I'll tell you some very interesting cases tonight. And most importantly, in real examples, I will show you again the difference in the use of the system in my normal life. I'll show you a lot of business, too. What's the matter with the agents? Hype, which has been in many bloggers for over a year, is especially telling people who don't really know technology. We'll give these references. Before, in the GPT chat, when you asked for some things, he often refused to answer, so that means that this system has anti-subsidence. Hello, everyone! You're on ToTheMoon and we have a special issue today that we're doing Wednesdays. I'll remind you that our graduations are coming out, our sub-cate goes on Sunday. Seventy-seven episodes and a very important subject I've been trying to take off for a while. The topic of this application of models, using GPT chat, using it in its life, using and using it in business. And somehow I've done this with my personal use. Now, given that GPT Five's chatting is now out and we've been doing this a few days ago, I want to, last Sunday, yeah, I want to take a wider look at this subject, and it's the use of chat. GPT Five in real life and show you what mistakes you shouldn't do with artificial intelligence. What is not worth doing exactly because a lot of mistakes have been made over the past few years. And most importantly, what happens in the front. I'll give you a simple example from the beginning of this business example. I have one of the business of education, and I don't take the operating activity in this business, but I'm so strategic.
And we talked and dealt with education with my partner, if we ask what happens to artificial intelligence, given that different systems are now emerging, which will be What happens in the future? For example, if people study psychology or people study programming, or people study design, how will they be? I'll give you some examples right now. It is very interesting for the general, to understand, and, well, to have such a broad, fundamental understanding of the use of artificial intelligence in its real life. And many very common examples I'll give you, very interesting, which are real. Write off, by the way, how do you get GPT-fi? Were you working with him? Who uses Gemini, Grok and some other system? What were you comparing? What are your conclusions now? Write, please, and don't forget to keep the canal to keep the releases. Sign and comment. It helps development. We're a good time, I'm like I said, the main seventy-seven editions are out. That's very decent. We're filming, doing all the videos and our own motion. Always need support for the move. Thank you very much. So we talked about it together? We were together discussing these changes that will be made in terms of education. And that was an interesting aspect that came up immediately. I'm saying we should look at two things right now. In fact, the first is not to do extra work that concerns, however, the use of such an illusion, not to do any extra work that is relevant to the introduction of artificial intelligence on the one hand. On the other hand, I'll understand, and how the user's scenario will change, because it's 100 percent going to change. And I was talking about it on Sunday, and I gave him an example. Look what's going on with the telemedical.
I mean, earlier, if there was a health problem, I wrote a personal friend. Someone was telemedicine-cientist. It's cheap. And I received a quick recommendation. Now, obviously, it's pointless to have a friend from the start for the first recommendation and it's pointless to call telemedicine with help if you have a feeble version, I even think it's a free version of it. different systems, but if you are, it's true, you have the opportunity to have a broad view of the knowledge of the world. I like to ask questions. For example, if I study the system with some kind of PC or, I don't know, we're gonna take everything from a person, for example, thyroid or, I don't know, heart, any problem. I like to ask how this is about all over the world and start asking for analysis. Please analyse how this is relevant in Japan, as in Korea, as in Europe, as in Britain, as in the United States. I mean, I'm asking for a very wide spectrum. I'll remember if someone's been watching from the beginning, I live in the U.S., in the Silicon Valley, and I've lived twelve years in Moscow. I've lived there for twenty-seven years in Minsk, and I'm from Belarus, Minsk, and I'm more aware of this wide range of different health systems. And I ask why I'm asking the question about different countries because I want to see exactly different approaches right away. And by the way, I'll notice you at once, let's record that there's a big difference. Here we have GPT-five chat and GPT-five chat, you have a top choice of model.
You can choose just GPT-five chat, you can pick GPT Thinking. It's when you start thinking or a point like that, plus you choose Thinking, and then you'll come down right away, you'll have to tell the system that it has to recheck and think. If you don't put Thinking, the usual system can think, maybe, you know, maybe throwing out quick answers, maybe including a Reasoning model. It's a model where it rechecks and rechecks the various data, and different questions do. You can choose Deep Research.
Nine out of the ten people I'm talking to, they don't know what deep research is, they don't use it. And if you want to get a quick answer, like what's going on in-- I don't know, put on, like, a heart or a PC, there, I don't know, a thyroid and ask your own recommendations, you might, for example, I'll ask ChatGPT Thinking. ChatGPT-5 just asks, big question. You can ask me to think. But when you say, "Tell me in Japan, there in Britain, the United States and Europe," he'll still give me some kind of answer, some kind of answer. On the other side. If you turn on deep research, he'll go and study separately, there's a hundred or two hundred resources, well, somewhere, maybe fifty, 150, and study more carefully, and he'll be more careful about this. The task and give you a very detailed description, a much more description, a much larger amount of data. What do these countries think about it? Here. And many people don't use deep research, but it needs to be used because there's a big difference, considering what kind of answer you want. So, my son and I are going to a few days tomorrow, in the mountains, in the mountains, in the lake, in a little, little house. And I can ask the GPT chat just, the GPT-5 Thinking chat or the GPT-5 Pro tip, I have a version, I can ask or Grok, there, four Heavy, or just Grok-4, I can ask, "Tell me what's in it. There's, like, what lakes are? I can start deep research. Still deep research, if I just ask, he'll give me, there, four-five lake recommendations, short, and deep research, he'll leave me and whisper me all the possible variations that are, and he'll give me conclusions, I'm going to be 20 minutes where I'm getting better, where I have to go. I mean, more detailed. I also asked him to know yesterday in deep research what restaurants are for breakfast, for lunch and for dinners, that we just want to cook houses, there, in the woods. Although we'll have a part of it, I think he'll give me all the different options. I'll come in and read fast enough not to come in, not study. At least he'll tell me what's in it. He's usually looking good enough. Same as my previous time-measure. It depends on whether you want a short or a long one. You're using different deep research. What's the ChatGPT-5? That might be the first time in the coming months people's troubles are gonna be, but he's the cool thing he's doing for you, you're gonna start some sort of more detailed analysis, less detailed analysis, and so on. But deep research you can still open. So I've given you this example from the position that telemedicine will disappear. I, remember, I started a history of education. I-- someone will say that I'm out of my mind, I'm not out of my mind, I want you to show you a wide range of perceptions that you really get your mind and make decisions from the standpoint of artificial art. Intellect. Most people can't make these decisions. They use indirect models. And even a video on, there, forty minutes an hour, see how much it-- I'll take it off, it'll open it to you, well, a much wider picture and save you a lot of time ahead. And I'm telling you, some areas will be gone. Well, as a model of telemedicine, we need to watch, and what in our business, in terms of education, can even come out or disappear, especially when in low business.
Honestly, yeah, especially when we're in a few businesses. And he says, "Look, well, we're already doing it, like an agent." It was very interesting when I said, "No, don't do it." He says, "Okay, I'll tell you now. We're making an agent." And I didn't even know what some agent was doing there because, well, it's an operating activity. They do, they do. "They do, he says, "the agent to help us. I was on one birthday on the other side of the Gulf yesterday. And again, they remembered someone saying they made some agent, and he didn't work well. What is it that an agent that's usually told about is everybody wanting to do? So there's a myth for about a year and a half when a lot of people say, "We don't need a real living man anymore. We packed his knowledge. And that knowledge makes the agent completely as relevant as the information, yes?" And there are about two casees, look, different. On the one hand, someone says I'm using now, I-I'm-- I'm making a lot of content, and I'm the one who created an agent who's responsible for me. There's a second one that's a case. People want to re-establish some old, 100-- another person who, for example, is dead, wants or is alive, but they want to take his amount of data and expand his data, like without him, right? I even have a friend here, a very famous American business business business businessman, known to him. He had the biggest online chat in the US, and at some point the company was worth six billion dollars. And he's doing a project on what's rebuilding people who are about to die. He keeps their data on them, allegedly to be able to communicate with this man after his death. And he says to me often, "Oh, it's nothing, there's literally a man who needs to ask, and the system can talk like this man. Or we'll put a lot of data in there and the system will be perfectly perfectly perfectly perfectly spoken." That means there's a kind of discussion. And these agents make tens of thousands. I know more than 100 businesses separately as start-ups and probably two-tier business, I hear people who created some kind of psychic agent, personal development, learning systems, and In projects, for example, where some man is kept, it is believed that the system will be better suited to the person. It's an illusion. Let me explain. That's what you're telling yourself why. It's very important to understand. It's a mistake. I was sitting with my friend here in the valley four days ago. There's a cafe in Palo-Alto Venice, there's a lot of people sitting there periodically. And we were sitting with her. She came from another state, and she said she wanted to make an agent to help companies develop in business, right? That she'll be filling up a lot of data. The Jacobs will save the company's resources. And what she's gonna do is make a cool agent. Let me explain to you two blocks, two, two, two. Why is this task now, at this point, with the agents behind it, unsolved. Now, at this point of time, it's very important, isn't it? And if the speech is made, it's worth tens of millions of dollars. I can tell you that, well, I'm a programmer myself, and I'm very good at the different technology systems. And yet I am very much aware of the large volume of data and of the business of myself, as I have managed, developed, and I always go out of approaches, from, here from, the way. On the one hand, I understand business, on the other hand, I understand technologically, at least at my level. It's clear that at my level, yes, I'm not a creator, there's artificial intelligence systems. So the first side of why you can't do it. I'll give you an example of a large body of data. And by the way, this is a fun example you can do, too. There's a author, like I'm studying a lot, Rudolph Steiner.
But we're not going to get into details and details. You can study anything, programming, astronomy, mathematics, hair and sewing, design, planting, anything, right? This author wrote about 5,000 lectures in his life. He died, and there's a archive of these lectures. There's a archive of some books, materials, etc. The archives were made in German originally. That is, this man lived in Austria, and there was Austria, Germany, Switzerland. He died in Switzerland, and a lot of material on German. This material is translated into English, Russian. I think it's 100, 150 books in Russian. Only in Stanford, I saw 100 fifty books in English in the library, and there are many languages, French and so on. But there's a German language as a fundamental. I'd like to see that author find some information. For example, the author considered, in terms of how this man was, that man who wrote about spiritual science, his own health system, his child-rearing system, his own development system Agriculture and so on. That's a lot of things. I mean, this man has been telling you something from a point of view-- I'll give you a specific example, right? He told me when a man falls asleep, what circumstances must be in the room. Well, that's what people say, like, that you need fresh air. And he's talking about it, too, right? Or that it's necessary to be dark that you're gonna have to get air all the time. One of the things he's writing about in his speeches, including those translated from German. And again, it's now letting you find a system if-- but it's important to look for a little more than just a GPT chat, it's important that you start deep research. Deep research is now a lot of systems in which systems, including those in Grok, is Ilona Mask xAI, they are in Gemini in Google, and they are in OpenAI, they're in DeepSeek, they're probably like, uh, they're a lot of them. eat deep research. Deep research is a story where the system is running through a huge amount of resources, online, and a huge amount of stuff comes in to look. To get information and, uh, to second a large amount of data. So deep research is usually a lot of information, a very large number of information. So, here you are, this deep research, which means you've developed this deep research, and I've got information where Steiner described everything about sleep. And I started deep research in a row, a few grand to get the system to analyze these 5,000 books. Well, 5,000 lectures. What is five thousand lectures? Five thousand lectures. Each lecture is about 10, 10, 20 pages.
Well, let's imagine it's $50,000, 500,000 pages of the text. She's analyzed 50-100 pages of thousands of text. It's obviously not gonna be able to get a hundred thousand texts for one deep research, and he won't do it. You must understand that too. I mean, if you just asked ChatGPT, especially for free, he gave you some sort of lecture, got some slice, found it, and gave you some. All this information is not a fact at all. Not enough, it could be hallucinated. ChatGPT-5 is considered to be saying that he even in the thinking model, and there were less hallucinations, he hallucinates by eighty percent less, ChatGPT-5 thinking than three is the previous model. And the previous reasoning model hallucinated less than, for example, four things alike. I mean, ChatGPT-5 is generally hallucinating, but much lower, although it can also. What's hallucinating? He can give you the wrong real information. I mean, it's important to me, for example, when I ask the author something, that I'm asking him that this information, and, for his word, it's him, right? You're gonna write this down in parallel, please, what information you've been looking for, right? And actually, what are you asking, what are your questions? I'll tell you some very interesting cases tonight. I'll tell you how you can automatically create certain pages, put a motorway. There's a very interesting case that people don't see what you can do. And most importantly, in real cases, I will show you once again the difference in the use of the system in my normal life, once again in my normal life, and in business aspects, you will also show many business aspects. So again, given my entry into dozens of businesses. So, here she gathered me from about three deep research, and I got everything. That's it. I've seen all the quotations. Or maybe the second one, there, there, there, there, there, there, there's a full set of data from the fourth. And keep looking, it's very important that she's got everything. That's all he said, and I'm interested. He says it's not good when plants are in the room at night. I'm not going to be in detail with you. Never mind who it is, right, right, right, or not. I'm giving you another example of how you're gonna figure out why this is an unfinished task. He says you can't have plants in the room. I have a further clarification: can I just put in a room what plants can be in the room, if there are any exceptions? And I have a further clarification: if the plant is small, well, like this? And we, by the way, actually had a question with our wife.
Just a little succulent. Can you stand or not in the room? What's the problem? What's the matter with the agents? I don't know what answer Rudolph Steiner would have answered. There's no information in this. ChatGPT may aggregate Steiner's information, sagate information from other authors, add, I don't know, Rerich, Krishnamurty, doctors, any country, anything. And some kind of information you want, scientifically proven, unproven, esoteric. So, personal opinions of people of a certain age who speak Korean, who you want. But there's a problem. If this information was not made up by this man.
The formula-- formulated by this man, you don't know what he'll answer. We don't know, his conclusion is unknown. And when people always said that you could do two hundred questions, my neighbor said, well, this businessman is famous, and you can make a man. I'm like a man who created-- creating an incredible amount of content in my life, right? I have a thousand, well, tens of thousands of content, hundreds of thousands of pages of this content. I know that the system cannot answer many questions. Even this system is the coolest. I'm gonna tell you the second thing, the second problem that's there. I mean, on the one hand, look, it's working out, the system doesn't know the answer. A lot of people, I have a friend who in business learns and tells the system to answer, she can answer, limited to your data. That's very important, limited to your data. So she can't answer the questions you've never told me. Of course you can make an agent that can help you find information and find it, answer quickly. It's a fact. I'm the one who uses such deep research. I'm starting it and looking for it. Only why would I need an agent, if someone told me, did, did what the system did, uh, in Buddhism, I'm gonna find you what the system did in the Buddhism of the Chat, the chat, and that means, that chat, it, uh, it's, uh, it's, uh, it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it's... it' The information that it's supposed to be super cool, which is, uh, to be released, which means it's a bot that teaches Buddhism and so on. I'm saying, why would this agent need? Because you can now turn into GPT chat room, uh, and other systems, in Gemini is staggered, and it's just that everything that's possible, right? There is a large number of books and libraries that are not even online, such as Buddhism or Christianity or Muslim. Well, what you're studying, huh? The direction or direction of the exercise you're studying, I don't know, theosofia, there, everything you're studying, there's a polyto or anthroposphobia.
There's a lot of information on it, and you can use different languages, including information that was in the script, information that was plunged into Hindi, information that was The Chinese language stitch, the information that is polished in Korean, Japanese, in a vast number of others, in other dimensions, is it? Can you give yourself this information, why do you need an agent? What's the second problem? All agents that people do, they do them through API. And usually in these systems, API uses low-cost models. So you just know that if you want to analyze 100 grand of the text, in some models, get in that volume and do a little bit of a steep system analysis that will make you reasoning, The amount of hallucinations, your requests, your training, first of all, will cost hundreds of thousands, millions of dollars, and your for- and infrastructure will be spent, and the software will spend an incredible amount of money. And your enquiries, and the likelihood that your programmer can do this at all, is a big question, that's a big question. And that he works for you, this man. Because those people who can do, they work for serious companies. It doesn't matter what country the world is in. I'm not talking about the States where hundreds of thousands of wages, millions, tens of millions of dollars are already in. There was an example of a billion dollars. Yeah? I'm not even talking about Europe-- about Europe, there, I'm talking about a big company in a small country that specializes in something, I don't know if in Russia a man works in Sberbank, right? You'll be able to pay him that kind of salary. I mean, if a man's a programmer, he's really good at it. So, uh, they use cheap models in systems, conventional low-cost models.
I mean, for example, in the previous version, the system would use 4O rather than O3 Pro or O3. And this is a different information, a different conclusion, such, very intermediated, or dry, or an extended amount of hallucinations. So it's not a system where I can do this now? I take, for example, a thousand pages of my speeches, I'm putting in the system, I'm starting, for example, GPT 5 Pro, and I say, "Take me the 10 largest quotes that I'm saying, or, there, give me the ten words I'm saying. He said, uh, about the right behavior with the kids. Or you can give me ten statements I've said about the biggest mistakes in business. Or give me ten CDs... that I've been telling you about how to build relationships with partners." And it's in the system, within the system, I'm gonna start looking and processing a lot of data. There's 500 pages or 1,000. And the quality of this response is uneven, once again, with any agent system that is on the side. Because my request is expensive. When you use free chat agents or agents that are there on micro-subscriptions that side people do, these agents will never give such a quality of information. Once again, they need to be very serious about this information and to shed just a huge amount of information. So you can really, like, have a thousand lectures, like you've got a thousand videos, like youTube. All this must be, for example, transcribated, transcribed. You're putting it in the mass, training the system on, uh, getting some, some data, building some chains, different, I don't know, mistakes, reasons, advantages, different scenarios, cuts, details, analyst. Get it. It's you, I don't know, there's 30, 40, 55,000 pages, like, or there's 20,000 pages. It is further added, the text is inserted and the person is given the opportunity to work with him. What's the problem? In one request, the system can normally operate approximately four hundred pages, five hundred pages. The dialogue window has now been expanded in GPT-5. Well, there's a hundred, a thousand pages of the text to be processed. There are Chinese models where you can do more contextual windows, but it's expensive. Oh, not expensive, dry, sorry, it's dry, it's kind of a language-processing thing. Someone, by the way, thinks that the quality of the Russian language is supposed to be exactly the same in Yandex GPT. I'll tell you right away that today two systems are winning all analytics, winning two systems as a language today, that's, uh, OpenAI GPT and that's Google Gemini, right? Number, number one. I don't remember Claude Antropic, too, as well as Russian, but it's important that the cool models are, they, uh, in language, in detail, they're just moving forward. And notice that the quality of even the Russian language is often important, he made some mistakes out there somewhere. It's okay. Well, it depends on your task, which you're putting. If you write a book there, it's another case. And the question is, uh, from the point of view of the normal life and the study of the ordinary life. Like, I used to give an example, you might have heard him. I have a Russian borze, and I like to ask, for example, in the Russian borze, what demands, for example, the community and the cinemas to the Russian brazies. In terms of, for example, nutrition when I give her a subproduct, and I give her, for example, a subproduct, some sort of subproduct, I wonder if in some countries it will be in some countries. It says, I don't know, some stomach, mmm, chicken is a subproduct, and in some countries it's gonna be written that it's pure meat. And I'm walking, I'm interested in the Russian-speaking market, Russia, I'm interested in Russian brazies, I'm interested in Europe and the US interest. That there may be some details different from the perspective of approaches and details, right? I mean, sometimes they write in Italy, for example, there are people who specialize in Poland who specialize in this, in Russia, they think that you should do this, which means in America they think that You should do that.
I get a general opinion. Yeah, I care, so I'm a general opinion. Again, it's another challenge. So what's the problem with the agents? And what I said to my partner, I say, "Stop, stop, stop the project, you can't make the system, you won't solve this task. So you're not gonna make it up because you're not getting a great quality." What I recommend to everyone now is that when people in business ask, "What do you do then?" Give your employees some expensive models, buy them models, buy them Gemini Pro or at least, ChatGPT Pro, or at least ChatGPT+. Buy them Gemini, a normal version or Pro, buy them Grok, a, uh, a paid, normal version or a expensive one. How much is she there? Three hundred or more dollars, I don't remember. I have a subscription that's got Grok 4 Heavy and so on. They open all new things, too. Buy them a pay-as-you-go and, uh, give them You're not getting that super quality differently. And one of the important stories here is the ability to ask questions.
Because when you give the programmer, you know, programming the system, you know, the programmer needs to understand, and what kind of data you have, and how it is, this amount of data is correctly set up. What's the third problem? She may be less important to people who, uh, don't get a lot of things into their business, but I'll tell you what, so you can understand it. See, today, technological progress is moving just at a tremendous speed. And that means that while you're working on something for three months, a system that's gonna be free for free, not just what you do, but even more often, and then it's gonna fly to the next plate. The speed is very high. Now there's a subject that doesn't have to be done, it's gotta wait. And that's why I said from the start that many applications are better off.
In many applications, it's better not to do. Look, I'm trying two, two types of cabs today. On the one hand, I'm sort of a business-like version of something. On the other hand, I show you the real life-use that's important in life. Put the liquor down, please, again, if there's no video yet, and it's very useful for my video. Write in the comments now, that it's hard for you to understand that you're easily aware of where the information was new to you, where it already existed. Video, by the way, always give friends, it's also good for development inside. And if you haven't signed up on the canal yet and you don't want to miss the video, sign up, yeah. And we'll move on with you now, and, uh, forward. You'll be very likely to meet people who are talking about agents right now. What I want to tell you about the agent work, and we're also waiting for the agent release, which is what we have on Wednesdays on the special edition. Last special, by the way, I had Wednesday. If you haven't seen it, look what happens in Russia with artificial intelligence, how much Russia has fallen behind or has not fallen behind artificial intelligence. There was a career graduation, which professions would be ahead of me, which I did not. Look, I've been doing some sort of thing on a different request, and I'm also in the GPT chat room. Very funny, very nice. We have a series of special subjects. And if you've been on our video for the first time, we're going every Sunday, we're four people.
Special-- technological news, Silicon Valley sites around the world, we're telling, yeah. So, the agents are asking for a lot of prom. That's a very, very, very uncomfortable thing you're not gonna give you. There's a lot on the Internet of the Haip when they bring some things that can be done, and in reality they can't be implemented. For example, a hip that's been in over a year with many bloggers talking about it, especially people who don't really know technology, say that all the calls of managers can be analysed easily. and allegedly draw conclusions on how managers work. If you remember what I'm doing right now, you're gonna analyze what I told you before, you'll see, like, every sales manager, if you make 40 calls a day, and you have, like, 20 calls. sales manager, you get eight hundred calls a day. You need eight hundred calls, like three pages of text, two thousand four hundred pages each day, to get into the system, process and draw conclusions from them. Imagine if you want to make a picture, like the manager has, started working worse or better, or what strong and weak points the manager has, or the manager uses or doesn't use a system-specific one. Lexico, for example, or what he uses standard questions- answers. You'll see that every day's analysis will cost hundreds, and in fact thousands of dollars. And the system inside will be hallucinating, there's a lot of things going on. As a result, all these systems are based on low-cost models. That the use of ChatGPT 5 Pro models is very expensive, so easy. The use of o3 Pro, for example, is extremely expensive, o3 by itself expensive, and o3 Pro is very expensive. ChatGPT 5 is cheaper, but still. Someone's gonna say he's gonna use, like Deep-- DeepSeek there and so on. Guys, a quality question, b is the question of how to build this whole thing and get real systems that do the analysis. A lot of things can be done. We can set up a system. You're on the button, like, clicking and sometimes you analyzed some kind of plastic like managers. But the question is, is your system as fast as other systems? So, inside I-- is she developing the same way? Is she going in the leg with time? You've got the programmers all over the peddium. What do I recommend to do here? I recommend you unload your calls. We're like this in real estate business, I've been doing it all the time. On the team, there's a team not superb. I'm telling you, give me calls, like, give me some examples of the 50, 60 calls from managers. They've loaded them into the system.
Said to the system, "Create, compile a list of standard questions and answers that are, standard episodes that are. Show the weak and strong side that managers use." A super-crunch paper was developed with a bunch of analyses, conclusions, details. Unreally cool. But it's all about keeping it up. If, for example, the O.R. is not in support of this, then you're not in the middle of a problem. You know, you're supposed to be with it all the time. So I want to give you some very cool examples, the last ones that are at ChatGPT, I think, in the Plus signature, including the possibility of carrying out the slips. I recently understood this dish, very interesting. I was sitting there, I had to sign up for the French embassy, and there was a system in San Francisco that was changing. And I was at ChatGPT asking what system the slots were. He didn't really go into the system himself, loaded, made some kind of crypt and checked whether there were slots in the system in terms of days, embassies, etc. It's not funny that he might not turn directly to the system, go to different boots, find out where someone had information, and he gave me information, whether there were windows free or not, and there was no window. I'm saying, "Can you put a reminder, an assignment every hour, check and report to me, send me reminders when the place comes?" He made the system, he reminded me of it. I took it, I made a system like, "Tell me the information once a day, how many subscribers I have, in ToTheMoon, there, how many subscribers I have in my personal canal, how many subscribers I have on my way to the next. Instagram. Send me that information." He's in these systems, he's handling and sending me information. These memorialists are really a very interesting subject, and they can build interesting things. For example, I wanted to buy one-- in America, you're gonna have different numbers on the car, you know, in many countries, you can be put, but there's a very wide range of visual numbers and so on. And there's a concept like electronic numbers, and we can set e-numbers. And these electronic numbers, now only one provider, and they're expensive.
I'm on different cars, I have different numbers, and I'm saying in the system, "You can put me in a system to make a reminder, "there's, like, two weeks, once a month, and see if I've got a look at it. "The market is interesting?" And, uh, some new cameramen about the numbers. And when they come, let me know. And he wrote, when there's a discount, there's fifty, twenty-five, there, interest, plus, when the operator shows up, I'll let you know. So he's working inside the job. You can put these tasks on your own if you have a story or a plus, and it's a clear movement on the agent side that's gonna move. Another interesting case you want to tell you today about the agency work of the last one. Again, look, in the semi- manual mode, we worked with my editor. We have people there who help in mounting little videos, big ones, someone in there designs, someone processing and checking the text from the recognized text to the article, and so on. And I'm very sad, very much, periodically, as people don't understand how to speed up with artificial intelligence. Actually, what things can be done to date in video, images, large data processing, etc. And so I sat there and said that a lot of cabs are getting more and more automated. And I'm telling you, there's a case soon when we give you a transcribation from the video, like youTube, there, there, the clock. The system will set it up and put it in Notion itself, and it will be on the website itself because we have synchronized. For example, I have a site without salt, a point of which, it syncs up with Notion through a super-point of sio or eso, and then, well, a man thinks it's a regular site, not a notion. And then I sit, I think that day, let me check on how it works. Got it, got it, got it, got it. So you know, you can pick GPT5, plus pick an agent, choose that agent. And I'm telling him, I want you to take this big video transcribation and write an interesting material on it as a way to people, how to get this short and place this in Notion right here. He says, no question. You'll have to bail at Notion first. I say, come on. He gave me a dialogue window inside. I bailed on Notion. They don't keep these data, the passwords, and so on, they don't. And he went to work. I'm coming in in a while, there, 30 minutes or 40 minutes. I see there's an article. I can tell you that's a very good job. I wrote it. I want this article to look cooler. I want this article to be-- what was the story? That he didn't add any color gammas in this article. I'm saying, I want to make extra colors in this article. I want this article to show some beautiful, well, some pretty, painted stuff. Let me see, please, do this to me. He gave me that article, and he did some quote. Then I told him I gave him a job, and that's where he stopped and didn't. I asked him to add to me at the end of the article references to other interesting articles I have. Well, you know, when you usually read the article, for example, and at the end of the article, you have some interesting references to others. By the way, I thought he didn't, he did. He's got me some extra material. I'll show you that. How can a man understand if he's in a state of consciousness? Then he gave me a relationship without borders. Ah! I asked him to make reference to two or three interesting articles. I asked him to give a link to my video on YouTube. Guys, he's saying-- I gave him my channel. Guys, he analyzed my YouTube channel, found a matching video on this YouTube Channel and gave me a link to this video.
Look, he's analyzed again, he's analyzed.
He's analyzed again, he's analyzed my channel. I'm going in now, I'm going in. I'm just looking at it, so he's got a few references to the articles. I'll tell you, I'm gonna be down in the tape. We'll give these references mandatory, there, when we describe the roller. And I see three references to articles working, and the reference to the video somehow doesn't work. Maybe he gave a wrong link to the video. That, by the way, the reference to the video is not working, but the reference to the articles is working. He gave me an additional description. It looks really fun. I'll give you a good one, and I'll show you exactly how this page looks like he did, yeah. We'll publish it. It's just not published, it's a page. We publish this page, or we give you, scrubbings here, let's show you, yes, how it works.
Very cool work in terms of simple things. So, with regard to ChatGPT-5, what I'm saying is that ChatGPT-5, first of all, if you're a free user, you can use ChatGPT-5. He chooses what system he's gonna work inside. Now Sam Altman, he's been approached by a lot of people. There, as always, people complain about all the glues and everything. They said they'd be back for some people now, I think 4Os, you'd be able to choose, but in any case, she's got less hallucinations, she's stronger, stronger, smarter. This model is definitely working faster. It's a fact. I'll tell you the key advantages that are at all. They say they have a lot of cool chips in the programming, there, frontend, and so on. Although I can tell you that in programming, Anthropic is probably in super-liberers and, well, a lot of different systems, there are many different systems. And it's probably not a programmist graduation, is it? ChatGPT has more precise medical responses, and that's very cool. Many medical systems are much worse, right? So they'll have more pedits soon. They're funny, by the way, before ChatGPT when you asked for some things, he often refused to answer. Says, "I'm not gonna give it to you." They have a concept of a system of rejection. If there's a retraction, then there's a explanation why there's a rejection. That's really cool, yeah. Here. And when-- since this system chooses, it includes, for example, reasoning-- it's thinking, thinking, or not. There are limits. If you're on a free version, on a fee-paying version, you have a limit, right? So the text says it's more floating than a stylist, better keeping the complex forms, the verlibre, the drama-dramotorgy, so it's more accurately disbanding images, videos, diagrams. The accuracy of spatial considerations is improved. I understand they're talking about videos when you're filming and asking in real time. I'm waiting for them to get a video. I'm waiting for the video, by the way. I'm in the professional business of my own business. One of the chips that's in ChatGPT-5 is cooler, it's less of a false confidence. It's better to admit the limitations. The deception level is reduced from four points of eight to two points of one per cent. It's a serious, really, downsizing. That's a nice subject. I'm buying a charge now, there, for-- setting up a charge for the electric car. Finally, I got a cell phone that I can drive around the US without my hands. He's rebuilt himself, the map understands, and so on. And I was charging. And the GM charge is Cadillac's company, yes. So, it's just the worst interface. And ChatGPT-5 says, "Attenly, the system is completely stupid." Well, very badly made UX/UI. So she clearly recognizes the limitations differently and says so to you. So this system has anti-subjection because, earlier, you know or not, the system has often been a little unsuccessful, just to accept you. There are some concepts of reducing psychophandy below six percent. It was four and a half percent before, right? They've made four personalities ready, and they're available. I have a scattered scenic, Robot, Listener, Nerd. I'm still using the usual. They say they've got better instructions, but I understand they still have a flight. What's the custom instructions? These are the instructions you're putting into the system at the beginning. I can see that the system has a lot of problems with it. Because I have a prototype, the system is analysing a lot of old chat, but she still can't analyze all chat chats. You must understand too much information, she has a narrow context window. You must be clear that she has a limited volume on the entrance, and you can take a little bit of a way out.
Even if you-- you gave her a document, like three thousand pages in the PDF, she's not gonna be able to process all three thousand details, she's gonna take the fragments, right? You must understand that. And in the proversion, it's a version worth two hundred dollars, I think two hundred is worth it. I've been telling you a lot of examples, of course, including. It's a big thinking, isn't it? She's a leader in some places. There is still a conclusion and a table in the view of these benchmarks that ChatGPT is still lagging behind Gemini, even falling behind some Grok model. See who's the first to reach the human level. There's a level of humanity. Let's see when it happens. Well, it's a graduation today. Please write in the commentaries, how do you have the impressions of such episodes as you're going to give us any more topics? Did you like this graduation? Again, the subscription, the like will be very helpful to the channel. Let's especially let this video be a lot of us today, especially those who have seen it before now, see. Let's get this video on a lot. I really want the channel to grow. It's hard, of course, always in-- in all the sets. The more you write commentaries, the more you can-- the more you send friends, the more the video gets. I'll see you every Sunday. Bye, everybody.