I went into the code today and said, "What have the salesmen been doing for the last fourteen days?" He's been there for me, like, that's what they've done. He's connected to the CRM system, he's found out, he's completely checked himself, yes. People will work in the CRM system through OpenAI. The CRM system will not exist. There's some kind of single chat where people do all the things. It recommended, however, that low-cost items in a certain segment of the types of goods.
Sounds like he already knows your fears, your values and knows what you can do, as a matter of fact, manipulate that you buy the best for kids.
I've turned up a huge infrastructure. Well, as usual, I had to hire a girl-devops, yeah. The quality of the entire infrastructure and the further writing of the adminki exceeded all my human expectations.
In fact, uh, dying, well, or ends with the era of the stacks, the frills, and so on.
Hello, everyone! We're on ToTheMoon. Technological news, Silicon Valley sites around the world. Aah, we have a new edition. Today, of course, there are different topics. From the fact that Anthropic has gone through the OpenAI in corporate companies. A huge theme that deals with cybersecurity. And Google, Anthropic, OpenAI, all released separate systems that-- that allow both prevention, uh, threats and a very cool threat to make, I think, threats. Aaah, big growth, uh, self-inflicted systems. And in parallel, we see, uh, some infinite number of new laws, laws, laws. Um, there's still a lawsuit, uh, between, uh, Elon Musk and OpenAI. We'll talk about this today. What's the point I want to start? Uh, you know, I-I'm in the middle of a research with one, and as part of this study, uh, there was a question like that, uh, that agent e-commerce. AI agents will buy for us, but whose side are they on? They wrote that. And I really liked the headline. Why? Because I've been facing the next problem in the last week, a specific problem, that buying, you know, from the river-- conditional, I'm going to ChatGPT, and I'm asking him to help with some kind of purchase. It usually comes to, uh, some things I don't know much about. Like, I'm a daughter here, uh, wanted to buy a ring in-- gymnastics, and he helped me pick it up. It's so cheap, uh, well, it's a pretty serious device. And then ChatGPT asks how to hang it right. He's recommending all these mi-micro, uh, micro-strengthes. And by recommending these micro-strengths, for example, he's starting to recommend a lot more to me than this ring a few times. And I noticed that he was always suggesting, well, at some point, that he was offering such super-cool brands. Uh, and I'm asking him, "Look, are you sure you need to buy this for $80,000?
There's a hook for five, you can buy five." He says, "Well, five can't be done. He's not checking the load." I say, "How can't you verify that? It says here, it's a product that's certified." He says, "Well, I'll be 15." I said, "Why did you offer me that hook for 70 or 100 in the first place?" And, uh, that's all. I mean, I had this very much this week when he tried to sell me some straight-to-road stuff, again, especially where I don't know. Where I told him, "Go ahead," and I didn't tell him where the brand is, what details, exactly. From chatter and hooks to, uh, some supplies to the house. So he's been direct recommending to me all the things that are very expensive. And that article was dealing with the question. It was a little wider, it wasn't just, but what you're recommending is a model in terms of, uh, product-- which products to buy and where, uh-- what it uses. For example, I'm OpenAI asking for the Amazon forever. He's doing everything he can to keep me from being Amazon. Well, I mean, I'm in all the chat rooms, I've got Amazon everywhere. I wrote him the rules yesterday. I don't know if he's gonna respect him. He periodically observes the rules, not periodically. But I noticed he couldn't hear those rules. So, some other thing is pulling up, and he's offering me different sites or other systems or producers. But Amazon as a heater, he's blocking. Maybe that's right. Well, from the point of view of the logic of people's lives to give producers, show producers, it's a good defense, that's all. I agree. But at the same time, it may be some kind of internal corporate-policy installation, for example, in terms of giving, not giving, or giving, and who, uh, what benefits to give. And that's what he says by agents? What-- wonders, as audiences, you've got this going on, yes, ours, our audience. But what does he say?
What's the deal with agents in different stores now? So you go to Amazon, there's your agent, you go to Instacart, like, there's your agent. So systems have their internal agents that help you buy. And the logic that was standing? This agent who helps on the website to buy something, he's interested in what? To show you some commercial and not to tell you it's a commercial or something. And he, he wants to help you with your personalization or not, right? And this article says what we recommend to you, uh, well, there's a general recommendation to create, uh, our own agents, right? So, agents who, uh, learn what's good for you, not the platform, huh? And and in-- and this article just divided into the fact that there are agents on the website, there are agents bringing your own agent. So you come with yourself and you're being followed by an agent, yeah. And I've dealt two cases. In both cases, the Agent was not on my side. What do you say about this? She seems to me to be very interesting, she's going to be very much developed in her, and I think we should be able to figure out at least, yeah.
I know what, Sas, I wanted to add about Amazon. First you're gonna pro-- I'm asking you to find Amazon, too. He's ignoring it first. Then, when I asked for two three times, he gives me a link, I'm moving on. And it says that this product or sold out to Amazon-
Yeah, yeah.
Or there's some sort of Indian Amazon going on. And so, come on, it's okay, come on, where you want, let me get this. I can't do it anymore. I mean, it's very interesting.
I didn't know it was all like that.
Amazon itself may be closed by searching for boots, search systems from GPT, as you and I have, remember, discussing that, uh, Amazon makes his assistants, his models are connected. Well, not the fundamental models themselves are developing, but there's a probable agreement with the Anthropic, and that's where Claude can use some sort of version of Claude, and all of them are used. The others outside may not let him. We even talked about this sort of way advertising budgets go away, something else. Well, I mean, there was a big one. And there might be a problem not in GPT chat, but, uh, in OpenAI, but in that they and Amazon didn't agree that he would go to Amazon and Amazon.
Elnar, I'm just gonna put you in here, and you're gonna keep going on until you're done with the subject. I mean, you might be very right, because GPT chatting is very strange about even the references that give him the Amazon merchandise. It feels like I'm giving him a reference to a particular product, and he sometimes describes me as if it wasn't that product. What am I gonna do? I've started to give scronychots. People, by the way, we have people who don't live in America who can think about Amazon? You can buy somewhere else, but you're jealous-- you're getting used to quality incredibly, you're used to service. I ordered myself one couch, and Tanya was there to help me with it. And I'm Tanya writing how tired they are with delivery, with answers. It's all glowing. The dates are not the same, they don't answer. Why? Because you're used to Amazon's quality. You're so used to it that you only want to order, and it's limited. And, Elnar, it would be cool if GPT or all of these systems and Claude were being chatted, and XAI, so they could tell you and tell you, I'm analysing this resource. Excuse me, we see what's in there... They're, uh-- look, GPT chatting is guaranteed to see Amazon wandering him. Well, let's get this to the man, let him know. Now, Claude, Claude is building a separate system. We may be talking today. They want to translate into human language thinking, model thinking, right? So, the idea is, it should take some side of the information, somewhere in the interface, write some of those resources, so the data can be imperfect. GPT's hot, by the way, is it a little overwhelmed by the pay-off news channels, right? Two seconds. So you're putting them in a pay-off report, he's giving you the whole thing. And here, horse-- well, Amazon is a major corporation. And you said it pretty cool, by the way. It is true. I'll take this next time and even in the lipstick, and I'll make a construct that if you have a problem, you can tell me about them. That's a nice subject, by the way.
Super useful, Elnar, yes.
Yeah. As for you, you're just making a reference to a resource that is not available, but he's still responding. You know what it's like to be GPT three, five, or four, just showing up, not even four. YouTube has bloggers that say I'm the one who's putting a link to the site, and he's on that website saying something. Oh, well, the models were not able to visit resources, collect information from them, and then respond. How did it work? You have a part of the information in your address line that you're throwing out of the reference you're throwing. Like you're throwing it down there, I'm not talking about an gymnastic tool, am I? And it says you're gonna have a...
- Yeah, and it says you're gonna have it there, that's the gym ring, that's, like, that. And he can already quite salivate what could be on that page, right? I mean, not even visiting, something can give. But like you say, he's probably just not getting caught, and you're dealing with it. I can't say 100 percent, but here's the thing I wouldn't rule out. Here. And as for the fact that your agents are going, again, uh, the head of these agents isn't created by you. I mean, it's not an open source model you've been stubborn, and it's some Claude, some ChatGPT or something. If they're gonna get preferences at this level, it's gonna be a lot harder to work with. It's just like we use, uh, not much biased, yes, not much shifted models that kind of give us the answers that we're comfortable with, but at some point, it can all go to the fact that the models are gonna be there are specific things to be recommended, specific producers, specific brands and so on. And it's gonna be hard enough to fight.
Uh, as for what he's recommending to you, perhaps again, it's not about the OpenAI building, but it's about you and him generally discussing some particular price segment. Well, we're gonna get a naphantasy now. If you're talking to him, there's Louis Vuitton, Hermes, Maybach and some other stuff, maybe he's just suggesting, you know, the algorithms that are inside, saying, they always suggest in the skirts, they just can't. You're gonna recommend a lower price segment and start raising it up there because it seems more relevant, in this price segment, to communicate, right? So it might be a lot easier, which--
Elnar, I'm gonna support you, because he's not offering me expensive things. I sometimes, on the contrary, would like to, uh, have, in my projects, expensive furniture. And I--
Your husband can't hear you, Tanya! My husband can't hear.
Yeah, yeah, yeah.
To train him.
Do I know what I'm saying? I noticed this, uh, first of all, Elnar, I'm sure I'll support that he's probably doing a wider analysis on the one hand. And on the other hand, Tanya here says an interesting thing. And you know what I want to tell you? I noticed this on these sky-- these weeks. Why? I've done a lot of things. Remember, we told you, I told you in the last edition that I had a friend of the walls there, didn't I? And we're there, uh, over-moving, there, part of the garage. I weighed it, I weighed a boxer pear in there, and I weighed a lot of it, yeah, that's enough. Then I was dealing with this ring. Then I started the pool, I needed different components for him. And I mean, that, uh, uh, uh, that bike I bought a new kid there, one of them. I mean, he did recommend me some expensive items in a certain segment of the types of goods. So he sent me this fun. I remember my daughter asked me to buy that ring. I came in, opened the prices at Amazon, I see it's worth $100,20. I say, "Ivette, look, we have 16 children's playgrounds, conditionally, well, on the ground, on the mo-- on, uh, on mine. And I'm not gonna buy you another ring, there, $100, $20,000, and that's it. And then when I started it again, I'm ChatGPT saying the ring is worth three hundred. Three hundred. And when I said, "Maybe cheaper?" He said, "No cheaper on the baby." And then he told me when I told him about keeping him cheaper, he told me, "Look, that would be a little tougher, I wouldn't recommend a baby." And at this point of time, I'm like, why? He told me why. And he described a bunch of de-- there's a nuance, because the ring manufacturer himself recommends a three-fold low or four-- yeah, on his website. There's a nuance, but he's still sending me to what I thought, "Look, what's the difference? I'll buy a ring in three hundred. And he did, and he explained to me that the baby could be smashing. There's a different quality, there's different sizes of different tubes and so on, that it can fall apart, you know, when it does these exercises, where, you know, the baby gets there, it's a little bit perpendicular. I'm gonna get up, you know, get out, yeah, and so on. I mean, these are all things, they, well, really, big questions. Or we, Tanya, when I was last in your guests, I talked to Pasha. We have a buddy, he's got a construction company, you know, they're doing repairs everywhere and good quality. I'm saying, "Look, Pas, can I hang a boxer to the ceiling?" There, there, there's a lot of balls in America, right? He says, "It's easy! You're hanging on one hook, you're gonna be fine." I said, "You sure? I'm here with ChatGPT saying, "I'm gonna have to do a sword in there." He says, "Let's go, let's hang. Let me come to your coffee and we'll hang." Look, I understand why he's telling me that. She's really gonna hold on. I once put a boxer pear at my parents' house outside the city, in the basement, just to the ceiling, and I put it in, and I hung it all up. But ChatGPT says from a point of view of the professional stuff that he's already doing, "Alexandre, what Amazon says is that this hook will hold 500 kilos, doesn't mean he'll hold it because it's that. "Maybe there's a good thing." I was just genetically willing to say the word Chinese, but it's not gonna be right, is it? Because all the good goods are made in China too often, and he's gonna be a non-quality product because it was previously thought. He'll be a non-quality product, there, uncertified. And this is the product that he's certified, and you're guaranteed to have a 500 kilo. And that's what he's been putting me in front of that choice. So, Tanya, I think he's actually offering some things weird, as you say, in terms of design, he doesn't know it. But, for example, he's a bicycle to me when I asked for the lightest, he gave me the most expensive bike right away. Though I might have thought, I asked for the lightest. Look. Well, he thought, since a man asked for the lightest bike, he knew the lightest bicycles they were the most expensive. Well, that's how it works, so it's a bicycle.
And there's number one as a cyclist parameter.
Sounds like Sas, like, I don't know, maybe that's the only way he's gonna sound like that, but as if he already knows your fears, your values and knows what you can, so, basically, manipulate that you buy what's best for kids, If you need to explain, your mind is a priceless quality. I mean, it's just that moment. I wonder if it's good to be-- uh, well, to know that next time you get that ChatGPT, you know that it's probably just that he's just running you a little bit.
This, Tanya, is super cool, and I think we've figured out here at the first part of our meeting a very important topic for most people in terms of communicating with the system at all. As important as modern models, they're still smarter and smarter, and nasco-- and they're still good between chat rooms. Well, it's good to read how important it is for you to understand what mistakes might be. This isn't a model of fault, is it? It's you, as you did in the first place, um, the request and how you wanted to do that, right? I mean, I finally turned this around, it's so separate, maybe an offtop.
I, Elnar, still decided to start programming, and I turned it into a-- not just a code, told me, "You got to do some little app." I've turned up a huge infrastructure. Well, as usual, I had to hire a devops, right? I checked separately, there, Supabase, database, Vercel for server, website, there. So I've got everything on the record, you know, GitHub's originals, a bunch of stuff, yeah? And the additional different API-Chem systems where I need to connect to all the reports. I mean, the quality of the whole infrastructure turns and the way the hell out of me has exceeded all my human expectations. I'm good at this, and I'm pretty good, very professional at writing technical assignments and business logic assignments, and business analysis, right? And of course I did work like that. I'm at ChatGPT 5 Pro Extended, giving him my demands. He wrote more straight on you in the code, I've been in high Fast code 5.5 to my task. And, uh, in two days, what he's done to me, he's done me the adminki, and then everything, everything, everything, everything, everything. And by the way, the code is in this cooler Opus because you're not in the thermen-- well, you can manage a little more of these infrastructure components. Although he recommended to me Opus 4.7 or whatever, or whatever, the number to check the whole architecture later, which I did, actually, what I did, and I did, actually, and then I did. And I was wondering. The other day, I went to bed, I was wondering. I have a lot of different projects, I'm in business, I'm mostly, of course, not-- O.R. doesn't come in any way, but I'm going to bed, I'm wondering. For the first time, I looked differently at what it was to put-- not-- programmers don't write code or what people don't write code.
This is the first time I came up with this, and I think I've even felt chapter Anthropic, who said that in a year, people wouldn't write a code that people wouldn't write a code, don't mean that they wouldn't be programmers, right? Here, you mean, I guess seeing, uh-- there was this interesting thing, I'm probably gonna get caught by technical people. I love having eight hundred lines when I did the Adminc. Well, who worked with me, programmers, they didn't love me all the time. I love being able to do whatever it takes to get things done without a code text. And that, of course, slows the system and stops all the work, all the storage of all these data. So, of course, in two days of work, he gave me an admincé that started to slow down the field shift because the submerging of each adminka loaded some huge amount of data. What I said to him was, "I'm in a hell of a way." He said, "No question, let's get this straight." He started to figure it out. Says, "I've got it. " I said, "No, I don't like it anyway. There's a million seconds left." He said, "We'll figure it out again now." He's got something else.
So he's got something else. I worked, worked there for 30 minutes, an hour, and I got back and I'm not stopping anything. I thought, "Oh, thank you so much!" And that's what I could do, right? Yes, that's what I could do. And I'm going to bed, I know-- you know what I was thinking? What's the horse-- uh, uh-- uh, I thought that, uh, I thought, if I was a programmer, what would it be? So how would he, how would he do that? Because I'm still finished with my programming, and then I stopped programming for two thousand tenths, right? Long time ago. And I, uh, built the database architecture, I think Elnar, and you and I built another, yes, two thousand twenty-second and twenty-third. Yeah, I'm a fan of this mess, always in the database architecture and, uh, and I've been doing it all the time. I mean, I'm well aware of normalization, especially, but that's not it. I mean, I don't know anything about all those smart words, names, especially DevOps. Well, that's just me, zero, zero. And, uh, I thought if I gave the software guy that I might have some other quality. I never found that answer.
In the morning, uh, my partner was talking, we have IT business, and he said, "The software guy would never do better than you are now." He told me that. Why? Because he couldn't. And today, uh, my, uh, friend and, uh, often a colleague, and partner Sasha Sagoon says to me today, "Of course he didn't, because you're more in business logistics than, you know, business logic, uh, You know, you're making things more than he does. You see all these things and the dolls in advance." Because, Elnar, he made me fully copy the roles of everyone, superadmin, and he wanted me to do it through email. I said, "No shit! I want a man through Telegram to copy and in Telegram to start a bot. I'm leaving in Telegram, warning that a new user has come in. Then, uh, I open the adminca, I see a user in the adminke, I authorise, he gets a notice, and he can log on the website through that account with his own in Telegram. It looks space. Well, you're doing it right now, knowing it's for people, because I had to-- I'm doing it for, uh, like, one, one, biz-- well, one of my own business, there, real estate. And we have a very large mortgage business with Tanin's husband in the United States. And, uh, I needed the sales managers to pass the accessories, tests. I'm not in the O.R. business at all, but sometimes, uh, I complain about, the quality of the sales, yeah. And I said all people would be on the board, I'll make the tests myself. Well, I gave them a study journal, but I didn't plan on programming it. I thought he was there, the head of the sales department would check with them. And I finally got on the subject. I wanted them to pass through Telegram the accessory that runs OpenAI, there's RAG or fine-tuning models, and so on. I mean, uh, where all the data are, she's leading them so they can't fool her so she's limited and everything else. And I was struck by this story, of course, and, uh, that's a very interesting subject. I'll try to make a special episode of this one, too. We have a series of special subjects already on, and, in particular, how future programming will be presented.
And that's the subject I want to let it down to the next story. Oh, there are three news, three that can be combined and that are very important to the market in general understanding what will happen in the world. Oh, I'm talking-- they're a little connected to programming, but it's important for ordinary people to know where Anthropic and OpenAI are going. And Google. Well, I'm not talking about Google anymore. Really, huh? Uh, you know my favorite opinion that I always tell them that people will work in the CRM system through the OpenAI. CRM system will not exist. There's some kind of single chat where people do all the things. And they're opening up, and they're telling them something, and they're gonna make the sound of the text set, and that algorithm is something inside it, and he's telling someone to call, like, yes, or what to do. I'm saying, why build a separate system? You can do it all in this one window. That's how you sit in the Codex, how I connected the Codex to the CRM system and I can communicate with the CRM system. Why do I need another program? I went into Codex today and I said, what were the salesmen doing for the last fourteen days?
He's in there making the sound of the text set. That's what they did. He's connected to the CRM system, he's found out, he's completely checked himself, right? Uh, I don't need to ask the sales manager, my partner's the CEO. No one needs to be asked, right? I mean, I got involved, I asked myself, I found out. And there's news of phenomenal. Phenomenamic. First news is Anthropic. She reported that she was turning her mind that Anthropic was turning its own, and, uh, her company to, uh, as her company's deployment, her company to develop into the outside market. That is a separate structure that helps companies to implement AI into real business processes. Bam, bum, bum! The market starts to crack, because you know what's going on with the market that's for the same companies?
Because of course there are the largest corporate names of the Americans, the largest names-- companies are coming up. Not only is it American, because OpenAI, for example, is now turning its security system for European companies in separate companies. The news is going on, and the OpenAI says we're turning our company's deployment, and they're... they're buying Tamora's company in parallel. It's a company, uh, it's a British consulting company, and it's got about 100, fifty implementation specialists. And they say we're gonna, uh, go through our technology in corporations, do it all in corporations. And that is, of course, a step, a very strong step in changing the understanding of the future market. Because now the whole market is still in the world, he lives old images of artificial intelligence. And that's me, I always say that, talking to my sister yesterday. I'm saying, everyone wants to introduce what you need, what you had to do two years ago. That's how we in Tuzemune discussed many years ago. That's all in this cough. But now the world is a little different, isn't it? And, uh, the creation of these companies is, uh, very serious, just an incredible movement towards, a change in the world's design approaches. And, uh, I always said that if artificial intelligence learns to politico-- learn to create micro-controls, create, uh, he, he, he, he, he, he, he, he, he's gonna have his robots that will create some chips there, And he can program them.
It may not sound super professional, but still. He can rewrite the whole soft. I'm probably even building myself, I feel that if I'd put my originals, architecture, um, like, a megaplan or some CRM system, like amoCRM, like, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, I don't know, I don't know, I don't know, I----------------------- I'd be an endless computer, you know, like Sam Althman, I'd say, "Take the architecture of this whole system and fix me a new system, more cool and weighted, and aggregating everything, all, all the functions, and "the modern." I think he'd get an answer in ten minutes here, ten minutes later, if he had built it.
It's like I did yesterday. I, Ilnar, tried to buy a speed in Codex yesterday. You can't buy a speed in Codex, yeah. It's like-- you're gonna have a Macrohard company that Ilon Mack-- he's doing it! And rewrite all the products of Microsoft. Yeah, yeah, yeah. Ilinar, Ilinar, look, but look. Oh, by the way, bingo! Like, like, how cool you said. I'm the one who's-- he's doing it and-- he's doing it. Look, he's doing this company, and they've invested a lot of money in it. And it is part of his vast infrastructure. So they're doing this company, which means they're doing 100 percent of this company. They may have one of these initiatives. And he called her that we'd sort of be writing the whole sophth. Because, in the idea, if they come to, here to these facilities and even now-- look, Jensen Huang, a month ago, said that NVIDIA could not be created by AI, but a good company could be. But you must understand that NVIDIA is still four trillion dollars or more. AI market is floating and now it's getting really big, yeah. It's four trillion. It's clear that you can't make such a company. But when he says another one, he, what does he mean? What volume of resources and everything, huh? I mean, still, uh, people who have this outside people inaccessible, but those who have access to power like that because I can't start and start, uh, buying Codex now, even, I can't do it now. I'll put it in there. I'd be putting my own sophth in the Codex now, uh, 1,000 dollars or, uh, I think I'd put it up to 10,000 because what am I-I'm doing now? I'm working with ChatGPT Pro at the time and sending some data back there, I'm gonna recheck something. And the idea is, if this system would have done all this stuff, it would have been a regular call to me, asking questions, I'd like to, uh, give it to you, because I'd, uh, get to another place. There's a place to spend my time, yes, and my resource if she'd created what I really want. Because I have, uh, some pretty serious final initiatives. It's just a little intermediate stuff I do in terms of, uh, tests, human training and everything. And I said I didn't want to save anything at all. I don't need any mini models, I don't need anything, yeah. I want people to get answers in the most expensive models just to see, uh, the fundamental possibilities are now out. Will they open them, Ilnar? That's a big question.
You said from the beginning that they would limit, and we talked about it, and that people would be constrained by systems. And now, the interesting news is that it sounds like, uh, that the answers of the OpenAI and the intellect in general may depend on the language of the information environment. In the study, Princeton and Euric Alert said that States and institutions could be indirectly, but they did another thing at all. They said they could.
They did the other thing. They said they could indirectly influence LLM responses by developing online media and texts on which models learn. Well, we talked about it two years ago. They've done this research now. We talked about all the interference, history. If you want to spread your information, don't waste your time, don't waste your time, don't you? But, uh, Tanya, that's the thing-- that's the subject, right, Tanya? Six months ago, you said you saw ChatGPT in Europe as though you were stupid. Yours, right? Your version is a feeble one, right? I'm stupid. And you're like, why? What the hell?
Amazingly, by the way, there was no such effect in China. I mean, he was actually doing great. Yeah, Europe is a terrible thing.
Although they are against China because China distils their models.
I don't know, Sas, but they all use Americans, too. I was shocked. Why? Everyone uses American LLM.
No, look, if you, if you get back, but the volumes are still users, so that someone from our viewers in the comments also put it, there's a Chinese market more detailed, but we talked about it, so, I know, I'm sure the entire American market is bigger. But still in China, uh, in China, the number of users is total, like, about half a billion.
I think those people you've been talking to, they, if they have access, they use American models, but there's a lot of restrictions. Now there's been a whole series of restrictions from OpenAI, like for the Russian market, right? And they find more and more decisions to understand that you're using it, for example, for companies that are blocked, there, or people that can't. And so in China, I think people use Chinese models often to get some independent, external opinion. Plus, uh, their data don't write inside the State because, look, in America, even though they're inside OpenAI, they're written within 30 days, or they can be accessed by a short time in court. There is a difference between sessions or yours, and in China, there is a default . Look, it's like, in Belarus or Russia, a phone call to anyone on urban communications has access to a regular policeman, right? In America, however, a court order is required to obtain access. It's a little different story. You didn't come in, just on the computer, you opened it, and you see, right? It's a little different. It's clear that there might be a little unusual, but it's a little different, different kind of access. And in China-- I'm not saying where the best place is, where worse, and in no case, I'm not comparing, by the way, here because it depends on, uh, laws inside, on different politicians, and on the politicians of even athletic companies that do.
I mean, I was looking here recently, and I had some riles, and Tinkov got caught. He said he didn't like Dorov, that he's doing nothing. Well, Tinkov, he's always a big deal, Oleg always made a big speech, even though I spoke to him. He never said that to me, but he's, uh, he's not talking about the Dura. And, but he said one interesting phrase at the end. Well, there was a very interesting phrase on the roller. He said that-- that's not Oleg's point. He said I didn't mind the State writing or the information system writing all the data, just so that it would help me, you know, live better, right? And I keep saying that, yes, that, well, I understand that ChatGPT, I've been looking for my own chat here recently, and he found me a quarter-year-old chat. And I realized that ChatGPT learned to look. Remember, Eldar, you're on-- and Tanya, you're on this subject, remember, saying he didn't lift my old chat once. He found my chat very easily. I don't even know when that chat was, do I? And that's why China is writing very seriously. And these systems are connected to some single Chinese core that weChat is imputing, yes, and some model is written there, and, well, DeepSeek is, uh, texting or writing an internal system, there, Alibababa, et cetera, yeah? So I think that's why people start using other models. But we'll go back to the question we discussed. Yeah, yeah.
A little bit of a comment about this phrase, which, uh, can write anything, just there, he helped me be happy. It's just a simple, ex-concitement question. Most of the audience were probably watching the Matrix movie. And the first part of it is the character of Saifer who betrayed them. And now he's talking to Agent Smith, eating steak. And if you remember, like when they're outside the Matrix, they eat all the balanda. And he's eating steak and saying, "I understand that this steak is not real, it doesn't really exist, but how it smells like normal steak, it tastes like normal steak. I'm totally comfortable with that. I'm willing to forget everything that's been before. Just stay here." Yeah, just make me feel good.
And it's like extremes.
Classic.
I can get a ride. Yes, yes, yes.
It's very...
You can go right that way.
You're saying that you're actually raising a very important human question. Our channel is still... and we try to wear a lot of artificial intelligence. And someone might think we're so dry electronic wires and we want everything to be computerized. And that's a big question, isn't it? They say that it was now discovered in a court between Sam Althman and, uh, Elon Musk, that Elon Musk, uh, ran to open the OpenAI when he heard from the owner of Google that that bet is more on AI than anything that does on AI. Maybe, and as a head-on component as a whole, it's a human being. That's what the phrase is. To humanity. I'll, we'll study these data, we'll tell them, yes, to be sure not to lie somewhere. But here's a very important thing that I'm putting on humanity. And right. And here you always remember, you're finally in the middle of what you're thinking. Because that's what people chose in Matrix. And this man, he said that you'd bring me back and make me rich. I don't care if I'm not real, do I?
And there were people who said, "We, we want to be real, we don't want to be unreal anymore." And here's the story of learning how to interact with AI and live with him in the truth. I think we've got a channel about this, right? To understand how he works. Even the subject we started at the beginning, why he offered you some kind of product. I mean, get this straight, turn your head on. It's not just a voice around here about something to say or believe everything that neighbors or Google say, or AI, aaaa, you know, trying to figure it out, actually, trying to understand.
You bought a baby device. There's now a whole research on the harm of child AI devices, because it's not just doing the data, it can manipulate the child very much. And whoever said it was a toy and a shot, yeah, you're still-- that's not, that's not an object, it's not a shovel, it's not a soft toy, is it? There's a, uh, pretty serious car inside that has rules that have rules for certain actions, even if they're good, but it's good, in the opinion of this company that created this company. Toy. And you and I know there's a lot of options, right? Just as you come to the doctor, one will say take the pills, the other will say fresh-- fresh air, third will say surgery. About the same thing! About the same thing, yeah. And here's the choice of you, uh, to figure it out. ChatGPT, I think it's very much possible to help me figure it out.
There was a study on the health of the creativity that most people already use ChatGPT in health, but-- for health. And I was here, and I had to go to the doctor, and there's one thing, to the surgeon, to the consult. And I came very well prepared through ChatGPT Pro, and my sister to me later, I described to her, she said, "The doctors hate people who came with the GPT chat." I said, "I didn't tell her I was in ChatGPT." But when I spoke to her, I felt very confident. I mean, I asked her the right questions, and I didn't put her in, and I respected her as a man with a separate opinion, and I didn't go through with her. ChatGPT usually destroys doctors. He says, "Give the doctor forty-nine questions." I say, "Is that a question you can ask?" "You have a right, so." And there's a-- "Death a doctor" called, right? And, uh, "What's your husband's slick?" There. According to that law, you must tell me. But I left a doctor with a very wide range, because I came from the beginning, reading and studying the world experience of different countries, and then, uh, I even talked to some friends or friends. And then when the doctor told me, I-- that made me feel confident. I mean, especially when she told me how to get the data, I heard she said things, well, professionally enough, yeah, wide. It doesn't mean it's gonna be right there, and you're not gonna die. When you get shots, you can still die, but it's useful. That's what we need to figure out to avoid being this man with the steak that Eldar said. It's a very cool thing. Here.
Eldar, would you like to add something to the topic where I was telling you about, uh, programming mine, then?
Oh, yeah. That's what you're describing, it's a show-- well, the programming is gonna die. In fact, uh, dying, well, or ends with the aporks, frillans, and so on. I mean, when you need to make a little project, and you're willing to pay for it, there's fifty, seventy, one hundred twenty, there, two hundred dollars, that thing is definitely gonna be done with the help, right there, 200-dollar. I'm sure you can do the subscriptions. Especially by the end of May, the OpenAI has raised limits. Claude is starting to lift the limits. Well, there's two outlets in a row saying that we're raising after they made a deal, right? And from, uh, HIAI, they've been diverted to their resources, and they're actually increasing the limits. A OpenAI announced that the Codex limits would be increased by the end of May. I think I'm following Anthropic.
I think after Anthropic they're gonna be the big ones either for a prom or something else.
But if there's a wish to try, you have to try now while the limits are raised. It's not worth losing time. About how people program. Look, in many companies I know, not only do I welcome the use of AI, well, LLM, but they also compensate for the purchase of subscriptions or corporate subscriptions. At the interview, there are many times where the Wype Coding interview starts to be conducted from the series: you're the target, you're the one who's gonna use what you want. You want LLM, you want Google, you want to write it yourself. And that's how you just follow the way you do it. And it's all going into the professional environment very quickly. I can't say I don't like it. It's accelerating the job. The job is more workable, but I think the job is getting a little more interesting. You're less engaged in routine things, and you're looking a little bit more on top of that story. The only thing there is, of course, is a very delicate line, especially when we're talking about some industrial code, selling the code, which is very carefully to keep track of what's going on, because we're gonna put it on. The system, well, is very easy or to keep vulnerability or something.
Well, if you're programming for Amazon or Ozon, don't say I have a glu-- I'm in a hellmine situation and send the whole code for change.
Get it fixed.
A completely working structure, perfectly alive, very fast. And that's why Ilon Mack must have done MacRO Hard. I think the market underestimated what he said it was and what side it was going on. Because at some point in time, it's clear that you can write a system quickly and possibly show up. It's just, maybe, the operating systems won't be necessary. I mean, maybe we'll come by the time the operating system doesn't need when the model is the operating system itself. Google does everything separately, which is somehow built into as an operating system or NVIDIA there together. I remember one of the people in NVIDIA talking about it, didn't I? In NVIDIA, they were together when they were with-- who they were doing, forget. We had laptops together. We've even told you about some video and they're holding chips, their chips, and how internal this move is to the infra--- to the operating system. I mean, basically, when you have a system, you're higher, and you don't need anything above it. I mean, you have this context window. If you have to turn this context window around some sort of visa or some application, but it'll just blow you inside. I mean, it's just gonna show you inside. It's a question when they learn to do UX UI and learn to make interfaces. Yeah, and, uh, do you need that interface? I asked myself to do an interface as a separate application for sensor because, well, as much as I wanted to press, I had to do some kind of a button on the plate, right? And or wanted to have some sort of data. And the big question is, since I'm still not, I'm still using, I'm using, I'm using, I'm using Codex, and ChatGPT and that's an app, I think, do I need this app? Or I would still be better off just being directly in a system like Codex. I mean, Codex, I think the closest thing is that it's your daily window. The only thing I don't like is that Codex is split up with the usual GPT chat, and I can't really load five Pros in the Codex. So they have some five High, which is clearly targeting other tasks, right? I want five Pro to turn on my extended. And this is a separate subject.
In fact, it's the same thing, only in the Anthropic ecosystem. And here I want to add a big story to America that Anthropic recently, just yesterday, I think the news was that Cloud for small business, right? There's a Cloud for small entrepreneurs, a small business tying that can connect to Intuit, QuickBooks, PayPal and other systems. And, in a rude way, you have to report, yes, or there's some kind of reporting to see for a period and so on. You've got all the conjectors right now. You can connect your system intuit, QuickBooks. Well, those in the States know about her.
Yeah, yeah. And this is all getting closer, you're getting closer. You don't even have to think about API. You've connected the conjector, you've got everything. You can prepare a report for tax return, see it or not, press the box. Well, it's best not to do that, but, uh, rudely, this whole story, she--
It's better not. America's better off the accountant.
You're a risky guy, Ildar. Don't look at the tax report of the box in America.
Yeah, yeah, yeah.
Look, but what I'm saying now is strong enough, Codex wins so far in terms of work in different systems. I mean, if I have GitHub separately, if I have one, I don't know, Vercel, if I have any Google API, if I have a separate Telegram API, win Codex, and Cloud wins from the Google API, The viewpoint of developing a more complex, quality code and architecture. And now he's-- I just-- I just didn't mind starting to connect myself to another system, you know, like a second system, a second system. He told me that when you write the whole architecture, you'd pay the money on the Codex like Cloud, like he's Max Opus or what, Max, who's twenty, right? Or how much it's called. Or there's two hundred, you know, that's worth two hundred dollars, you're gonna pay for it, and he's gonna analyze your whole project. You're just gonna spend some money on this through the API, including some extra, yeah, more. And he'll analyze everything and give you read-only, for example, right? And he'll give you and tell you what quality the code architecture is, what you do, and Ildar, I'll do, I'll be sure.
I, Sasha, to see where the market is moving because all the last six months are repeating Anthropic. Anthropic's gonna roll something out, everybody else is running after him. And, in fact, Anthropic went into a big business, into the enterprise, in fact, the OpenAI turned up that way. Now, it's working out that Anthropic's targeting a small business. They even write in their news that, like, 40 per cent of US GDP is generated through small businesses. It's a huge market.
You're the one who's right. So they said we even built a company to automate AI. Yeah, yeah.
Here.
It's a very good one.
And accordingly, the automation company comes along, and that's the story for a small business, when you don't have the money to pay for this big one. The company's deployment, and you're just doing this to these conjectors. So they're all going to the big business first, now, respectively, to the small business and the same side, and everyone else will go as they say...
And will this deployment be a company to create companies like Intuit, inside which QuickBooks, companies like Salesforce, inside which CRM system? Would this deployment be a company to make decisions? Because I've made one decision there to assess, you know, for the sales associates. It would be very hard to migrate to another field, because, well, it's not a story, you sell tomatoes and you get cucumbers, and you're gonna sell it the same plus or minus. And that's not even a tomato you sell, you need to sell meat. It's another business, yeah, inside. It's been hard before, and now it's very easy. Well, I'm in the picture because all logic and infrastructure are basically the same. There are differences in the knowledge that lies inside. Yeah, just like ChatGPT has inside knowledge that lies. And so, of course, these companies, I have a feeling, because you're just saying that Anthropic, in fact, through these company deployments, will make its decisions, not the conjectors, and at some point, they'll say, You don't need to connect, we'll give you our own-- our e-mail infrastructure, we'll give you our customers' communications infrastructure, we'll give you our massage aggregation infrastructure, We'll give you our infrastructure and system, there, data storage from the database and everything. Well, I mean, they'll be giving their infrastructure. Why would I buy? I just bought it to start it, so I had to buy a bunch of signatures from different people. And the question is, why would I buy those signatures? He says that you don't need to buy anything, we'll give you everything. Or why would you have a report? I have Codex reports in Excel, and he also gave me HTML. I open up, I look like, oh, man, that's more fun watching!
Right away, right? A staggered, fucking report. And you're like, a spider!
Look, at the vip-coding of websites, what kind of problem does it have to be? I don't know, there, she's still relevant or not, but she definitely showed up that all kinds of integration, uh, the system starts to pee from scratch. Well, for example, integration with PayPal, with Stripe, with some systems, with CRM systems. And it's not that much of a Tilda thing, like, yeah, but now we're just going to the technician, but there's a Tilda inside that already has integration with all the systems you need. They're high-level, not glue.
High level, no gluten, work. You're gonna put it on, you've got it all on your website. When you start this story, it's not always that way. It works either inaccurately or once, and when it gets loaded, it's getting over to work, something else. So there's still things to look at. Noo I think it's gonna be a little quick, um, healing.
Oh, look, you're talking fun stuff on the one hand in terms of integration. That's how I got on the CRM system, there'd be some conjector, but on the other hand, when I got directly through API, and told me to study all the documentation in full, to get maximum access, Uh, that I can condom, actually, if I had a conctor who doesn't have anything, I'd have a problem. And on the one hand, you're right to connect such systems, absolutely right. Because when you write from scratch, there are some, uh, things that are checked on the two sides, checked on the two sides, for example, PayPal, and the CMS system. On the other hand, you might be limited. I always don't like restrictions, like in the Whoop bracelet. I can't access certain raw data. I don't, I don't have access. I mean, I'm very limited to everything that's going on. I'm saying, if I can wear a sensor of the worst quality, I'll just get the raw data, and I'll get the ChatGPT myself to build the reports I need, but I'll guarantee, well, I'll know that my data is, retained, even if I'm out of a signature. And the point is not a subscription. At least I'll see these raw data, watch, and the system's wrong or wrong, and where there's a mistake or a misstatement. Because whoop won't tell me: I'm sorry. That's how oura Ring. I'm at my wife's place, this is the ring that wears, the purtness ring. I see, she's got, uh-oh, data intervals like they don't. I'm saying, what is this nonsense? She doesn't tell you, you know, I haven't received a signal today, like, from the ring. She pretends it wasn't. Okay, well, maybe at this point of time, she had, uh, heart attacks, there, or at that point of time, yeah, she had something-- that's when her heart was pooping. I mean, the difference is still in terms of getting it. That's gonna be interesting, of course, how it's gonna work. Uh, please write what you think on these topics. Write, our viewers, what would you like, uh, to start with the subject from the standpoint of special subjects, huh? We're still planning on going out to the specials' worm now. What would you like to hear more in some fields, projections? Tanya, by the way, in the comments, people wrote a lot, like, how good it was that Tanya-- when Tanya talks about the topics she knows. Tell me more about design.
Is that a hint I'm not talking about topics I don't know?
This? I don't know what they're implying. I'm not talking about that. I'm not saying what Tanya says. Oh, we're in there, I'm being sniffed, there, two, all sides can be floating. I know or I don't know. Illnars can be recurrent, too. It's more like a topic, uh, it's more of a story about what, uh, there's more, like, thematic episodes that people are interested in. And why don't these thematic episodes be discreet?
Yeah, I'd love to.
Yeah, because there's one--
If there's any questions, let them write in the comments.
Of course.
I'd love to answer that.
Of course we'll try to see them. I'll see you in exactly a week. Bye, everybody.