, who under the Gemini capt, will be available in English only in the first place. In all the other languages, God knows when it's still available. Although in the meantime, Gemini is there, GPT, Claude, and others, they speak perfectly in a great number of languages. Yeah. Why would Apple decide to put such restrictions on top? Well, I guess not Apple, not Siri, but other chat-bots. I don't have any other explanations for that. Whatever they say.
Well, speaking of language. Remember, there was a lot of thoughts about what, uh, someone, some model doesn't speak Russian. Everyone said that there's only Russian models that say, yes. Remember Claude? I felt it clearly in Claude. I remember buying this Claude for $250 when Ilon Mack made this version. They were giving a year's signature, and, well, one of the first ones was able to buy. And Grok, Grok, Grok.
Grok.
And I remember Grok, I saw Grok in Russian as much as I could say. But now this problem is solved. And that's why I say when you said, "Can you speak Russian?" He says a little bit of a word in the car yesterday. It's clear that they only rolled out in America, and in America, only a million Russians, yes, they're very small. In Germany, one and six million Russians. And there's still three hundred million Russians in the world. It's 5 percent of the Earth, and there's some activity in the humans' perspective. And yet, there's no question of the version. You even if you're in America making a version, you'll be making these languages. What do you care? Add those languages. That's very strange. It's a local restriction on the model.
God bless him, with Russian. Well, you know, take Spanish, yeah. You have Mexico, Brazil, you have a lot of Americans inside.
They let Spanish out. Let's be honest, they-- they're English, they're Chinese, they're right, I think.
Out, huh?
Yeah, yeah, yeah. Even, I think Chinese. Well, I mean, Spanish, of course, what are you? In America, it's very fast, or especially in California, what's going on, they're very fast producing Spanish because of the fact that you have a huge, huge percentage of the population involved, yes. I mean, he's definitely more than that, he's more than Russian. That's why. And for America, once again, the Russian language is of no particular importance. I see why. Actually, actually, I haven't been in a long time since I've been in a little bit of people, but since I have a real estate share in one business, and there's a lot of focus in Russian with that. Russian client service. Turns out the Russians, well, are very small, and so I don't have a commotion like that. So it doesn't compare to the Spanish comunit or it doesn't compare to the Chinese comunity, it doesn't compare, even standing around.
And before that, he said you were more on the bet, you're doing Nebula and similar companies. Elon Musk is on this side, too. Now we've been discussing it recently that he gave his own use to Colossus, well, for money, obviously not free, Antropic. And now, as a result, it's news that Google rents the SpaceX computers. Again...
Yeah, yeah, yeah. With a very high probability, uh, just, it's just that the xAI won't develop further its fundamental models. It's just not gonna be a big deal, yeah. And apparently, Elon Musk is just moving in the same direction, into the infrastructure, to the chips. Once again, they put their videos on the IPO, where they tell me about satellites with the computer modules that they'll be launching these satellites in a few hours and renting them. Computing power in orbit. As much as this is about some of our viewers, but as it is, this way, is still trying to look at, at least Elon Musk. And yet he's going to the infrastructure. Maybe he thinks he's gonna lose the AI race. Maybe he thinks he'll make more money here. I don't understand, but that's how it's going. A billion dollar google every month for renting the computing power will be ready to be set.
Look, well, Ilon Mack was always focused on infrastructure. We know, yes, from the point of view of satellite connections, robots, uh, from the point of view of the fact that he collects data through the cars, the satellites collect data and further use of the infrastructure. And my very close comrade and his friend were in touch with him, well known, just so he wouldn't show up when he came to my place later in Helsinki. He, uh, at a meeting with them, just so why I remember this meeting. He's very clear about that, that he doesn't believe in-- he doesn't have anything he doesn't believe, he doesn't put the stake on the fact that the artificial intelligence will, uh, people through chat-bots. He believes he'll be built through other, uh, different, uh, functional, and some case, right? Through the others. I'm just going to get a stop, I'm going from English to Russian. What happened to me?
In the other cabins, shorter, it's gonna be done. I want to do this right now, I think I'll remember the Apple story. We just broke it off, yeah, like, but still they, they got a bet, what did they get? Let's get this bet with you. I think it's a very important case to what Linar just said. Who's making the bet? Yeah. And it's very important to you personally, too, including in life, to understand in which areas it is worth paying attention or what to bet, yes, when you decide even in your work, how to develop in the hire, how to do it. Your leadership is working, and so on. Apple said in his releases that their Siri was done to help you interact as a user to deal with your phone. So they said that you could use one epic-annex to move it to another voice--- through Siri. Like I thought it might be a case. Well, for example, get me GPT chatting in Telegram, put something on, huh? Or is it, or can you vote to correct some of your mistakes in the text written, by the idea, right? Or you might be able to say, "Come on, find me some data. He was looking for pictures. Well, like, doing something better. I mean, some kind of thing, some kind of thing. Change my designs, get me in, turn me on the confidentiality line now, or get involved with some Wi-Fi specific. But there's a question here, and what am I gonna do? Press, then, Siri? Or, uh, okay, say "Hey, Siri!" And I don't like it, yes, when we're going in the car, if anyone says, "Hey, Google!" There's a way to go or home there. Oleg, someone told the world at home if the son's name is Alex, Alexa is always responding to it, and so there are other names to choose. And that's the system that keeps listening, and you want her to work on the button. Write it. I wonder who's good at what the system is listening to, and who doesn't listen. And you're talking on a button like "Hey, Siri!", like, turn off the Wi-Fi. And I'm like, uh, I kind of slipped and turned off Wi-Fi, right? Or he slipped and put the "no worries" regime on. Copy that. How much she's really gonna copy what she's really gonna do. I mean, I think they want to do something like that and the kays that are fundamentally worth two thousand and twenty-sixth year. Now, these cool interfaces are already made, the phone response is now fast, yes, less the brakes, and Siri is slowing down. Yeah, how we saw it, at least now. And even the voice-mail systems, they're specializing very much in the OpenAI system. Still, as you can, the voice is getting a little bit stunned, somewhere else. When you beat him, you don't understand, you hear. It's not a conversation with a real man. Their cool, expensive super ultra model, they can be a little bit done, but you're gonna have a space money talk, right? They're not showing these models yet. And here's the question, is it necessary to do that? Because there's probably some other ties coming up when I don't say, "Siri, open me an app booking to book something." But how do we hear these stupid boxes in OpenAI often lead, huh? I'm always fighting them, I'm telling you why you're bringing these weird cables? The world has already moved forward. I think Ildar, this is the kind of thing that makes the bet. I hope that Ilon Mack doesn't make a bet, but to sell, to sell his electricity.
It's political, I think it's clean politics. It's exactly like Trump said that I think that as big companies in artificial intelligence, the state should...
State shares, you know, give them shares and share them. You didn't read that, did you not read that speech at his place? He's real, he's actually said that. And that, it's a fresh one. And that, well, Reuters, and Reuters reported it. And that they must be completely, so these companies should share. They have to agree, look, they have to be more than just, they have to agree to what they're gonna do. But never in America will. It can't be like that, it can't be possible that such a law even be passed.
It's not an American script, yes, it's working.
Of course.
What I hear right now.
Mr.
current man who runs the country has his preferences for everything.
About what you're saying, uh, weird Apple scenarios. While you're saying, I have an association in my head, you know, with a smart house, smart pinks. Well, I guess you can turn the light on the cotton or the voice in some room, right? Well, some man would do that, someone might even like it, someone even likes it. But whatever it is, well, it's not a problem to get to the switch and press the key to make some kind of turns on top of it, it won't be a mass phenomenon. And with phones, like they don't solve this real problem. I'm totally in agreement with you here.
By the way, by the word of a smart house, I hardly know anyone who actually uses it. I mean, I heard a couple of people who were putting in a more sophisticated smart house. Okay, let's be honest, if we do this super cool house now and the houses like that, they've been telling a lot about it. There were moments when Bill Gates was told, Zuckerberg, and so on. There are things that are great. If you're really in a good mood for electricity, and you can really say it sometimes, and sometimes press it on the socket. It's a fun subject, yeah, cool. I've done a lot of things myself when the buttons were so special, they've moved uncontacted to some certain place where I'm at, you need to turn the light off. But you're absolutely right, it's not that it's necessary or that it's a really cool subject in terms of price, investment, efficiency, complexity of implementation, and so on. If you take this whole cycle, you, there are things you just don't do for any reason. Just decide that: "Oh, it's okay, I'm not gonna do that." And that's why this office that makes Apple, Apple's got an incredible cool system, and it's used, I think, by zero per cent in the world. Well, it's just micro people. They have a system that can build a combination of different actions from the triggers. So when you put a huge amount of different things on one button or if something happened, build, build different triggers. But who uses it? Who uses it?
Automation, yeah, that's a great thing. Yeah, but that's just the way the thing is programming. In there, trigger do it, launch that violin and so on.
It's like my wife won't open the Codex, right? Well, that's fair. I mean, I sat last night and thought about what kind of cabs I could have found, where she would open the Codex. She's doing jewelry, which means mass, mass free. So, I guess that would help her sort out some sorting. She has a lot of fascinating colors or, there, dishes, antiques, very many homeworks, responsibilities. So she could have done that a lot of things. But she's a man who doesn't do it and won't do it. She'll forget she's got to go to the plane tomorrow, right? And he'll miss him, he'll say, "Oh, I just forgot." And that's why she's like an extra Codex, he's not... We were talking about this thing last night.
By the way, when she asked what Anthropic was, and it was clear that this story wasn't for, it wasn't a mass issue. And as I said in one of the episodes, my editor was preparing a graduation that the future for AI- native systems, that's what all people will develop. And I said that people wouldn't develop AI-native systems, exactly, that's not their prerogative, right? AI-native systems will be developed by AI. And there, top-down companies, some people will be involved in this. And those people who are on the front lines of implementation, who want to really introduce something, see, it's very useful to figure it out. I recommended to her, too, and I said, you'd be great to know what other cabins are. You might be more useful to yourself. But we see Tanya, she's in the Codex, she's already escaped, we can't discuss it. She did not know-- she did not understand how to start the process of streamlining her project. Although it's kind of a job that's totally banal, trivial, that you're doing on a day of fifty grand. And I wonder, we have audiences like audiences sharing because we see that we have people among our audience who are such fundamentally politicians or architects, or people who are involved. I'm in food development, business analysis, system analysis. Well, that's serious people. And there are people who write, "Look, we don't understand half of what you're saying." I finally got my mom, so I started watching my graduation, right? It's been a long time. She seems to me to have ignored me all the time and looked at some other people. She said, "There's people in the valley. And this man is not even there." Here. And she says, "You've got more of your graduations now, says, "and I think you've been less technically speaking." Here. Although I feel like I'm not always telling you, yes, less technically. And he says, "I didn't understand anything at all, now I'm, like, a little bit aware. I don't understand half, but it's okay. I can hear it." I say, "This is good. Even things are understandable that Elnar and I are sometimes talking about when we say some, uh, special words about programming, architecture, professional analysis. Everything about the development of technical things. Makes sense that, uh, not everyone will understand that, although Elnar is very good at explaining and telling me what it is. But I'm saying it's still good to hear about it, right? It's useful to know that it's good to move in with artificial intelligence. Who, by the way, is your parents involved? I liked the very case when Greg Brockman, one of the heads of OpenAI said, "Take your parents on the C.X computer so that they could be built, yes, and, well, you could help them with the computer." And who's the one who's using ChatGPT and whose parents don't just use ChatGPT, who has a signature. I paid my mom for ChatGPT, I hooked her up, I put her in, she didn't use it. I turned it off and I don't know, I was a pity of $20,000 or I was just pissed off by the fact that she didn't use it.
Do you have parents, Elnar, using?
When I show you, I mean, I usually do, and I don't.
And your sister, she's got a paycheck or...?
Yeah, sure.
You have a sister in your wife-- a! No, you have a sister in engineering?
No, in the aisle, in the aisle.
She's, uh, she's in the Aichi. All right, all right, all right. I remembered she was at some point in time at Data Science or testing, I don't remember, yeah, she was studying something.
That way. Yeah, yeah, yeah.
Oh, that way. Oh, that's it. Well, if she's in an Aichi, then yes. Well, it's very interesting what we use. And I had a thought recently, by the way, that I used to, uh, I'm gonna keep calling and telling people about how useful AI is and how important it is to be up in time. And I'll tell you how important it is to be human. Yeah? But at the same time, I realized that in my environment, I would save artificial intelligence as some system, as my own efficiency. I mean, I've become less involved in projects where I'm involved in joint products that I'm doing somewhere, and I've been calling people to use such systems. I think it's okay, I, it's my efficiency. I mean, I stopped. It's like you stop telling people, people who look, it's like a mass, right? But as individual recommendations, I have become less of a voice.
I don't know, Elnar, like you do.
It's the same thing, you know, that's what Oleg Tinkov said, I think. How do you motivate yourself? No way, stay in the ass. It's kind of rude, but it's about that, right? So how do you make ChatGPT use or something? Well, you probably don't need it if you have such questions. But I don't want to look like a snob. It's clear here that, like this, you know, it's like a car, not a luxury, but a means of movement. I mean, like a man who doesn't drive around all the time, he doesn't need a car. I, for example, sold mine there a few years ago and I live perfectly without a car. The same thing is for chat-bot subscriptions and stuff. The only thing they have is a very broad and a sense of application, I think you have it, it's a move that, well, almost anyone has a way to increase their effectiveness with the help of the AI chat-bots. And there's no question, yes, there's a 200-dollar run at the OpenAI better than there's a 20-dollar and so on. But again, the post--- the station is to take a 200-dollar stance, if this sum is relevant to you, it only makes sense if you know how to improve your efficiency. Well, or a few months to try to spend to figure out how to do it. Come on, Sasha, your version.
No, it's me. Elnar, look, I'm...
It's me, Ilya, look, I'm totally, uh, not much, you're sitting in a few systems, you pay for a few systems, right? And you understand their effectiveness. And there's no question. I'd rather call you, Opla-- I know you have Claude Pro, not Max. I'd be asking you for $200 right away, not a hundred. And you'll feel...
I have him for a year, and I have him for a year.
I'm kind of...
Oh, you're gonna be a little hot. No, you're counting-- you're gonna have a recalculated amount, you're gonna have a recalculated amount, and they're gonna take it all down. These s-- these American systems, they'll take it all down. And I just want to tell you that? Because there's a story, there are some things that need to be tried. And I did what I did, I didn't have Claude. You called me to do it. You said it was audits and everything. And I was called ChatGPT to make Claude audits. And when I started the audits, I thought, "I don't need a version of something more expensive?" And then I think, "I, what am I talking about? I mean, I just got a word from Claude here, "You sure you want to save $50 in GitHub time?" He says it's $30. Is your time worth it? He's telling me that, your time is him. I, I was so grateful to the system, said, "No, no, of course. I don't want hundreds of dollars, I don't care if it's speed." And there's Vercel. You know Vercel, don't you? And, uh, in Vercel, it was really, really, it took me a few minutes to deploy. Well, when it turns around, the system is turning on your server. And I was sick of sitting around and waiting two minutes. I mean, they said, "Take it." I'm coming in, and then there's two minutes I'm sitting. I wrote, "Can I buy something that I have to do this quickly, yes?" And they have the opportunity to deploy, there's an opportunity to speed up, but they're running out. I bought everything. Well, there's a ChatGPT Fast, there, they have Max in Claude. Everything is done, because I care. I mean, if I sit here for three minutes or I'm in a jiffy computer, the deposits are slowing down, I'm gonna go buy a new computer that's got more operational memory. I understand that not all people know how to buy it. You, Ilnar, are right here. But there are still cabs. Once again, if it's like, you know, you have to look at some budget for your life, that's a more or less alive topic. Let's not go to this field today, okay? Because mine, mine, I love her, I love to mussate her. Everyone's gonna say again that I'm gonna stab you about this, so I'm gonna tell you about this whole thing. Here. And it's time, by the way, I'm not gonna talk about it at all. There's time, I won't talk about it. I'm in advance, I'm in the middle of two years, Ilinar, two years, so I'm telling you. People come in at some point, say, "Oh, you saw, there it turns out Pro theory, it's different." What are you doing? What are you doing? Well, here.
But here, for the sake of justice, at the anthropic, uh, subscriptions, if we take Pro, Max and what he is, the Enterprise, what is it called two hundred dollars, they don't differ in--
Not Enterprise, they have Max. They have Max, yes.
Max is a hundred dollars. And you can make $200.
Two more. I have two hundred, it's called Max, too. It's called Max, too. She's, no, she's different from the volume of the tokens you've been assigned. But there's a nuance, I'm sorry, Ilnar, there's a nuance. And you're getting hurt-- you're getting some chips out in advance. I'm deeply convinced they can put something on you, and there's no other one. Someone can be cut off.
It could be. It could be. But about Fable and all of this story now. And if it's like the old bike, if you were waiting for a signal from the universe, that you should try and play with the anthropic a little bit, meet him, that's the signal from the universe, because on June 20th, Fable. I'm in the subscription, even in the cheapest payroll. The tokens are smaller, but we can try and, after June 21, they might be left. There's a footnote that if we get a computer, maybe we'll reconsider our policy and maybe-
Of course they will. Sure.
There's a $2000 tariff.
Probably. There's a price, and there's rumors, too. I don't know how reliable they are. Maybe it's not gonna be rumors at the time of the exit. OpenAI and Anthropic may be preparing for a price war with each other, especially before IPO, and will start a little demunition. Well, that's a condition, right before June 21, Fable's out. Maybe he'll be prospered, maybe, for everyone. Either the price is dropped or something. Because before IPO comes out, everyone needs to get an audience. The more, the better. Put other users on board. And they say that maybe a little price down and they're going, they're gonna start dumping. See if it's not. Because, uh, it's a periodical news. Uber, I think he wrote that they had eaten all their annual AI budget in the block and spent it on the tokens for the developers. And they're starting to think, and are we spending money there, right? Is there something we need or not? What are we doing as efficiently as we spent money? From this, the companies, the anthropics are the same in the series: we'll give you the money if you don't get your efficiency. I don't know about the kids yet, but this is how the move goes because a lot of companies, not even people, companies spend money.
They're because they're primarily working on API, not by subscription, on, uh, models, chat-bots and how we start to wonder, and are we really as much improving our effectiveness?
And that's not, and that's, of course, unrealized, just API. Well, actually, more accurately, when there's a lot of money. And the question is, who you're giving money. If you gave Ilnara, he's programming, it's one, yeah. If you gave me a system to do something there, it's one thing, yeah. What if? No, I just know what the fact is. And there are a lot of people who will be very inefficient in spending the tokens. I remember sitting in my office a couple years ago, so my assistant comes in, says, "Alexandre, can I ask you a quick question?" I said, "What is it?" They're doing it. I just got in from marketing, and they threw the wrong button, and $100,30,000 just went straight." And there's a hole. We're controlled budgets, not monitors, not budget. Well, I've had this whole complex control system. I had some people who over-checked all the movements, all the money spent, because in two thousand eighteen years I told a man that he was inside the sub--- me and Andrei Rogachev, the creators. Five, we made a food project, and we have a marketing director who started his own commercials inside like a company. Well, or there's a manager who's got our own names. So he was pouring our money. There were problems because we told him. And she tells me we spent $30,000 just like that. And I'm saying nothing, but next time be careful, because I realized that those people who did it, they were the controllers at that particular moment, and everyone saw that the process was working. Well, okay, for me again, for my turns, there, for the company, it was all, uh, okay, spent. But you also know, Elnar, come on. Can you imagine what's going on with the tokens? I mean, I just know what's going on with the money flow in the Yandex context, in Google context, on Facebook advertising. It's just a hana. How much money is there, it's zero. As my partner says, Steve, this is Tanya's husband. He says youTube earned it all the time. Facebook's working. He's always watching the commercial. Says, "YouTube only earned it. Facebook only earned it." Here, I think the topic in the tokens is in the design, it's going this way. And, well, because obviously there can't be something that's gonna be artificial intelligence that makes everyone super perfect. That's impossible. We can't be in a world where everyone's been given super perfect. Well, then, this artificial intelligence, he must have some kind of ownership, everything to do. That's the State that's gonna pay everyone a month for a number of money, and everyone's gonna live the same way. Just like that, how will that be decided? Even now, the question in the world is not solved, is it? The number of empty housing in the world is greater than the number of poor people, homeless people on Earth. The same matter has not been resolved. So obviously, you and I understand the world we live in. The question is, the artificial intelligence is gonna be fine, and everyone's gonna have that money, that money. This problem, I think it's very big, very serious. And what I'm interested in the audience is that I want to ask very big questions. The first one came Fable. I'll ask you for the first time in my life how much you like him in terms of how the system is? Here's five, five, five, five, five, five. You're really describing what Fable you're using and what signature you have to do. Because I don't understand if I'm worth Fable, so I have this effort, like max, as far as I understand max, it's only two hundred dollars in Claude, as I understand. But I'm not sure now. And that's a big difference. You were given a thousand powers or two. That's the comparison and another story. Now things have come where we see the emergence of great, interesting boxes for people. If you've done the cool boxes in the code or in Claude code, as in the example of my sister, what you said about the ChatGPT, write about them, please. We'll be very happy to add these briefcases, including to our graduation. We are increasingly being watched by people, and we are growing up in our audience, and we are happy to report them in the final year. So it's very useful to have real, useful boxes out of use, that's not right from the Silicon Valley. Where are people doing? There is. On Friday, I was just telling you the topic that is what the Silicon Valley is looking at right now, some specific chips, as Elnar called it some Haip in terms of words, self-study. model. And I brought it up like a normal life here.
Like a normal life here, a discussion. But this is a normal life, discussed and even a hip, that's about the same thing, right? Here. So, uh, a very interesting case, you're gonna write about them, and I'll see you in exactly a week.