Hello, everyone! We're on ToTheMoon. Technological news, Silicon Valley sites around the world. We're going out every Sunday. It's the forty-first issue. Join us! The amount of data available for training is almost exhausted.
It's not gonna be a huge scale.
He gave people notices of the missing news.
Chinese models, which are specifically trained in English, may not be subject to the law.
There's a separate autonomous agent. It is designed to carry out the complex tasks in a self-sustaining manner.
There's no special service.
Well, the question is, should we be chasing a fall, yes, price? I mean...
People are just tied up with bonuses.
I'm really curious how cycling the world is, aren't I?
I got hit by a vacuum cleaner who drives you and your iPhone wears or brings you a beer. I've got something to do with a crisis of some kind.
Last time, I recall, we've been taking a very cool issue of the 12 major global technology sites that will affect technology development in the twenty-fifth year, especially as it relates to the development of technology. ordinary people in general, and everyone who's interested, who's interested in technology. You'll have to watch the last edition. Very cool.
Hey, guys! There's a subject I want to start with. Elon Musk, there was an interview where he told me that there was a pic of data at the moment. The volume and peak of data for AI training, yes, that the amount of data available for training is almost exhausted. And he talked about it, including the former Chief Scientist OpenAI, Eli Suzkewer, that the world has reached its limit and that now. What does that mean for-- that's all that's understood. So when different models are created, they learn from certain data. So we told you, for example, that Harvard had a million books there, I think, Microsoft for training or that OpenAI was on the data in open sources. The pros were checked by people, people, staff, like, from Africa. The whole Internet has been written over there for a long time, and a bunch of libraries and everything. And now what's going on? Models have started to be developed that are taught on synthetic data. What do you mean, synthetic data? It means that these data were being generated by some artificial intelligence.
And it is that there is one of the problems of synthetic data that can be the risk of a model collaps when artificial intelligence becomes less creative, more biased. And plus, he has mistakes, yes. And these mistakes can cause learning problems. What happens in terms of data ahead? It was understood that, yes, artificial intelligence would start producing a large number of data, start to produce a large number of articles, and in fact, at some point it would be impossible to even determine what was on which I learned where, what I've been generating. So, some model, you might say, in theory, she's been learning from some sort of thing-- I don't know, Internet data, Internet data are just unbelievably saturated, including artificial data. Intellect. And we see the articles that appear when they write that, for example, Facebook has literally now closed a lot of profiles that were a profile of artificial intelligence and, like, Apple was literally a week ago, it's a... I've been making news and, like, the summit on these news. And there were hallucinations. Accordingly, he gave people notices of non-existent news that is. I mean, it makes sense that when these data come, they're getting someone on them right now, right? What is your vision of what will happen next in terms of all these data?
I think that this is an incredibly important question from the perspective of humanity at all.
Two or two years ago, maybe Forbes and some of the tech companies were conducting a study on the influence of artificial intelligence there, well, on the CIS market on education. And that's where we were, too, involved and reached what? That there's a reference date, that's a dateset before the AI era is created, yes, where all the data are on the par with people, is intact. It's like, you know, there's a reference kilo in Geneva, I think, a reference meter under the cap. We know that this is the metre. It's not a abstract thing, but this is the specific reference. Well, they're doing differently now. But it's been a long time. And in the same way, the dateset is likely to be a turnout that is not, for example, contaminated with agenerative AI. Why is that important? Because if, for example, a model of genetics with a slope on some side, it is so positive or, on the contrary, negative feedback. The loop is like that. He generating information in a direction, then retrained it, and it's all oriented. And he's got himself. Actually, that's when they say about the collaps, which is what this thing means. Therefore, the emergence of a reference dateset that should be taught and addressed is probably not that fantastic. We'll probably get to that. Although it must be understood that the date is limited in any case. It's not gonna be a huge scale. And here, Ilya Suzkever is probably washing hands and saying that it's good that I started a company that's looking for other fundamental approaches to modeling. Here. Because the currents can be exhausted just because there's no data to scale indefinitely.
But it's probably not even a question of scale. The question of correctness, including how to check. Because the reference dateset is too much, I think the data on how this reference can be checked and how the reliability of this one can be checked. It's like a story, right? Very often, in different countries, the history of the same event is very different. And even with different dates. I'm not talking about people who won, who lost what really happened. That's where we're half-- we're going to a new one, because information is just some endless lot, we're going in there, and there's a multiplier inside. So the correctness of the information becomes...
As math, Sash, I live in a world of pink ponys where everything proves to be, everything
is clear.
You think like math, because there are laws of physics that, as you know, there are no exceptions, but there are some philosophical questions or there, well, something between that, which can't be a reference date. Seth.
How important is this for development, Sasha?
I think that this is our prism through which we look, it will not be applicable to generative AI, because now, by implication, the ameri- Chinese models that are specially trained in English, They are quite good, these Chinese models, they will not really be obeyed-- they may not be obeyed by the copiate laws, they may not be subject to any reference. I don't think I can imagine what the solution is, but any restrictions are likely to just throw American companies back because they're gonna have to follow the restrictions, and Chinese models are probably Not a-- not a-- you don't have to follow the restrictions. So the last time this DeepSeekV3, it was completely naive, which led to her being there, saying she wasn't trained on ChatGPT data, which I was like, a Chat model. GPT, yes, because she was so incredibly trained. But I think it's not like that, but it's because nobody's serious yet, well, as long as they're serious about the models, not so many companies, but in the open source, I don't think anyone's really training. I think the next generation of closed models will be very clean than this generation.
What is important is that new training data are important to, for example, create AGI or go further in terms of revolution? How does that even matter?
Because data in general, why is this even a problem? Why do people think, well, look, if she's a smear...
Well, people say that AGI is a system that improves itself, right? I mean, for example, I mean, by imputing, humans are falling... well, like the number of people. So AI is not limited to the number of people. And AI can, for example, find new ways to train itself differently and so on. So it's all very interesting concepts, and, uh, it doesn't need much more data for their development, does it? That is why it requires very specific data on how these models are being trained. Yeah, I mean, this is RLHF data.
But it's like if there's any data available, if some AGI is created, and it starts with these data inside, when we say it's AGI, it's gonna be synthetic data anyway. If he's independent of the man, he'll create something new, it'll be synthetic in any case. And there's no problem...
Yeah, yeah, yeah.
In synthetic data, because everyone wants to create someone who can generate synthetic data.
Yeah. Yeah, yeah, yeah.
But now your synthetic, she's exposed to hallucinations, and on the synthetic, learning from it, you get problems.
Yeah, but basically when we're talking about AGI, it's what creates new synthetic data that many people will be hallucinating-- how to make a mistake. And many people will agree. Just like some man, he's talking about, uh, some kind of performance, I don't know how to do business or how to start kids.
I don't know how to do business or how to develop children or how, uh, plant. Some other guy comes in, says, "I'm totally different. And it's like we're making this mistake as a sort of mistake. Look, it wasn't. Check this information. But with things, uh, such as scientific, inaccurate, it's basically any AGI--- yeah, AGI says, "I'm just saying that I'm supposed to be. Yeah, I just thought that's what I should do." And there's, uh, there's, like, a hallucination block where they say there was a president like that in the United States, and that president wasn't there. And the other point when AGI is generating a synthetic in terms of, I don't know, human development, there, or educational methods, or some extra stuff in terms of, I don't know, he's gonna say that, you know, that's what you're gonna do. The next medicine, there's three hundred and sixty-five days, yeah, some kind of test there. I think it's gonna be totally new. And there goes, there goes a separation, which is hallucination, which is hallucination.
At this point, the model itself cannot describe where it works. If you go back to the world of pink ponys, AlphaZero is a move from Google, from DeepMind, which plays chess very well. What happened there? They described the model. You're gonna have to play chess like that. The valid moves are like that. And they did. She was playing with herself for a few hours. There's a ridiculous number of parties that played with themselves, just observing the rules of the game, so there's no hallucinations that the horse walks like a man and so on. And by creating a model that works within this, the conditional AGI has been made, yes. She learned to play herself. New strategies have emerged, which have not yet been applied, have changed attitudes, positions and so on. I mean, it's all been there. And on a big, unlimited world, unlimited, just chess board, it's not yet working.
Apparently, whoever can do it is that that's all that and that's how to make such a revolution.
Well, here I want to add. You're just talking about this self-learning, that the CES show is now 20-25, 25-five. There were different technological developments around the world. And at this NVIDIA exhibition, the technology of the autonomous players was presented. What does that mean? That when a man plays a certain game, there are characters that, well, they're called non-migrant characters, yes. They have the term NPC, yes. And in fact, there are some characters inside that they used to be programmed, his behavior, that character was programmed. They have now presented models that allow for the creation of characters that will develop themselves, that is, do what they want to do inside. And I don't know how to create viruses or, on the contrary, like some, uh, positive, negative viruses are different. Anything can happen. And I understand that if I run my characters, I'm supposed to have limited action earlier. He might be in this self-standing, a character playing here, if the NPC is a non-growing character, yes, I don't play. And this play character will also evolve around some of its models, yes, how to live his life. It's not Tamagochi who lived there, we're two buttons, three pushers, and he's gonna be. I left him on a daily basis, I don't play them. He's developing himself. I'll be in the next 24 hours. There's something new up there, anywhere, and it's all over and moving. And that makes sense that it's limited in the game. And that's a lot of different movies. Here. But the subject, here, development and learning, on which data all this is learning, yes, what happens next, it's crazy interesting. And I want to get some more medicine. Interesting, like, boxes were right now.
I was just going to the tooth a few days ago, and I was told to do, uh, like, the channels, yeah, in my tooth to treat the channels, that I had some untreated channels. We gotta figure it out. And I had a question of this kind, because there were three different doctors, they gave a similar, but partly different, data. I took it, I went back to 4O about my photos. He said, "If you tell me, please, I've been given some information as correct as it is?" And of course, the quality of the conclusion I received, well, it was amazing to me because the system gave me a lot of information. More data and details for me as a man, so I can make a decision. Do I need to go consult with some other doctor, or do I have to trust this doctor? So, this is the picture and the details. I guess it was impossible to check. Actually, well, that's impossible. Because it, it told me the information that, for example, the current doctor didn't tell me, yes, that you're seeing some darkness there, you're probably in there when you were a kid, and you're probably already treating it, There was a problem, and that's a story, that's a story, that could lead to something like that, that kind of a thing. I mean, the level of the incredibly detailed. The Germans are doing research on breast cancer. And as part of this study, they , uh, tested four hundred and sixteen thousand patients, and there was about 20 percent accuracy, the accuracy of the diagnostic-- that is, a very large sample, a very big sample. The accuracy of the diagnostics from-- using-- when artificial intelligence was used, increased by almost 20 per cent. Well, that's a lot, yeah. From the point of view of the extra details, a man is clearly less and less able to determine, you know, some incredible gap. There will be some agents or self-study systems in this part that are, again, learning about some data.
Yeah? Sasha, what do you think about OpenAI launching this Operator or want to launch Operator?
I mean, it's, uh, it's not-- it's just independent AI agents. It means independent, autonomous, right? That's what I'm doing right now.
Yeah, that's, yeah, it's independent autonomous agents. So it was before in-- it was like some code where an agent could be made, and then they say, "We already have a separate, uh-- there's a separate, autonomous agent. It is designed to carry out the complex tasks in a self-sustaining manner. I don't know, can write a code, book travel and use a man minimally.
It's like Google Deep Research. What Anthropic did. I mean, I think we'll be seeing only the first versions that aren't exactly the same as we do. I mean, I'd say I have very low expectations from the first quarter of this year in terms of the agents' work. But the level of the agents' work at the end of this year, I think it's gonna start to make sure that there's a lot of interest. We'll assume that the number of people who have completely changed the business process is measured by interest, right? I mean. I mean, not that they're replaced, but the process will be completely different. I think it's gonna happen before the end of this year exactly because, in fact, the three most important players are putting a stake on it.
They started talking about it in November, and they started talking about it in November. Now they're talking about what they're gonna let out. Yeah, yeah, I get it.
What this operator will release, yes. I just didn't even get a word with that, yeah, it's pretty straight.
Yeah.
And they're straight--
You know, it's important for them to say first. First, it was Anthropic, then Google. OpenAI is as old as as it is, but it's important for them. I mean, they were, they weren't the first to say it. But actually, we'll forget in three months who said that, yes, before. Who's gonna do the first question, huh? Maybe that's how we're most often left behind by not the first one to say, but who did the first thing. And OpenAI in this respect, of course, is not behind anyone. Because Google and Anthropic decisions don't go anywhere, in my view. I haven't heard a successful example of the introduction, so the OpenAI has the chance to be the first. As usual, they always lead.
Please write a comment on who's watching us now. In the commentaries, write what process or function, or what task you would like to trust in artificial intelligence. Imagine she would have worked. That's real, yeah. What would you like to trust? I mean, I know there's humor. I'd like to trust that I have artificial intelligence, I don't know, bought my kids or walk with a dog. I'm not talking about that abstraction, and I'm not about robots, but about some real subjects. Write them. It's really interesting to see the list. We'll even put them out separately, or in a comunity, or we'll figure out the comments. In the notes under the video. It's gonna be really cool to see. And if you've ever done anything or you've made a bot, well, it's gonna be interesting. Or, in fact, some agent you used something, it was really a job to do.
It'll be great to know. Interesting story is that, in general, agents and bots are talking about travel orders, but one thing is to write a code, something complicated for people, a very specific topic, when we are We're talking about ordering a trip, and how much man comes to doing this. I mean, I'm using, for example, you and I use a lot of AI, but I don't think any of us use an agent who's doing his life in parallel, yeah, well, that's what's really doing. And there was a study from Adobe. They said that the purchasing power of Americans had grown very much last year, and especially in November, December. They've been studying the period, from 1 November to 1 November.
They studied the period, took from November 1 to December 3, and that the retailing web trawl that came from artificial intelligence chat rooms grew by 1,000,30%. And so, the cyber-week rate has just increased to about 2,000 percent growth. I see there's probably a little chat room in the past year. What does that mean? It means people trusted their purchases and the system bought through chat-bots. I mean, something happened. What's interesting is they don't give the systems for these chat-bots because what are these systems? And you realize that these are chat-boats that must be created somehow. It's likely that these purchases were in the retail business, including those made, I'm guessing, people who bought something for resale, you know, or for some benefit, because ordinary people through chat-bot are doing something. I don't know the people who do it through chat-bots. Here. And the trend itself is very, well, it's starting to move. These are still programmable things. We'll see if this year, one of us or any of the people we know will start using an agent for real, which will start to do, well, some kind of fun, real action. And it's not gonna be hard, it's going to be a little real. That's gonna be very interesting. OpenAI, by the way, as the audience sees OpenAI-
Of course, Sasha will, of course, be. I mean...
By the end of the year, huh? By the end of the year.
Right before the end of the year, yes. I mean, for example, we-we were looking for a way to get cheaper than just to fly from Europe to San Francisco, yes, from any city in Europe, because in Europe it's easy enough, yes. Oh, and now we need to find a site that can, like, behave like this, logic, yes, support that from any city in Europe, and, well, that takes time. I mean, it's gonna be all over the year, and it's gonna be working. I mean, in principle, it's not very sophisticated technologies in that the auto-filling on the sites is often already working. Google's already got data, you know, they're on it, so they're asking for a name, a name, a address, yes, maps, a bank card. And I think that's right, we'll see it before, we'll definitely see it before the end of the year. Why else, huh? Because there's a way of verticals that bring too much money. It's not a secret, is it? I mean, it's travel, it's e-commerce and it's finance. Too much, uh, it's like, say, a way too cheap. It's a fairly narrow and sufficient cash opportunity in a con-- in a consumer for all three giants. Here. And also including Perplexity, which will do it, too. Oh, I mean, I mean, yeah, I just did, just roll up another chat-bot model with a resoning for everyone, right? And that's a little controversial, because resoning is expensive, right? Well, this general resonance is expensive. And as Altman said, they lose money.
I think these verticals are very profitable, and they're gonna try to make a decision as soon as possible.
I want to add, Sasha, you said the subject about the flight. It's a very, very, I think, live story. There is a Flight Connections site that allows for the view, for example, of which country aircraft from Europe are flying to San Francisco. I'm surprised if we really start going through all the websites that sell tickets, that's not what we're doing. I mean, it's almost hard to find. It's got to go in on a separate, often searching for a different airline. There's no detail. There's a site like this, I once hit him, and we'll light him inside, let's show him. It allows any city to click and see which city, for example, where to go, how many stops. All the routes to be made, yes. A lot of different cities. I mean, to make any deep-blue analyst, yeah. It's not funny when I can get out of a little airport next to me, I have a little airport in San José, I live in a cremnieva-- a little south of San Francisco, in the Silicon Valley, see where I'm at, and where I'm at, The world is flying these planes, yes, sending them there, and they're drawing very beautifully. Many sites don't do that we, we use in our lives because they don't need it. If they do so and use this easy way, people will buy the most advantageous tickets, the most advantageous routes. It's like Amazon, it's a shitty search, actually. Can they improve the search? They can improve search, but then people will buy the most advantageous goods. So everyone who sells the goods to others will not work. It's like a commercial, right? If only the best people can show, or the only interesting ones. Who's interested in that? Something that is, for example, to change this industry, change the industry, someone has to try as if it were, create a new, totally market. We are at the threshold now, of course, when in a vast number of industries this happens, and, well, as far as all sides are concerned, everyone, including professionals, is not professional. Someone might have a chance because, for example, in the same tickets, I think standard operators, they're like, if they do, they might lose their market, yes, here. Right away, right now. They're not interested in showing only super cool... Well, it seems Sasha's a light case, like, how to get out of San Francisco, for example, the cheapest. And if you say you have a wife, a child, and you also have some extra parameters, some aspects that fly, which class we, fly business, not business. Are we ready to spend a lot of time or not? When we're ready to fly, we're not ready to come. I mean, give it to me, but I'm not going to these airports in Europe, but I'm going to go to these. Well, how do you ask for a little bit of a complicated request? And he's looking for you. There's a company like Kayak.com, which allows for the search for outlets from different cities to different cities, and the construction of different junctions. But it's a very small, you know, a little restriction. I mean, I think that in the world of these boxes, there's a crazy number now, just a lot of them. Which is what is being done is another matter.
But there are many.
Yeah, yeah. I mean, yeah, like a lot of cabs. And I think the main thing is, well, all that's related to travel, yeah, it's-- uh, pro-existent consuser market, yeah, it's just a huge consumer market, and, uh, where, uh, say, no--- k-k-whatever. It's the most standard pitch, yes, it's always been. Here we go, uh, aggregating all the services to plan travel from and through. But no one could ever do that, right? Well, it's the most standard start-up pitch that everyone's allergic to. But I think that with AI it's gonna be possible, actually. I also feel like it's good that all of these APIs are already in existence, all of these data are already standardized. Ah, I think there's a lot of historical data to re-check something, right? Uh, I mean, I think it's really-- it's possible to minimise hallucinations.
About Aviasales. There are resources that, well, um, are on the rumours that are used a lot, but there is, of course, AI's appendix can change the services they can provide.
Well, Aviasales, yes, very standardized, just like they give you, uh, no extra water. I mean, I don't know what it is... Sasha, you can talk about it. I just want to say there's no special service there. He gives you the same thing as Kayak and Aviasales is the same plus-minus.
It's used in America,
yeah.
Yeah, in America, just like Google Flights, Google really gives it a great deal. Well, I mean, Google-- and they're not much of a lot of things now adding extra possibilities inside, but still! There's no simplest parameters. I mean, imagine we would make a good classification and ask questions: what really is going on, right? Where are you, where are you planning to be? So, normal extra--- they don't even give a good deal of extra questions. And it's visible in-- some very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very, very I think we're in a world where we can add much more, question a man, and they don't ask for it, and, um, just while we're looking for some minimum transaction, because it's simple because it's simple, She's simple, people often need a simple case. People don't want to get excited, people don't know how to buy tickets. That's what money's made. People can't take hotels. That's what money is made of. People can't make advertising. That's what money is made of. People can't make choices where they're cheaper, where they're more expensive. That's what money is made of. We thought we had a surgeon who was making it cheaper to buy. But you're tied to this aggregator like Amazon, and you're just not checking out where it's cheaper, and you're just buying it from them because you want to do it. And you have one position on Amazon with four or five different prices. It doesn't have to be that kind of thing, but it's a post--- it's a lot of sight, right?
There are other loyalty programmes. People are just bonused with some airlines, and they're there with three triple-price transplants, but they're going to fly this one. And somehow, Sasha, what he said was standard pitch, but I never encountered an app that normally either boozes or finds hotels or, uh, cars, Uh, rent. But for all this to be on the same platform, I never get that kind of thing, even though many offer as an additional option, but they're usually good either in one or the other. I mean, even--
The market stopped! The market stopped ten years ago.
They couldn't have been the basic model, yes.
Yeah, the market stopped. Airbnb released, I remember what Airbnb had-- this very cool new search, a new UX/UI that allowed all the interesting stuff to search and find, uh-e-e-e-e-
Search and search, book, for example, housing. Yeah, and all that wow! I mean, first, it and for years, it was super cool, and then it's the same degradation. So it didn't go into some super cool stuff because there's a problem there, right? I mean, if you start giving only the cool spaces, only the right prices, only the stories you've been tested, you know, you can filter at the price. Many services often don't get a good look at the price. If you start doing this normally, you'll basically lose a lot of other castomers. And so those people who are housing are renting housing, just like the Amazon. As long as you're in charge, you're showing a tomato, a tomato, a tomato, a real tomato, seeds, and you're even, you know, and you add. You've got a search problem all the time. I mean, good search. I've been looking for a lot of food. So far, a great search in foods that would allow for real use, as if it were done for a human being. So, people get the opportunity, they say you're used to buy two products, so we'll recommend them to you, too. And this standard is the same, so it's the same.
And you're just being put into an era, in the era of current, current technologies. It's really interesting that you'll show up or show up, right? So, this is how the market changes, what services are going to start to be real, that will change? I don't know. Last year, such services were in general, which led to something radically new in normal, ordinary life. It's not about the video-engineering. Probably not yet, huh? No one, no company shows. I guess it wasn't.
I don't think so. And now Perplexity has first started saying that they're going after finance, travel, e-commerce. I mean, it's a specific vertical, which obviously makes it possible to make visual, rich even chat interface. And that's why I feel very, very much, very promising in general.
Well, by the way, the search must have really shown that the new one showed what Perplexity had shown to look for, it was the OpenAI search. What was shown in the search showed something else. When we can find a friend through chat-bot because when Google showed it, or did Yandex show it to me, it looks really raw, doesn't it? Somewhere on top of it, a little poop of aggression that was shown and forgotten, huh? It's not developing. So there's no such thing as browser and browser constantly developing, and they're super perfect. I don't know, Chrome or Safari have become a space-based tool. He could have developed very much. I can't see that. Here's a chat-bot, he's probably growing inside. The search is still on the move. Well, remember how many statements Bing had? We heard a lot of statements, and then they stopped somehow. I don't see any progress at least. Maybe I'm not a Bing user, maybe they've done a lot of things there, but somehow it's all stopped very much.
Well, I'm not quite clear here because, after all, Google's extradition is much cheaper than the release of the same OpenAI. So you have a request yourself through a big, genius grid, it's much more expensive. And if we're gonna come to this, not just on the dotage, but when the company pays most of its profits or something to show the technology before it gets there. self-propelled. It's either price rise or cost reduction. Here's inference, yes, what's called, that's the chat-bot response. I don't know what this is going to be about.
Well, is the question to be chased for price reductions? So, the news was, like, that one about the alleged loss for the OpenAI. What's new is what the point is, if it makes sense that GPT is only lost before, and OpenAI doesn't plan at least $100 billion to earn it up until the twenty-ninth year. And logically, that's what's gonna be invested in. I don't think they have a question now, you know, profitability. The question is how much it takes to do, how much support should it really be, huh? And what you're saying is, if you've created a browser like that, right?
Or created a chat-bot that is everywhere, spreads everywhere, that is, through a normal person.
And from this point of view, it's really great to know that the OpenAI is now actively looking for a development front, right? I mean, it's like AI's thinking this morning, but they're hiring a development front. But I think the agents are, they are, they are, uh, agents are there, and they need rich interfaces to make them feel good to be involved, right? And so OpenAI, I think that's why OpenAI needs the front liner.
It's very interesting how the world is cycling.
Sorry, they're very, very sorry, they're still UX/UI, sorry, Sas, that everyone who interrupted, but sorry, they have UX/UI better than anyone in general. I mean, from all the AI videos.
Well, you mean just chat-bot? You mean just like chat-bot?
No, no. Sorry means, right?
Oh, sorry. Sorry, sorry. Sorry. I thought I was sorry, but they have more. Oh, in Sorry, yeah, in product Sorry.
They have new solutions there that they have.
Yes, I do. Sorry about 100 percent, yeah, yeah. It just doesn't work.
I'm not sure it's just like absolute. I'm not sure it's convenient, but there's definitely some innovations. They tried to make UX/UI.
Yeah, yeah.
Yeah, come on. I'm really curious how cycling the world is, aren't I? So, in general, it all started to communicate with computers through the terminal, through the command line when the crews were just writing, something was done. Then, now, we're working on the chat-bots, too, talking to the system and going on somehow. I can't help but share it. At the CES exhibit that you started talking about, you showed the machine to make gas.
And I don't know if you're up or you're not getting the Soviet syphoons when you could just get a gas tanker to make a house. And it's been shown in the twenty-fifth year as some sort of invention of know-how. I was shocked, of course. It's so cycling that at the exhibition of the achievements of the twenty-fifth year, the same syphony is shown, but a little in another design.
That's right.
It's a matter of size, I think. There's literally a lot of things in America, as you remember, Sasha, we've been puzzling about it, that typical American, he's coming with this aluminum mug. Here, and they invented this, yes, because we need to observe the water balance and all that, protect the environment. Which is basically just for me. But this ciphon, he was that size, probably at the Soviet time, I don't even know. And now he's just like that. It's a little rusty, and, well, it's convenient, I guess, well, I don't know.
Well, I, by the way, agree with Elnar that some things, to be honest, are very strange. So one thing shows some new prototype robot or says we've developed some kind of glass technology. Very cool glasses, we'll talk about it in other things that pass near certain images, like, and nobody sees how they write, right? Well, I mean, there are some technologies that are very interesting. Another case, when they come, even shows the computer that has a screen out there somehow, changes more. But, to be honest, it's just not interesting. Or someone says, "We made a new or new TV on the wall. Yeah, a new TV on the wall that has less, more wires hidden and a screen. You think, "What kind of body-- what kind of TV, huh?"
Yeah. There are other ways to hide.
Yeah.
I got hit by a vacuum cleaner who drives you and your iPhone wears or brings you a beer. I've got something in my head that's a crisis of some kind, I think.
Something, yes, something very, very strange, very strange, very strange, which, I guess, is not, well, totally not in the era, even what we're talking about here. I mean, if you see what we're talking about, well, that's what it's like, it's just some kind of parallel movement, right? And that's what's coming very hard. I mean, I, by the way, asked ChatGPT to be a very tight issue, this exhibition, to gather news that was the coolest. And he's been collecting, like, absolute shit that, uh, I don't think it's interesting to discuss, right? Well, I understand you can discuss TV if someone chooses a TV or can discuss a new table or discuss a new sketch or a new vacuum cleaner, there, a robot. But I think that when a robot vacuum sucker still glows when the map is being developed, it's really hard to do.
Or can't get down the
floor.
And it can't, yeah, can't get off the floor, from the step to the step, right. He's stuck in a lot of places, and they're showing, so, a robot that's supposed to raise and move it. I'm just getting funny because I know there's a lot of things in the world trying to make a hand that can take any detail, and they're driving around in a robot vacuum-sucker. Yeah, he'll be 100% of the curve. And it's a very strange skies when the world develops robots that walk and they wipe the walls and everything. They're in a robot pep. Well, I mean, I don't think that's the hand. I mean, you just need to improve and do this thing super cool with the design. Well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well,
From the point of view of the construction. Well, look, there's one simple function that doesn't work there. You can't change the room that was scanned. So you can't tell him where the room was scanned. If there was a stain of the mirror, the glass. That's who knows who, for example, has a lot of mirrors at home or glass, thinks it's an extended room, and you can't change it. And a perfectly elementary, stupid, stupid script that's not solved in a vacuum that buys tens of millions of people in the world. Yeah? And they say, "We're gonna have a hand right now." Yeah, yeah, yeah, okay, and my baby's over there.
"He'll pick a landing." Yeah. He'll pick a landing.
Yeah, he'll pick a landing.
Something's gonna clean up.
Yeah, and he'll take it, yeah, and he'll take something. And if it takes the electric wire, it starts to blow it out, and there's a short circuit, right? So, or will you pull some more story, which-- well, that is, if there's no way to decide, how do you decide this step? Oh, yeah. That's the same thing, I think we talked about travel, yes, about the fact that travel has not solved the question of such a cool, interesting search. And in parallel, people are talking about some space ships going to Mars. And, uh, I guess there's a very big break, a big break, and it's gonna be great, as we said today, if the AI agents come up with interesting scenarios by the end of the year. Please write whatever you want and think that the services you'd like to automate in your life, which you might be rabid or willing to do, as I have now given the example in Roboroke. The point of view of the inability to change the room will be very interesting. We'll meet again in the week of ToTheMoon. Technological news, Silicon Valley sites. Likat, please, our subcast, ours, ours, ours, our canal is very supportive. If you first joined, don't forget to sign up, share your friends. Until the next meeting in exactly a week. Our blank-- always- four. Bye, everybody.