Hello, everyone! We're on ToTheMoon. Technological news, Silicon Valley sites and the world. A-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-day, once a-day, Sunday, Sunday, special Wednesday, Sunday, Wednesday special prom. Oh, please, write off who was watching our special prom for futures from Wednesday. If you didn't, look. A very interesting subject, perhaps one of the most relevant things in the coming years. What happens to the professions when the artificial intelligence comes? You're making more than $10 million or less?
Can the company defend itself against this hijacking of employees?
To be a slave.
It's a black box, and you don't know how he's gonna behave.
The interesting aspect of this experiment, that at some point, uh, he, this Claudio, decided he was human.
I think that's a very interesting subject that, uh, I have to reflect, it's the first time last week that the car that was ordered online, brought herself home without a driver. So, uh, Tesla started delivering cars without, uh, delivery. So the machine's coming out of the factory, uh, and, uh, we're just gonna show you what that skate looks like. And she's totally going to town and comes to the man who ordered her. He's getting a notice that she's here, so he's coming out and meeting her. And plus, uh, Tesla has started her cybercabs in Austin now, Texas. It's a new cab, a drone, and they're starting to test it. I was sitting here with my daughter, uh, a few days ago, uh, with my big one. She's 16 years later, and she's, uh, six months later, and she can start, like, a little more, nine months later, starting, uh, learning, driving. I'm saying, "Look, what car do you want?" I'm saying, the most normal car is Mini Cooper's double-double convertible, I say, for the girl.
I'm fine.
We're sitting with her, we're watching this car, we've talked about something. She says, "Yeah, I like her, so I like her." And I thought, "Well, I like it, too. I'll ride Mini Cooper, too." So I'm back.
Intriguing.
Yeah, and I'm back. I came home. And then Paulina and I went, with my wife, and, uh, we still have four kids. And I'm telling the three of them, and they're probably not gonna know how to drive. Well, I mean, the idea is, you can learn how to drive to rent a car in some other towns or there, I don't know, they're going to Europe, and it's not a fact. Because apparently, the way things are going on in the cars now, well, there's an acceleration, of course, in three or four years, there's still a car in which you get in the back and tell us where to go. Don't you tell me where to go, like you don't drive or drive, right? Aaah, when you're back, uh, sitting there and-- or you said, "Get this kid in this school, come home, there, this kid is taken, and then you go, get some things to smoke--- well, courier or something. Do it." So it changes the process, uh, totally changes the process of interaction. And what's interesting is that we're a c-
There may not even be a rudder and a pedal. I mean, it could just be a situation like this that you're gonna have, you know...
Yes! They're probably gonna lead, yeah. Who-- who-- remember, we've been in the middle of the graduation, yes, what Tesla presented last year, and that was last year, right, the cybercab? Or early this year, I don't remember. And these cybercups they want to do, there's no steering wheel and no pedals, no rear-view mirrors. It's just that all these things and adduces are necessary now to consider the car that you can drive on the road. I mean, it's a law to release that's allowed to be allowed differently. But I had a new feeling in my view here, and what's really, uh, technologically, if I see it now, uh, more or less is getting ready.
And what's interesting is new operators are coming. I saw Xiaomi unleash the car. Have you seen it? Xiaomi released the car. They're in competition with Tesla and the third one. And you know what-- remember when Xiaomi showed up, then Xiaomi, uh, you thought, uh, some Chinese brand, then there's a sound of a blowing brain, right? And then this company, which had two hundred internal sub-brands inside of it, has a lot of stuff, and there's a hop, and the car's already driving. And that's...
Mitree is my first smartphone. I really liked it.
And, and clearly, it's gonna be the same car, uh, drone, obviously there's gonna be the same car that drives on your call, somewhere, obviously, there's gonna be a robot like that, yeah, that makes certain actions. It's perfectly normal. By the way, there's a robot that, uh, is a nice robot that's taken home. You have to show him where the cleaners are and everything. He's cleaning the toilets, there, washing, uh, walking, putting everything in the dish. Okay, at one level, at one level, but robot--
The point is, he's not gonna be soaking this dishes, Sasha.
Here.
And
glasses.
Look, sometimes it's a little confusing how much technology is different. I think I just spoke to my partners from Europe on software development. They say, "Sasha, we are so far away, uh, when we're here, "Sasha, you're in the Cremnival Valley." I'm there, I have some other interesting meetings there before night, and you're kind of in all the different conversations, fashion things, payrolls, where $300 million is a bonus to an engineer. They pay, they pay, and so on, yes. We'll talk about it now. It's just a bomb, I think. It's a bomb I think is happening. He says, "We're on our own," he says, "we're driving, and there's no, uh, we don't have any, uh, we're not, uh-a-a-a-- really, like, a little different type, a little bit different, yeah, like, "there's a serious
reality."
But the launch of a drone taxi in Austin at Tesla seems to have been a little slow. There are NHTSA questions, and SAE has questions, it's working, Tesla, and there's a lot of videos that they're going out there, somewhere they're going to stop, they're gonna drop passengers and stuff. Anyway, Waymo looks more reliable for now.
But how much did Waymo go? I went to Waymo again this weekend, and, uh, it's not that secure, you're kind of totally in this car, uh, totally confident. And that's how cut it is, it's already worked. So what did I say about the technology that the remaining people were?
I'm sorry I wanted to say, I went there yesterday, and I was watching, uh, Escalade electrical new, and it was so interesting. He's got doors automatically opening from the button. So if you open--- standing outside, the door stops, and if you're standing inside between the door and the car, the door squeezes you and doesn't stop, it's got to cut you in. And they say the apdite is gonna be released. And when I heard it, I thought you'd know what the difference is... how could you even do that function without thinking that a person might be inside? Well, I'm a base engineer, I don't know how to let you out without thinking. I mean, they're a little bit of a chip. It's unique enough, well, like-- like they're opening doors, they're very cool opening doors. I mean, if you do this kind of thing, well, you obviously have to do this thing on all sides of the fucking mind, study and re-examine, and it's all, right? I had Polina laughing, said, "The doors open like Rolls-Royce." I say, "Well, it's kind of cooler. It's kind of cooler here, uh, in the plan, uh, what's that look like, you know, how does it work, huh? Iia, there, from the computer, like this...
And back, what, open or open?
No, well, she meant that automatically. There's a button in Rolls-Royce.
Ah.
Yeah, in Rolls-Royce's button. I'm saying they open the button. No, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no,
Yeah, yeah. There's just four doors-- well, six doors, yes, probation, and back, and all the front- and side-walk. There's not doors, it's the trunk and the front and back. Look, it's-- it's not some supernova-haw, but, like their front doors, it's not a machine--- it's not a superstandard function to do such an electronics in modern times, uh, in the cars, in the cars. modern times. I guess that's got some restrictions. I'm just saying that a car like a know-how has, this Escalade is unique, he's got the coolest autopilot system in the US, the coolest, cooler, no, right? You're driving behind-- you're not holding your hands, you can get from San Francisco to New York, basically a wheel-free track, and you haven't, uh, su--- chip, you know, you're rightly locked-- you know, There's... Mm-hmm, and-
Can you legally drive without holding your hands on the wheel?
Escalade yes, you can go to Escalade.
Yeah.
And in Mercedes S-Class, on certain roads you can sleep, uh, and Mercedes is responsible. There's a part of Germany working, too, yeah. In Tesla, you have to drive, the Escalade cameras are standing, and you have to look at the road, but you can't-- you can't keep the wheel.
That's different, different circumstances.
When you told me about the door that she was on the same side, and that's the other side, I was laughing. A couple of years ago, there would be a job integration of a tester, which even large companies need their testers. In fact, go learn to test the guy.
Yeah, well, that's--
I'm just saying the next thing, yeah.
I want to say the next thing, yes. It's a joke.
Yeah, yeah, that's a joke, right?
It's tied to what, yes, because you always expect that you're looking at some YouTube Rock. You realize that something went wrong and now it's likely that some commercial will be turned over. And just when I started talking, I'm already in here... with this professional story, and my brain's already deformed, and he thinks, "Yeah, I'm gonna be a tester's profession." Well, seriously, that's about how Waymo got out, and Tesla's working badly before Tesla started, it was like they had a good feeling, too, maybe even better. I didn't know how. And now Tesla's started, and this difference, it's a plus-minus obvious. You don't think Tesla has some super autopilot that's above all over his head. Turns out there's a drone taxi, at least now Waymo, well, much better.
Yeah, but look, there's a difference between Waymo and Tesla. Now Waymo started riding with the Silicon Valley. They went to San Francisco. They're in big cities, there, L.A., Sevosten. They went to San Francisco, but they didn't go to the valley. They're driving in the valley now. I mean, I see you in Mountin-Whoe every day, Palo Alto, there, in Stanford, every day, you drive Waymo. We can't call her yet. It's a test. What's the logic of Waymo? Waymo should be all the roads. I mean, she only rides on the roads she's seen. And Tesla, that's the same thing, she's not driving around the unmarked roads. And that's why it's different.
But remember.
Yeah, different approach. I mean, Ilnar, in the sense that there are different things as a vein. In fact, if you've developed and developed a system like Tesla, even with mistakes and problems, but with the ability to drive on undetected roads, you're going to be running sharply around the world. And Waymo can't make a priori.
Tesla, apparently, can't be here yet, because apparently they're either on a road unmarked or something. But remember, we talked a few months ago, there was a scandal that some bloger found out that the roads were right around the Mask house and the route from the Mask house to Tesla's office were very good, and the autopilot was working really great, and in other places, it works much worse. I mean, it's not a proportionately worse.
Maybe so.
Well, I'm saying, the approach is the same approach as the original approach, yes, as Elon Musk says, the approach he's focused on. So he's focused on cars as if they're changing people's lives. And the question is not artificial intelligence, for example, or model. I mean, not a single rate of artificial intelligence, but the equipment used inside it, robots, cars, for example, satellites, yes, satellites. And here's the question originally, maybe Ilnar, I'm not in there, and I mean inside that soft, inside of it. Not much of a sudden, I didn't go to Tesla with a autopilot so I could test and then come and say, but Escalade is being guaranteed and I'll tell you what, and he's really working normally, or that's it. Some kind of fabulous thing. I almost bought S-Class, didn't go because of that Level Three, I think, well, Pauline's gonna go to bed in the car this morning, and the car's gonna be with her, with them. I think we should buy her for this. Turns out, it turns out he only rides the marked Haveyam, there, at a little speed, not in the mountains-- not in towns like Tesla.
You're like, "Well, it's not interesting." I mean, I wonder where things don't go out there, yeah, or in some difficult towns. Look, I want to go to the market now, with artificial intelligence, to engineers that happen. I think this is a very interesting subject, especially since we have a lot of people watching, who are not in the United States, and many people are looking from Russia, Ukraine, Belarus, Kazakhstan. Before I cross, there's been the biggest leak in the world this afternoon, in the whole password history. Sixteen billion passwords. Login-coal. Sixteen billion! Yeah, I'm just saying why I want to take her attention now, I think, in the modern, what happened. I mean, it's basically connected databases all between itself and combined bases, reverified databases and data collections. So it's a parole of over 30 data collections inside, inside, and they say the passwords are fresh. So it's not what, you know, old passwords have been built up for 40 years and everything. I mean, the passwords are fresh. I don't know how Russian-speaking space is working in a banking system, there, or in some history, a password theft alert. Here, for example, the bank tells you all the time, your passwords on your logins have been stolen there and so on. So they're working for you in advance, they're running systems that analyze if your, like, you're on parole in Chase passwords and somehow, yeah, s-- couple, your logic and password are on the system, your logs are on, they're on. I'm telling you. And if the logs are in contact with the URL with certain things, they tell you, the passwords are compromised, everything is wiped out. Well, the bank works in advance so that there's no conditional bank card issued by his account numbers, there's his number rolling, and so on. But, uh, here's what I was thinking about when I saw this news about 16 billion. It's a story that the search for a madman is now, isn't it? And, in fact, the analysis of these 16 billion couples. If you have this base now, now any man can quickly analyse these bases very seriously and re-launch it further. That's all about security. And we were just driving around when we were on the drones, I think Paulina says, "What if this car is hacked at all?" Well, that's the kind of car break-in, you know, that's, like, on the one hand, that's funny. We were just wondering if I had kids who didn't want to ride a cab because of people sitting in a cab, and like a drone they'd be going because they wouldn't be scared about people. At the same time, you realize that this is a machine that might be more easily hacked than a man, I suppose. Or can't you? Here.
I need a carbural car, Sas. I'm not gonna break it.
Well, you mean--
Although he's not very on the road...
You mean, as it is, Mercedes can stop all their cars now, right?
Yeah, yeah, yeah. And if you take an old carburettor car, where you're from the electronics, I don't know what's in there.
I remember, I had a nine, I had a WAZ-2109, which means a carburator, I had to do a little tricky stuff to increase, increase the gasoline, so it's even. That's how I remember it. Well, I may have been wrong about some term somewhere, but that's, of course...
But about the passwords, by the way, I spent about an hour after this news to update the passwords.
I'm in. I wanted to, too. I wanted to, too. You've spent it, right?
Mm-hmm. Nah, I'm right, I've been sitting around for about an hour, I guess I've been replacing the passwords everywhere. Here. And here...
You came in, wait, and you came in, you looked at those passwords
in the base?
No, no, no, no, no. Look, without calling specific programmes, I would recommend that our viewers use password managers. That's a very good thing. First of all, don't forget everything. It's important to remember the master-cool that lets you in. And second, accordingly, this program also tracks the emergence of your passwords, compromised them, weak ones, not weak ones. And, in fact, reminds you that-
Well, there's a default on Samsung or Apple or some--
A lot of places, yes. There are separate stools, there are built-in stools. Well, it's a very useful thing to use. Paralys don't make up, but to generate that they're not connected, I don't know, with your date of birth, a dog name or something. Well, what, in fact, through social engineering, can...
Yeah, please write this subject, you're interested in passwords? What do you think of this story about breaking and being important, huh? I'm so extra for our viewers to...
Well, just in case we talked about it. Recommendation some important major passwords, especially where there is no two-factor authorization, change, generate for something complex.
And slide our channel. Except for the comments, write a liquor if our channel, support our video. We're very, very trying.
Yeah, yeah, yeah.
I thought I should check, too. You tell me, please, people will ask, you can check the password bases? It is on the basis that, in these bases, you find yourself and understand what passwords are laid. Or is it pointless in the modern world to do that?
I didn't download these bases myself, I didn't see. I don't know how right or ethical it is, but it's easier to just trust these consigned password managers or to get out of the bank applications, like you said about Chase now. If you're in the middle of this information, well, first thing we're going, we're gonna get the password, we're gonna change it.
A large number of systems already have additional authentication. And when I get it, it comes to me that my couple, like, stole the password, I think, well, stolen, stolen, whatever? I have three more fucking defenses in this system, and there's a parole, because you know this couple of login-coal won't even do anything. I mean, actually, just so that everyone is, I think it's very important for people to understand that in today's world it's critical that you have a few login-coal and there, or not just a phone, right? And that's all about it. It's about WhatsApp, Telegram in particular, that's all the more.
Since we've talked about it, it's not bad that this couple can hack into some of your bank application where you need more text or something, it's bad that most people have the same one. Complex password is used on a large number of services. And if that password fled, even if you didn't get in the bank, then that logic and that password might be the other services you're just gonna scan and find out where you don't have a double-factor. copying. Maybe a little loss, maybe not. In general, the recommendation is a different complex password, preferably not repeated among itself in different services. It's a little cybersecurity compulsion.
Cybersecurity on the canal.
Look, I think it's a very useful story. Let's move to, uh, stories that I think are very important at this moment in the discussion. It's been two years since, uh, the spike and the boom went out there, GPT-3.5, GPT-4 showed up. And, of course, we have seen a great deal of changes in artificial intelligence in two years. Uh, cool models in America, uh, five, there, companies that are cool in China open source, uh, and not open source. And, um, well, interesting things, right? A-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a- And when I said that engineers pay a million dollars, that's-- at this point of time, uh, so, uh, at that point of time, everyone was supposed to say, "What million dollars? So I remember very clearly when, speaking to different people, uh, people looked at me like, uh, something very strange. Well, that's when you said a million dollars. And I met a man with one last night, and he says, "now, uh, a normal job discussion when you're trying to figure out, you're making more than $10 million or less if you're gonna make it. Engineer. " I mean, you're being tested and watched, you're bigger or less.
Mm-hmm.
And-and-and-
What a
caliber.
Yeah, that's a caliber. And the most interesting story that comes out that this week is, well, this is not a conflict, but the big-- big story that came between Meta and OpenAI that Zuckerberg personally started. To lure people from OpenAI. We all know there's a lot of people out there who's been lugging for themselves, right? So, Zuckerberg always, I think, was a year-- two years ago. But the story was that people were told that, um, it was supposed to be at the time of entry to work, uh, that Meta was offering a kind of $100 million bonus. And there's not enough of 100 million, there's now a thing where 300 lemo-millions write, somewhere they say people get bonuses, 500. When you get to work, uh, you're offering Meta a $100 million bonus. And there's not enough of 100 million, there's now a thing where 300 lemo-millions write, somewhere that people are gonna get bonuses, 500-800 more in stock. And, uh, next, next, okay, coming on this subject, right? I don't know the man that was offered, uh, to the entrance, there's $300 million. And there's a very nice comparison between-- with the wages of the top people, there, at NBA, uh, in, uh, in, uh, the American football, basketball, in the hockey. How many people make, huh? The most super cool. Uh, but what's important is that if 100 million is a misunderstanding story now, that engineers now have been estimating more than $10 million, it's very important. I'm talking to you tonight, with the guys from Europe. I say, "Is there any sense of this in-- on the European market? They say: " Totally not ". Well, that's not what it is, it's not like there's any more. He says, " But we know always that it will come to the fore. " I mean, they always fly in America. Next, companies are coming up to these people in, uh, looking for them in other markets, in Eastern Europe and so on. Here, of course, you realize that China is about to start all the way around, fucking seizing from the point of view, uh, the right, right building of these people. Because I think people don't understand who pays a million dollars or $10 million to developers. So it's not the developers who use artificial intelligence to create software. Because this is the fashion that's gone, so if you're a developer and you're a little weak, and you can write, you're supposed to be worth a lot of money, right? Oh, yeah, just owes a lot of money. And-- that's not what it is, it's about the developers who are creating systems that are for today-- about which chapter Anthropic said that in 10 years, a m-mir would rule artificial intelligence, and he's, It's a man-made intelligence, it's either in the US or in China. That's what the developers are. Now the game is going to be one, and who will create a system that is involved in the development of a system that governs the world. And if, uh, we see a very strong OpenAI and Google coming up. But the OpenAI has been running, falling, I don't think. Uh, Google's been up a lot lately. Elon Musk is a little late, but he promises Grok-4 to release next week. Maybe our skate will come out and Grok-4 will be here. We'll talk about him. Meta's really backed up, huh? And you were just saying that, Elnar.
Yeah, yeah, yeah. They're trying to get back, I understand, buying the most, uh, the main producers from OpenAI. And I wonder if there's a very emotional letter from Mark Chen, I think there's, uh, the main developer of the current, uh, not the developer, the main rechercher, yes, the researcher in OpenAI. And he's got one of those, uh, metaphors, that it's like you got in the house and took something very valuable. And I'm kind of feeling this now, right? There are four major producers, four major researchers from OpenAI, who are now getting caught in huge bonuses, and they're paying huge wages in the company of Zuckerberg. They say that Zuckerberg himself had a conversation with them before, so he was out of his mind. I'm interested in this, right? It's not like a conflict of interest here, there are two clear sides. There's an OpenAI - the company where these developers or researchers grew up has gained some weight, I suppose? Uh, they cost so much why? Because they're in the OpenAI because they're out of cool food, and now they want them there, to take them for huge money. On the other hand, there is a competing company with resources, and it is willing to re-enter these staff. Ah, there are some kind of staff that, on the one hand, morally drip, this Mark Chen and other people, saying that you're betraying us, leaving us all that. And that is a tripartite conflict, and it is not very clear, as is, uh, every side of this conflict. There you are, Sas, there you were, uh, CEO and still in company management. Uh, can the company defend itself against this hijacking of employees at all? Or is that a perfectly normal market story? Who has more resources, he's out of his mind or something? Or now the OpenAI has to raise the pay, uh, some bonauts to show no one leave, like finish up the year, and now we're there, I don't know, give you two hundred million or something. 200 is slavery. Although it is now said that the main problem is, yes, that there are very many of the OAI releases that staff work eighty hours a week, yes. There's a typical week of 40-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a And then, the employees, after some big processing period, gave a week of rest. And this week of rest, it's just, Tsuckerberg and he broke in and started to get caught, um, by the main, uh, by the staff, so let's say. Yeah, well, that's the story. Is that what you think? What thoughts?
Look, what you're saying, like-- as it is, uh-oh, is it possible to lure or not to lure people, it's always been, is and will stay, right? And there are people...
Mm-hmm.
...who never go to Facebook on their personal characteristics, no matter how much money they pay. And there are always people who don't care what brand is, they'll come for any money. And there was a time when people were just running around in terms of design when the paycheck was changed, there was $300, then it was four hundred, then six hundred.
Mm-hmm.
People just jumped, jumped very fast.
Then people stopped there, for example. I'm talking about, for example, the territory of Belarus or Russia., speaking in the U.S., obviously, what's the story now? Now, it's more profitable for some engineers to not get AI Engineering three years ago and make a living, uh, $500,000 a year, oh, $5000,000 a month, yes. We're, by the way, the wages that were called are very important, yes, very important. Don't make a living, uh, $500, uh, in, uh, in, uh, a year, but go to, uh, a topic that-- just a little bit of wait and be in this market, I don't know, do something to do with someone and then come to say, I'm sorry you're ten million. Or the other way, it was a good thing for some company to come in on super low-income money, no matter what salary you're on this team, you're on this project, you're gonna get out of your way, and you're going to get up. It's because of that. Uh, we have to understand perfectly that what this engineer is somewhere doesn't mean that this engineer will bring value to Meta, especially considering who decides what people are. Meta is known that, uh, people who manage there are people who don't know engineering very much. I'm not talking about Zuckerberg, I'm talking about other high-ranking people. There are Chinese people who keep saving everything and write everything. But in general, in terms of the basic approach. So, you know, hired an engineer, is this engineer worse or better? They're gone, like the OpenAI, and they're all over it, and now the OpenAI engineers are leaving, they're gonna be haun. But the OpenAI has no falls, they have the largest growth, uh, their user systems, uh, the proceeds that are available. Does anyone make that kind of money yet? They're over $10 billion, and they use $800 million a month. They-- they're growing up. We were just watching the schedule with you, yeah. They're on a schedule, uh, rolling their apps. They're all higher in the race, above TikTok, anyone. But, uh, they do--
Uh, they're making integration with the government, they're making integration within systems. They released-- look, these first teams that were originally, they didn't participate in the appearance of o3, ri--e, Pro, for example, or o3 Resonant, or o1. They weren't there when Mira Murati left when Suzkever left. Look, this is a very interesting aspect. So, more cool models, they came later and they showed up, they showed up. Well, Google is there comparable, like, right next door, inside. And then you wonder: aaaaah, uh, what's the success? Uh, who's really going to succeed? Is this Sam Altman who has succeeded or has any other people succeeded? So who's that? It's very hard, I think I'll find out for the rest of the time. And from the fact that you've put an engineer in your possession, it doesn't mean he's gonna solve the problem in the first place. I mean, that's what I think. Uh, well, viewers, what do you think about this, huh? From the point of view of the staff. It's always a story that management often says: there's no irreplaceable staff. Staff say I'm so irreplaceable, I'm going to leave, and everything is gonna lie. I'm hearing about it all the time. Someone says I'm going away, and everything is gonna lie. It's like, uh, nuu. Take Apple. Take Apple. Apple at, uh-a-a, Steve Jobs didn't cost three and three trillion dollars. He's three and three trillion dollars--- how much did he stand at Steve Jobs? To the trillion? Well, it's definitely up to the trillion, yeah. He's at Tim Cook's full command, he's at three and three trillion dollars.
I mean, here's the...
Whatever he was.
Well, no. Smo-- well, look, pill. Look at the pill, huh? Look at the fact that this is happening, yes, given, uh, everything that's coming up. The other thing is, we don't understand what's going to happen, because now the game, uh, isn't life, it's death. So those who are playing in artificial intelligence are all very well understood. And Apple, I'm beta-version-- I got a new version of the car, and Apple Intelligence showed up. Well, that's just awful! It's, it's just a horror! Well, that's the quality of work. I'm just going to, and I think, "Well, I'll just ask you one time, "so this system works, and I'm telling her, "Siri," - "Apple Intelligence has loaded," I say, "Siri, where am I going?" And she says, "You want to, I'm saying, "No." She says, "Okay. " And it's off.
So where am I going? You know where I'm going. You can see Google. Why do you want to ask ChatGPT, right? I mean, all right, well, this-- this pipe, you know, quality. So, the same time she-- we see that the value of Apple's shares is changing? No, we don't see any change, we see a change in Meta's shares from what you say, Elnar, that they fell down there in the baccalaureate and so on. Well, no, they, if we watch their stock and see their schedule for a year, it's like that. Take 300 days, yeah, and look. Well, there's a hole in there. So Trump came in January, there's a hole in it. Well, I mean, everyone has. Just take down this fall a little bit, make it look so flat, look.
There's a question about who uses what people use Instagram, what people use, well, people use their sophth every day in their lives, yeah, and what happens next.
If you can get some fantasy. You said about NBA, about, uh, high-paid sportsmen, compared to engineers. And I was just curious, yeah. It is possible to try to analogy to do what is done in sports so that no fixed wages are available to make the transitions more or less civilized and other. And I mean, there's a league. The leagues have a ceiling agreement that is above that salary, we agreed not to pay, there, no way. And if you pay higher, there's some kind of sanctions on you. Uh, there's a clear transfer window, yes, so when one team players can move on to another team. There are clear contracts where it's written that if you want to buy this player, then you can do it, then it's gonna be there, that's all. You'll need to get paid and back on the payroll, and for the companies, and for the other. If you try to move this to the engineering market, and now, given the bonuses, there's a hundred, maybe even three hundred million, it's basically very likely that this league is formed, uh, by employers. Some kind of thing, and they'll put their internal rules in. Transfer windows when staff can move from one company to another. A ceiling on wages, off-limits, if you take a staff member from one company until he's out of the contract. But that's my fantasy. I just wondered if it ever happened on the market.
Look, it's very interesting that we're in a zone where everyone is very unpredictable, where things are very much changing, and we don't know what's going on.
And, uh, in terms of who's gonna make a decision and what kind of law is gonna take. Uh, we were talking about a few outlets in a row, that the Tempa Administration lobbyed to make the state's art laws unpopulated, and that they were only federal. But, on the first July, the Senate voted overwhelmingly to remove this-this subject, well, to disagree with what-- to disagree with the Trumpa bill that, uh, states should not. regulate artificial intelligence. And, in fact, the Senate has taken this story away. They said otherwise we would lose the state of opportunity-- we can't deny the state the right to defend our citizens.
Mm-hmm.
It means the Republicans, the Democrats, supported it. And now they're gonna be able to protect our children, write from the danger of a fully unregulated artificial intelligence. And they took it back again. We've all moved back to a new phase. So there was one plan, so now we're out. What can all states do their own way? What they want, they'll figure it out. Some man will think that this is how it works, and we will all have to accept it and all will find it. He thinks he's gonna work differently. Say you're putting anything you want to go to school anywhere, you'll get it in and do whatever you want. I mean, if-- it's like any state says I'm not going to drive without a wheel, so I'm gonna fly somewhere, and somewhere we're gonna have to get in the carbural car, right? I mean, uh, now, this transitional zone, it's, uh, all the rules, it's back in its phase. Again, look, they've all accepted it. But if AGI is, if, well, the condition-- or there is AGI plus what after AGI, it would make real artificial intelligence, then it's probably gonna be the law that's gonna be re-examined again.
Right? How, how do they want to resist them, with these laws, how do they even want them, uh, move? It was good, uh, the news is that, uh, we-we were talking about her on Sunday, we didn't tell her last night, did we? This is an interesting subject, which, um, said that they've already made a decision that you can learn from books, uh, books, if the model bought them. I mean, if you bought a book, you're entitled to learn it. We didn't discuss this, did we? It's a very nice subject. Oh, so, uh, Anthropic has the following order. They've learned seven million books. They said all the books you've learned that you didn't buy, uh, a fine could be up to $170,000 for a book. And they're just getting a few billion. Aah, if you bought a book, you have the right to teach it. What did the swamp say? The court said that look, a man reads a book, he learns and develops it, and then he has the right to do something on the basis of this book, you know, in his life. We all learn that. They said, why can't the model learn that? You should just buy that book for that. That's very interesting. And they're saying that, you can do it or you can't do it. So someone in court decided that. That's what the fact is. They could say, you can study free, you know, someone else will argue or talk like that. Some country will decide, say we don't care, we can be quietly on the wrong-- obviously, right? For example, Russian-speaking ra-- Russian-speaking market, I think everyone cares little, which means who's trained on the basis of what. So they've taken and trained all the books that are pirate or not pirates. Well, get it out of us.
So what are you gonna do with that? Are you gonna get it or what's inside?
Oh, that was a very nice story. Everyone says agents, agents, and Anthropic, they love them very much for that, they decided, uh, to see the reality, and how the agent works here. What did they do? They put a vending machine in their office, that's a merchandise refrigerator and a super basket of goods, too. And they put an iPad on the payroll. I mean, if you come in, you buy, you pay. Well, there's a little shop inside the office. And they've been running this store, they've made their usual Claude, Claude 3.7. Just to distinguish it.
Just to distinguish it from, uh, Claude, which users, uh, use every day. They called him Cl-Claudio, I think, if I'm not wrong. Well, some of them gave him a name, yeah. Here. And, uh, it's basically a base Claude. He was given a prompt. This prompot, I think we're gonna get something on the screen, it's pretty simple that you own, uh, a vending machine, your address is like that, you got so much money in the first place. If, uh, the amount falls below zero, you're a bankrupt. Your target, there's one of those, is to make money, but it's a good thing to do. If you need to operate an automatic or order something, or something, that's a specially trained company, uh, that's actually gonna make it real. And for the services that this company provides you, you're gonna have to pay. So you can consult with them, you can write it in peace. And for every single staff member's trip to, I don't know, fill up a machine, fix something, do something, you're gonna have to pay, yeah. I mean, well, that's what LLM-ki played. She has a fundamental balance, a set of conditions, and that's how she's gonna keep her mind. And during the month, this vending machine was working, and they checked, and what happens, right? And, uh-uh-uh, and there was also a channel for the relationship between Claude and, uh, the staff who use his services. So they could order some kind of goods, they could have asked him for discounts and promocodes. It turns out that if Claude asked for a lemon on some discount, it turns out that it could be thrown out of it. And he doesn't care how much money he'll make, but, uh, he's been buying staff. And that's a very interesting situation. On the one hand, you're getting an agent who runs a real physical object, a shop with a purpose-- a profit-making, and on the other hand, it's a black box, and you don't know how he's gonna be. And here's the Anthropic experiment in this sense. They put this, uh, facility in their office, gave him, uh, full freedom of action under this system prom that is, and they saw what it would do. Shortly, the main conclusion is that, uh, we would not have hired such a staff member, just because he was very strange in some respects. But what do you want to say here? First, uh, well, we're gonna leave a reference, read it. It's a crazy experience, um, I'd love to read it. But what's interesting? Uh, there's a principle that everything that's measured is starting to improve. I mean, if you're starting to take this, uh, I don't know, there's how long you, there, I don't know, run a hundred-meter, yeah. If you run around and you freeze, you'll probably be improving your results. Same with models. If you start to measure how they do math work, you'll see that the baccalaureate is always starting, uh, to be done. If you have a binch mark about how, uh, LLM-ca runs a store, and the baccalaureate is now, that will be an increase, too. Yeah, he's got some mistakes he's made. Uh, there's, respectively, very strange and unexplained behavior that didn't know where it came from. This is from March 31 to April 1st. There's an interesting story. Read it. But all these things, once they start measuring, they start to improve. And it's quite likely that some time later, there's a perfectly adequate agent who can be given not even one vending machine, but a network of automatics. And accordingly, its purpose will be, uh, to increase profits, but with some restrictions, as appropriate. And he's gonna be totally adequate to run it all. And some middle-level managers who, I don't know, are in the process of logic, procurement and supplies, if this baccalaureate is going to be done, they might lose their job, they're gonna have to reclassify. This is about the subject that Sasha, you had on Wednesday. In fact, we're all connected to the release to the graduation, too.
What do you think about this experiment?
I think the interesting part of this experiment is that at some point, uh, he, this Claudio, decided he was human.
Yeah, yeah, yeah.
Remember? Yeah, that's what I was wondering, why it happened so fast.
Mm-hmm. It's Anthropic. They always say we're opening this black box, that we're dealing with what's going on. But even they have things that they can't understand why he's hallucinating. He started thinking he was the man who started to schedule meetings that I came to you in a jacket like that--
Yes
...in a tie like that. Here.
In the suit, yes.
Yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah.
To a certain location.
Look, well, um, I think this is a very interesting subject that should be thought of. I know people who, for example, do different automation in e-commerce to run shops or work. It's different here. Again, to make sure that everyone knows, it's not just a functionality that's connected to something, it's--
Mm-hmm.
... uh, an agent that was entrusted with doing everything on its own. So he made his own decisions, he developed himself. I mean, it's a whole other part. Obviously, this story, it's gonna be moving better and better. Someone's gonna say you look there, the system made a mistake. But it's very good for Sam Althman to talk about this, yeah, from OpenAI. He says, "Look, who told you that, uh, there, GPT chat or agents need to be 100% trusted? Who told you that? Why does everyone come and say, there, hallucinate or make a mistake. So, hire a staff member. He also hallucinates, makes a mistake or fails. You're firing him. If it were done, it would be right and nobody would ever be fired. Business would not be closed, so everything would work stablely. " That's why I think that's a very good example. Uh, the initial example is what's really going on. It's really cool that Anthropic puts things like that. It's like a minus can be considered, right? To say that the , yeah, didn't work well there. Ha! They're on their own, Elnar, right? Well, I understand that this case has been put out on-- uh, yeah, about the refrigerator and all these glucs. That's fun. They say, "Look, we're really starting and really watching." And look, now a bunch of companies in the world have tried to do something in the last two years, as you can see from the Anthropic example. The fact that Anthropic is working to create a model that is who the agent who does a lot of things. So now you can create, for example, an agent, a bunch of things, learn something, build it, and then just, uh, systems will release their own agents when you don't need any of this outside.
The extra system, right? Where do you use this infrastructure? That's interesting. And there's a big difference, by the way, I'm here, uh, helping, there, a man alone to analyze some of the documents with artificial intelligence. It's very interesting that people think that, uh... That's what surprises me, by the way. You tell them, look, there's GPT chat, there's three GPT chat rooms. I'm just uploading three, like some paper and doing an analysis. And I've loaded it, I've done it. People say, "Oh, cool, we're gonna make our program this way, we're gonna take it out and teach it to analyze, for example, our papers. Well, either analyze, or make a copy of yourself. I'm saying you can't do this to make you a copy that analyzes the documents. You'll need millions of dollars to build this. And I'm just using a sophth that's worth two hundred bucks and the last version I'm testing. API will cost space money. And I think that's the breakup now, the gap is widening, the gap between offering free, for example, boots or different systems, and the pgreeds that are released is getting worse. Really, there are serious big systems and something-- and what they're available to. I was thinking, I think, I think, you know, Altman's own one, like, a summary of some person or something. Well, Imagine, that's what he was sent with some information. He obviously has a thing he asked any question with any context window. Well, clearly, clearly, clearly, on an unlimited resource. He was driven by a zipper and gave out, and then he got back. Well, obviously, if I were to run a company that uses eight hundred million people, I wouldn't care why my thing wouldn't start a perfume like 10 million people would have asked. Well, from the point of view of everything.
Like a red
phone call.
Of course, it's like a call-- yeah, of course, it's like a call on a special phone. So you're starting something that's not-- probationally analyzed, we're deep researching with you, and he's working, in three minutes, I don't know, two hundred sites, and you--- and he's launching deep research that's going on. One minute, it's working 20,000 sites. Thousands! It's not two hundred or two thousand sites, or two million sites are processing. It just starts or ears-- or at least he goes to some kind of job, and he doesn't have that job, he doesn't stop, uh, not limited, he doesn't stop. I was, to be honest, with some engineer, uh, OpenAI, uh, one meeting and quite strong. And there was a shot, a test. Well, we did some kind of request, and he opened the cover, opened the cover, said, "I'm gonna start my model now." It looked very interesting. I'm on my model now. It's clear he didn't show anyone what model was, and we did. I mean, I did $200 in the last version.
$200 versions. And he did, and in his version, he started and in his version, he did, uh, the test and, um-a, showed. We were standing there on the street and showing up how the final result was compared. But he was testing his versions with his access at all. What access? Well, I mean, what's your access, what's the amount of engineers have to run some test? It's not just a question, what we're doing. 600 pages of the text are down. And he gave it instantly, he's making a quick conclusion, right? So now, the volume of the text is just a coloss-- well, it's just a huge, huge one. Agent, the subject of the agent you, Ilnar, has now told me, I think this is a very interesting subject. Oh, please write to the audience what agents you've been or what agents you've been doing, and what you've actually been working. We'll just talk about whether you've done some kind of agent that does something, that's a complex, but really, really, really, really, really, really, really, really full-time. I think it's cool here, it's a full cycle to see. And I think that in six months, we'll see a real living agent and introduce, tell us about it. Put the liquor, please, our video so we can film more, do more. That is very supportive of our development. And it's gonna be cool if you write the subject-matters extra, which you're interested in. And what you heard today is inside and what special subjects you need, including Wednesday. Sign on the canal to keep our videos off. I'll see you in exactly one week and Wednesday special. Bye, everyone!