While artificial intelligence is just helping a man write a code. It's a big, big market, big development. But when artificial intelligence starts to help research, design, test, improve the next model, there's a loop of acceleration. And that's the loop, if you can call it that, can sharply reduce the time to the general artificial intelligence. And today, we will see in the heart of the Ring Valley and what happens to AGI. What is an incredible phase of artificial intelligence that many people are waiting for, many people are very seriously afraid, and many people think it's never gonna happen in life.
A very interesting discussion was held recently and was hosted by the editor-in-chief, The Economist Zennie Minton Bets. And, uh, I think she's built a very interesting discussion with two incredible people who were present at this meeting. And at the East-- this meeting was the head of Anthropic Dario Amadei and , and the head and co-founder Google DeepMind. It's a unit that handles artificial intelligence in Google. Demis Hassanis. Very famous people. And it's important to know, by the way, that it's not just a little fun out there, Google. And there's a chapter of Google, who's above all the stuff. Yet, Demis Hassanis is an unbelievable person in artificial intelligence. And this editor, she didn't just ask any general questions when some discussion took place, as usual, on the market. If we open up, uh, YouTube, and we start watching and listening. The interview is different. She's always pressed into real contradictions that are in the market. She always talked about the timelines that people call incorrect, about the money and the cost of everything. She always talked about jobs, she talked about geopolitical. She said a big--block about the chips export, and the risks that models might fool you and deceive us. A, the trust of the people, society and the role of the State in this regard. I think this is an incredibly important conversation, and this conversation took place in Davos, and I want to-- and there was a continuation of this earlier conversation, uh, in Paris, yeah. Specifically, it was, this session, yes, that was the continuation of their previous conversation. And, uh, in this session, which was in Paris as a continuation of this conversation, A-Amoday, Dario Amoday said that AGI was a common artificial intelligence. Uh, what is considered a common artificial intelligence.
We need to understand that this is a certain notion, very abstract. No one understands what this is, but it is in fact considered that it is a system that can carry out a wide range of intellectual tasks at the level of human beings or above, yes. And not just answering, there, in the chat room, and not just creating and programming. Although I'm thinking about the time when the first version came up, there was a GPT chat and Anthropic started, and there was, yes, and Gemini showed up. And yet, AGI was talking about when they said they were missing a layer that is now. The layer of creation, uh, systems with artificial intelligence, yes. The word "agent" is still in use, and I don't like it very much. You know. But it's still used. Before AGI, it was just... unbelievable everywhere. Let's just look at this, here, two of these people who are really, you know, serious leaders of opinion. Again, my reservation is that I see how these people are often wrong.
I understand that this is a very relative story to refer to these people forever. So I've been recommending to all of us that it's unbelievably hard to understand, to actually use it, to try, not to try any systems on the hip, but to study, study, study and test. To introduce it every day, yes, into your real lives. Not to devote ten to fifteen minutes a day, but to introduce every day. So, AGI. So, Dario Amodey has a line that's pretty tight. He believes that the AGI window has been very realistic for several years. And because he explains that, to date, coding is all that has to do with the design of a sophth and the study of artificial intelligence, it's research on the establishment of Sami systems. artificial intelligence, they're accelerating now. And Demis Hassanis, the other one, on the other hand, has a very careful line. He says that coding and mathematics are actually accelerating, but full scientific invention, there, physical world, experiments or is believed to be the creation of some smart game, for example, some kind of game. super cool, like a go, like, once was a jerk. That robots and constant training are different until it's, uh, until this subject is closed. I mean, if you're really talking, yeah, about physically artificial intelligence from the robots, well, it's not really closed. And the difference between them is not that they believe in AGI or they don't believe. I think it's like demagogic to go this way, there, believe in AGI or believe, there's still a term like ASI, another level of system, yes. It's important here that both people think technology is huge. Difference in criteria, yes. Amoday looks at the practical ability to do so, accelerate modelling, and Hassabis looks at a higher level. It's the ability to ask the right questions, for example, scientific questions or hypotheses, or to check them in the real world. And there's a lot of things, like all references to AGI goes to that, and at what point in time, artificial intelligence is... can do some super-crunch research. And I think again, it's still some kind of populist, because this endless search is probably going to be AGI. Write it up, by the way, what do you think, huh? We can just distract and argue indefinitely. Now or not AGI, yes. We are now in time for some incredible systems to be built with artificial intelligence. And how would it feel like this artificial intelligence, it's like, well, it's not AGI, it's something, something a little weird, right?
So, in this conversation, in the general, that I'm gonna be there for you, there, to bring some things. So, uh, the motorist is the headmaster, she's the editor, yes, she's not the chief, she's the editor, the economist. She, uh, she had an initial title of the topic that the name of the "AGI Day" conversation. So, and, uh, she's returning Amadea just like I said, to his past prediction that a model that can do almost everything that a man can do in a nobel laureate in different fields can show up. by the twenty-sixth, by the twenty-seventh year. And that means we're in the twenty-sixth year, and still, you're not on that project. I have a question for you: do you think the model can do as a nobel laureate or not? Because on the one hand, the model, I think, can do a lot more than a Nobel laureate. And on the other hand, there's a model that can't do things and can't do things. And here's the question, and we need to exactly take AGI as a system of accurate cloning, copying. It's like watching robots, I-- indefinitely show, like, I even made a prepayment in this robot, but I still haven't been given the opportunity to buy it. There was a commercial in this robot, totally stupid, I think. There was a robot that was dusting the usual carpet. I always said, why would I have a robot that sucks the carpet? I don't need a robot that's gonna blow the carpet. I have a robot vacuumago who's optimised under this, does it with a crazy speed, builds incredible cool cards. I mean, it's built for that. I don't need a human species robot to vacuum the carpet. And how I don't need a human species robot to wash the toilet. Or I don't need a human species robot to protect space. Or I don't need a human robot that I have some clues or something. I mean, I don't need a human species robot to walk a dog. I need some device that might, for example, in theory do it, which might give me a bag of land or, there, a plant to land. I'm not talking about industrial robots anymore, am I? But if we're considering a robot common, which is doing some kind of work somewhere, then first of all, who said he should be alone? Not them, I don't know, there's five, ten houses, that the absolute rule can be. So, is he the only one? And here's the story, I think with AGI, it's very fundamental. So we really want a copy of a man's life and constant parameter and never do it, never get a man. That's what I'm talking about on my other channels. From the perspective of human development and human spiritual development. But even if hypothetically, you think of something like that, then you know, maybe we should or not? Because today ' s system, in terms of, for example, the volume of knowledge, is obviously modern artificial intelligence, such as ChatGPT or Anthropic, they've swallowed a man in knowledge far ahead, yes, any nobel laureate in knowledge, speed, calculation, approach and everything else.
So it's very important. And they're talking about this conversation not like "what's better" but with this most aggressive hypothesis. We are already at a point where anthropic or OpenAI believes that-- or Google, yes, as DeepMind thinks that AGI is practically a horizon of his horizon for several years or some distant horizon. philosophy. But we're gonna stop at AGI, right? I am.
Yeah, I'm talking about not sharing it, but we're gonna stop on AGI because AGI is some high super-court intelligence. And of course, it's not an artificial intelligence that is shown in the film, I don't know, Oblivion or even, perhaps, not an artificial intelligence under AGI understands that it's shown in the First Player to prepare. They probably lead to it somewhere.
So can such an artificial intelligence come up? So Amadei doesn't say the exact date is guaranteed. Yeah, that's very important. He says his prediction is probably not so far away, and he's writing a simple but very, very powerful chain, so much consistency that models are getting stronger in programming, then they're in the middle of a chain. It's getting-- that's, I think, the most important chain, isn't it? Models become more scheduling, then they become stronger in artificial intelligence research. This is accelerating the next generation of models. Acceleration of the next, accelerated next generation is further accelerating the next generation. And that's the development loop I was talking about. By the way, I literally, I think, a few weeks ago, I took off the skate and told the skate that Anthropic was actually telling them that they had a big piece of code today writing still artificial intelligence, The artificial intelligence that makes the decision itself to write when writing. And he's developing himself. And I'm honest with you, I think I would have unrestricted tokens or unrestricted access to the model, in particular, not just one session, but I could manage many sessions between me and you at the same time. It's easy, and it's probably a lot of systems I'm doing today, and I just recently put on YouTube-Canale in the tusmour, and I've been very detailed on the case. Super cool, I think the case is in the business. Well, look who didn't see. Well, write what you're saying about this... on this case, you think you're gonna take off the cabs, right? Of course, I would have put in a system that would develop itself, develop and develop. Now I have a very serious case right now that I want to create a big one, only completely without the software, no one. It's a big case, right? And the ads, the websites, the Lenings, and the leathers, their post-processing. And the name-- well, well, a lot of things. And to put this in some of its existing infrastructure. And I understand that I will create this exclusively with the help of the code and the claude. There's an important aspect here, that is, with artificial intelligence, I'm gonna adjust it a little bit, right? I would be fine if he could adjust himself or come to me periodically for some kind of small aspects, but rather advice rather than, uh, some small micros. Adjustments than those for real technical assignments, huh? And so that's what Dario Amadei is saying, explaining this, this story that is about the development loop, he says, cites the example of Anthropic, that some engineers tell them they're not in company anymore. They write the code in the old sense that the model writes the code, the man edites, checks, where something goes, goes around the code.
And, of course, I think that so far, a large number of engineers have left even-- behind, even those who use artificial intelligence systems to create a code. Because it's probably just about 100 percent of the world using it for the programmers today. Anything, huh? Something. You should be a very strange man to not do that. But the majority of people still use it as if it were at some tenth stage, right? I mean, it's not when the artificial intelligence system is self-sustaining, and you support it with certain conditions, such as a robot pulmonary. I'm not describing robot pipsos, I'm not telling you how to clean my room. That's very important, isn't it? So he built all the maps of all the spaces himself, and in fact, he's already making a decision when and what to vacuum. I mean, not today, not a very strong system, which determines when to go, where to see where to do. But, in fact, a modern robot pilot, he's doing everything. He's been cleaning his own car, he's cleaning it, he's changing it, supplementing the water, he's drying this thing himself, right? He's cleaning the collection completely with the trash and the cheeks, he's already cleaning it himself. That's incredible! I mean, it looks like a simple thing again, but we don't tell him what to do. He does it himself, he'll recognize it. Modern vacuum cleaner, he's actually gonna recognize that he didn't do dirt, yeah, or clean the carpet. And the idea is, he can tell you somewhere, "Look, I'm here with a tool, washing a carpet, because there's a vacuum cleaner who was washed with the funds, yes, I couldn't wash it. You're here to help me out." But you're not really in charge of them anymore, are you? And I think it's, uh, an important design. Amadei says that models can be six to twelve months away from doing most, which is likely to be almost all the engineering work on software development. And as they say, end to end, yes, from beginning to end, that is, from complete task to its outcome. But he's right here adding that an important restriction that not all parts of the cycle are accelerating artificial intelligence. And there are many problems with chips, chip production, physical location of data centers. All that has to do with the time of modeling is not accelerating instantly. And therefore even if coding is accelerated in itself, the infrastructure remains a brake of real peace, right? There is, of course, the question of how this is related to the AGI brake in terms of infrastructure, because we're still talking about the introduction somewhere. Well, it's clear that there are cities and arrogance-- most of the world, it's not that covered by a good stub-net, is it?
What kind of artificial intelligence, like, can we talk about? But still. So, Hassanby agrees that progress in the coding is strong, as I said. But he says science is more difficult if the system offered, for example, a chemical connection there or a hypothesis in physics, it's not just a "check" button. We need experiments, need labs, need time, sometimes years. He therefore considered that the way to a full-fledged common sense of this artificial intelligence and the way to AGI was not to make the model well-written. I think it's a very important design, which can also be hypothetically challenged about research, because we see that a large number of studies that were, like, proven and done in science. The world, then turns out to be inconclusive and wrong. When new research came up, yes, or some new evidence. Is there a question here, but is there any kind of proof that makes artificial intelligence? Should we even do that? And I remember, sometime two months ago, when I started so much more serious systems to do, not some micro-agents, yes, some less artificial intelligence soft that you can call. artificial intelligence made AI native, yes, software. What Elon Musk says is a lot of things, creating MicroHard. I remember when you were putting a challenge and a system to do something like serious reporting or to achieve serious integration, she had to give a little more in-- well, demands, for example, I'm writing. The systems are now very serious to study, to look for and to conduct research. And we see a lot of cabins. Even there are special platforms now, where artificial intelligence, model or some soft is written, actually some kind of code, it's not supposed to be artificial intelligence, some kind of code. He, for example, asks and hires people to work. Well, that's basically perfectly elementary. Could the usual procedure be to send a request to a site for the purchase of people's time to get a person to come and fix, for example, a toilet? Well, maybe. Did she hire him or not? Well, is it something that artificial intelligence should be done for that? I don't understand. I can write a code that is not connected to any artificial intelligence, put, I don't know, a camera that will monitor the dirty room I'm not having a dirty room. And if the room is dirty, the system will come in and pick, I don't know, ten people to work, so someone can come and clean this place. Well, it's a very simple case. What does that have to do with artificial intelligence? So we have to understand that software has been hired before. Yeah? Now, it's kind of exaggerating that there was an agent who hired a courier or did some research, but it's a little bit of a life, right? So, by the way, the most important point that Hasaby puts to you, to remind you, said that the highest scientific ability is not only to solve the challenge, but to come up with the right question, theory or hypothesis. And it doesn't look like a closed power that's now in existence. And the question, how does he decide, does it look like a closed ability or not? Because I'm getting a big question about people between me. I mean, we're comparing people. Look what's happening in the world of people. People go to church at the same time, cross and kiss icons, and say they're completely material and believe only in physics and science. How can that happen? So we're still saying that this is the Nobel laureate of this AGI level, that we are what AGI says, which is still limited or not limited in our perception. And this AGI, if he starts telling the public that there is something that Archangel Michael is real, not a imaginary--
A real, not a fake character. How do people treat it? And what evidence would people demand?
So there's a key difference in all these considerations, their thoughts, my mind. And yet on this channel, I add my opinion that Amadei, he looks at practical productivity, that if the system does work, for example, a strong intellectual worker of some kind, and speeds up. The development of artificial intelligence is almost like this AGI level. Yeah? Hasabis, he looks wider, and he, well, what at least can be seen on this discussion. Again, he doesn't argue about AGI, that he says we're going there, but he thinks there's no way that there's anything. Yeah? What do you think you're here again?
I think the question is very big. Yeah? So the editor is this economist who has ever questioned whether the artificial intelligence model can close this loop. I mean, not just helping a person, but accelerating your own development, yes, through the next generation of artificial intelligence.
We have to talk about the day after I. But let's talk about the cycle set. What are the chances of you having models that can shut down the cycle and, you know, provide energy? Because it's a really key moment for the "winner gets everything." Do you still think we're gonna come to this, or is this gonna be a more normal situation? Normal technology in which followers and catchers can compete.
Well, you know, I definitely don't think it's gonna be a normal technology. As Dario has said, there are aspects in which she already helps us coding and some aspects of research. However, the complete cycle-down is, in my view, an unknown value. I think it's possible. This may require strong IIs in some areas. Again, where it's more difficult, it's not that easy to check your answer. There are areas that are classified as NP- difficult tasks in a strong sense. So once you start getting more data, I'm, by the way, putting in a strong AI physical and robotic tech. Work, everything is in that spirit. And you also have hardware. This may limit the speed of self-improvement systems. But I think that in the programming, mathematics and similar fields, that's gonna work.
And then there is a more theoretical question: what is the limit of engineering and mathematics capabilities in the field of science?
Who, by the way, had to do such a thing. I had a big story in one of the systems, for example. I wanted different people, both business and editor, and marketing, to report system errors, for example, through Telegram or through WhatsApp. And reporting the wrong-- mistakes of the system to make these mistakes appear in a certain baccalo, and Claude's code itself was reset. It doesn't matter if there's aPI or even a timer inside, they've solved these questions. We're back, uh, we've been turning the code and fixing these mistakes. I mean, basically, to remove a man in the resolution of these different mistakes and to put a large number of different controllers inside of people to some orchestrated agent controlled to a written, straight-up hard-core. I was. They're not agents of some written program that test themselves and they're completely checking themselves. Like, if I have different systems to date, I'm sending a small number of triggers if something's broken or done. But can the system be such that the cycle is as endlessly decided? I went and decided, went and decided. Somewhere she involved a man, somewhere she involved someone else. And by the way, this is the most important story, don't exclude a man. Why would you rule out a man for the rest of the day? A man needs to be treated just like we've seen and considered before. There are aspects where human beings are needed, there are aspects where human beings are not needed. When they came up with a drone, there, I don't know, ten or fifteen years ago, and started to develop it, they knew there was no man inside. Yeah? Well, no, I don't. Easy, very nice to this, totally calm to see it. Just like a lot of cabs ride with the wheel. It just happened that there's probably been a Zoox taxi in San Francisco for six months. And it's a cab, it's from Amazon. Yeah? They-- it's not a wheel, yeah, totally free of the wheel, it's not, obviously, without the pedals. It's not Jaguars who drive from Waymo. But it's clear that since Waymo's project is Google, driving Jaguar in one of the most difficult cities in the world in San Francisco by road, it's obvious that Google will have no problem putting Jaguar in there. There's no steering, there's no paedals. Just like every day forever, because I live there, five minutes from Tesla's office, there's a Ts-- Tesla Cybercab without rear-view mirrors. Yeah? But there's still a man sitting there and there's still a wheel. So it's all micro, micro parameters are some. So it's important that when we talk about shutting down the loop, it's not a little thing. Yet, yes, when we're talking about the real shutdown of the loop, where the system itself is fully interacting, fully operating, fully operating, fully fundamentally operating, completely developing, that is, functioning. Absolutely independent. I mean, Imagine, there's a company, some bank or some kind of structure, or a system that operates completely without a person. But the structure is more serious. I mean, not-- not a vacuum cleaner who's been out there somewhere, but some bigger system with different, different, different, different processes, yes, different process characteristics, different: and offline, and online, different. It's not a little thing, of course, because it shares, in fact, two worlds. In the first world of artificial intelligence, some technology where competitors are moving, they catch each other. So the market is spreading between countries, between these companies and different. Users choose what product to use. And in the second world, where the closed loop is, this is the threshold when some winner takes most of the market. And I am deeply convinced that I have said many times that if someone reaches AGI, it's basically just gone. And by the way, Amadei is very serious about it, yes. About Anthropic. And here I want to raise one thing, I think, that there is a doubt on the market, whether some independent model companies can survive while all the calculation costs are very high. Giants like Google, Microsoft have very large infrastructure. Well, I see, yes, Apple, Meta, Aliba have, there, big clouds, big distribution. Could-- can there be anyone else who can develop at all, some companies can develop, even large operators like Booking.com or, there, I don't know if there's any small search engines that can develop. Operators, any locals in a country? Or you have these giants, they start eating everything, taking everything completely. And the model answers, as if there's a concept of economic curve, it answers through, through it. He says that if there is an exponential connection not only between computing resources, what is called compute in artificial intelligence, but also between the cognitive ability of the model, The ability to model and manual. He says that, well, he says that this connection is present, yes, and that the Anthropic extraction from zero to $100 million, there, rose to $1 billion in 2003 in the twenty-fourth year and to 10 billion in the twenty-billions in 2003. Fifth year. And his point, he says that there is-- if the company can do better models in the chosen directions, the revenue begins to catch up on the scale of the cost.
And once again, that to date, all the big companies that are developing artificial intelligence are created only because someone subsidizes it.
You know, we're trying to start from scratch. This is crazy. But I'm sure if we can create better models in the areas we're focused on, everything will be fine. And you know, I'd say that, in my opinion, it was a good year for both Google and Anthropic. And I think we have something in common. The fact is that both companies or at least research units of both companies are led by researchers, which focus on models and addressing important global episodes, right? They are guided by complex scientific challenges. And I think it's the companies that will succeed in the future. And you know, I think that's what you and I are gonna do.
We can talk to you about how companies are perceived to be incorrect and so on. But this market was actually just because someone subsidized it. Just like Elon Musk once gave money to the OpenAI, and then Microsoft gave money to the OpenAI. And Amadei says that both Googlenthrop DeepMind and Aic combines one incredible thing that research leadership and orientation into complex scientific episodes as some North Star. Speaking of OpenAI, I'm focusing on Anthropic and on Google DeepMind today, but I remember that the OpenAI was generally being done to make AGI and make an open AGI.
Do another open AGI. And literally, at some point in time, I think that six months ago, Max Grigoriev and I were filming the production of the Kremnival Valley culture, in these companies, the culture of these companies. And it was said that the OpenAI is actually treating all people as homos, in fact. Why? Because they're doing a model, they're doing AGI. They're just to get people to use and ask themselves for answers or some kind of micro-agent, I don't know,
right?
No one wants to create a better chat-bot. And even more, the leaders of OpenAI are talking about it, they're not listening. They say we're not here to build chat-bots and Soros. We build, uh, AGI, a common artificial intelligence. Here.
Yes, general artificial--
They build a common artificial intelligence. OpenAI was founded as a company that would build AGI against Google. Because Google has in principle started in the middle of the tenth, and, uh, working on these, these models, and so on. And a few of the directors realized it would be dangerous if Google was the only company that would have this AGI. And to make a counterbalance to create a counterweight that is accessible to other people, not Google, but for that technology they went and founded the OpenAI. It's-- they weren't gonna build a chat-bot, they weren't gonna build a video editor, they weren't gonna produce any products. It was, it was a laboratory that was doing a study on artificial intelligence to build a common artificial intelligence. They're still doing it. So nobody's listening to them. They're talking about it. Yeah? That, that-- they don't have a job to release some product. And therefore, when they don't produce another product, it's not worth surprise.
And what you can see from the OpenAI action now, and what people inside say they've been very serious about switching to commerce, not losing the market, to win this race now IPO SpaceX, Anthropic, OpenAI to get money. I mean, it feels like this North Star, it's supposed to be around somewhere. But it's hard for me to judge, because I think that it's time for the OpenAI to be in Anthropic all day, you have to be in Google DeepMind all day long to be in Google I'm sure you're gonna understand that. Because otherwise, well, it's hard to do.
So, model, the model has a positive scenario, and that a powerful artificial intelligence can help cure cancer, fight some diseases that are very serious, for example, tropical diseases, that it can do. Raise science, better understand the universe. But we all know you, huh, serious things, right? And this is not a decoration part, it's a story that, obviously, is necessary to avoid talking to fear. I saw this short-sized video today, where the former LinkedIn and the governor of California were making some discussion, and he was telling me what it was possible to create, so in California, access to chat-bot for any man, To free chat-bot to help him there on medical, legal episodes. And I thought it was like an old market. Because all systems are now free of charge. With medical assistance, legal aid, they are already financial, they are already available to people on a large number of episodes. There was still an education issue. They are already available because today the quality of free systems, even limited systems such as Gemini, ChatGPT, even free models, is superior, incredibly surpassing their results, which were unbelievably high. In the most advanced road models two years ago, right? Maybe he doesn't exceed the pro model free models, there, ChatGPT, when he first got out there, two hundred dollars. And it might be getting better, right? Because there's a very serious reasoning behind it, so much can be done. So, to date, people have access to incredible things, and it, it's already present.
How does AGI have to do with that, huh? So, actually, with the artificial intelligence, of course, the future, this negative turn through these images of a film like the olivion, I said, yes, and if you could ask a more developed question. TV, one question, he'd ask: how did you get through this technology teenager, that's Amadei talking about it, or didn't destroy yourself, right? That's a strong, strong image, I think. And technology is already powerful. But it is understandable that the State, institutions, the economy, habits, that is, control, debate on economic rights, it is completely immature. We see today as unbelievable these discussions, which go, especially in the United States, very much on the topic of words that deal with inequality, the topic of comparison of different races, and bias, including, It was like a white race in the US, which at least broadcasts that the fact itself, the fact that it was a comparison, right? The very fact that the word was racist. It was set out by the Ilona Mask schedule, and in general, it was very much in the X-ray on the subject, as the use of racism in the past, 10, 15 years, has increased, as it has dramatically changed. I'm in line. But it's a incredible shift, just, like, a fantastic shift. And we now understand that society has not taught to work with a person like this... to live, to work with such a life, to work with concepts as racism, and to use it, including, you know, in negative stories. And then we have artificial intelligence with you. We're talking about AGI, this is another kind of problem that the problem of racism, it does not exist as to the complexity of the issue, the complexity of the issue that actually exists in AGI. So we got a fire and we haven't learned how to build safe ovens.
That's what I think, a very good rhetoric. It is understandable that there are many risks that when an autonomous system becomes smarter than a person when the abuse of certain people begins, terrorism is incredible, bioterrorism is there, Cyberterrorism is completely different. Very complex actions by different States, authoritarian regimes, etc. I am, of course, when I look at some leaders, political leaders who speak from different countries, some of which is the trend that we are incredibly strong in the world, which we-- very much. highly described and visualized, over the past seventy years or the last eighty years. And you understand how many leaders don't understand what's happening in the technological field. And what is a artificial intelligence. And that question, which is bragging about the existence, I don't know, of nuclear weapons or the brag-- bragging of what is supposed to be domestic, means that technology, now, of course, is not comparable to what it is today. There's a lot of artificial intelligence in the world and where these systems go. These systems are not the ones that have launched drones or which systems intercepted some kind of weapon, but where they enter, and that these systems can analyse what they can learn, what they can do, what they do, They can be packed. That's not the fifteenth, the twenty-fifth, the thirtieth, is it? It's not somewhere, some teacher has been translating something. So it's a completely different system of general perceptions of everything. The other is the very reality of everything. So I-- one of the discussions, I think that's very important to people watching this tape, of course, is a topic that concerns jobs. And there's another important topic that concerns China and the United States.
And on the one hand, there is a debate and there is a debate in the world, which-- and by the way, there is a contradiction of Jensen Huang's perception of NVIDIA and Amadiyah in Anthropic, which, on the one hand, we can very much. I'm in a hurry to win. The US may rush to, for example, defeat China in artificial intelligence while creating a system that will be higher than all the people who created it. On the other hand, if you don't rush too much, you won't be able to develop artificial intelligence in terms of a closed loop, you can get off China. If China does this closed loop, that's it. He, he's gone, so he's up ahead.
And then the U.S. will be gone and never will ever catch him in his life. What do you think about this? And the other question that is worth, it is about these projections, which are that up to half of the initial office jobs can be lost in the horizon, there, a year in the lap of five years. And the topic of discussion is that until labour market statistics show a clear impact from artificial intelligence, and some changes can be explained by over-service after the pandemic, by a change in general. market economics, markets, new companies. In this case, for example, Hassanis, he believes that there will be a normal technological evolution in the near future, that some work is being disrupted but new work is being undertaken, new work is emerging.
New jobs are emerging, new skills are emerging, new professions are emerging and perhaps more valuable, by the way, well-defined. But at the same time, of course, it recognizes the influence on the initial levels of specialists. I don't see any problems with the initial level of specialists at all. Yeah, I think that all levels of people are being affected because a huge number of people in the 30s, 35 years are used to living like they've been in the 20s, and they can be very serious, too. To suffer, right? Just like the young people are, who have been making billions of dollars or hundreds of millions of dollars to date. We see them in the market AI, yes, the very young people who were the juniors. How did they do that, huh? And so all the advice to students in the series that are now given, they can only be given one advice, it's a strong user of artificial intelligence instruments. And here we are at the graduation on the recent sub-category that was Sunday, we had a discussion with Tanya about what Tanya said she wanted to go to school now, take some kind of art course. Intel in terms of design, a, different applications. And she said that there was like one friend of hers saying it was very important to spend 30 minutes with artificial intelligence every day. I said it's not a decision to give 30 minutes to artificial intelligence that artificial intelligence should be as stacked as usual cell phones you have in your pocket, just so you're the only one. You're learning, right? It's just like you can learn to cook food. You can learn to cook food if you cook different foods all the time, different completely. And if you go-- go, sometimes go to some kind of class or you'll consider it as just some kind of training. Let's, uh, set up 20 minutes and do it. This will lead to some partial development, yes. And that's a problem, because yet-- it's like being a good musician on piano, you have to devote every day to this incredible amount of time. And to see and see, and hear, and to learn music everywhere and everything, right? And the 20 minutes a day, the hard task. We need to start a life, we need to start this incredibly inspired, together with this, at any time, to pay attention to it, yes, that, that, that's, that's, that's, that's, that's, that's, that's, that's, that's, that's, that's a life. And, of course, what happens in artificial intelligence, my personal advice, again, to all students and to all people in the world, not to fear artificial intelligence and to use it indefinitely. I used to-- we did a lot of training to-- development IT in Russian-speaking space. And I always said that the biggest thing that could be done for development, for the development of computers, and not for the influence, but for the development of people in terms of computer knowledge, it's all. People, so all people can go through blind seals, right? So if a man passed a blind seal, he stops being afraid of a computer, he can easily do certain things. It is difficult, for example, to write a letter, to get some text in Word. He's getting easy. He's got a very easy perception of the keyboard, he's running his psychological barrier. Once the psychological barrier goes away, so he starts using a computer as not as complicated, he starts using it as something natural. Again, it's like a knife in the kitchen to cook food. So you're very calm and easy to do that. You don't see any problem in this, you don't have any, uh, stress. My mom's got a stress if she needs to go to another country. She can't buy tickets. Even my wife has this stress. I'm like a man who bought tickets incredibly, a lot of money, I don't have any stress on this. I mean, I automatically know a lot of systems, parts, everything. It doesn't mean I'm super smarter, it's a venom-- just a part of my life, yeah, that I live. There's a very high risk, and, uh, I think it's one of the most important parts of the discussion, and I want to give it to you at the end of the roller. And listen to him carefully. There is a story that concerns globalization when a part of people felt lost, and the political backlash was very, well, very strong. I wonder if the same thing would happen to artificial intelligence. And of course, that risk is very high. And the industry, in terms of discussion at all today, in the artificial intelligence market, the industry must demonstrate, and, look forward to, the public goods, not just the future benefits and the future benefits. The incredibly high number of cabs in terms of disease treatment, the creation of new sources of energy, such as some examples of people ' s assistance every day. And show, you know, teach society that artificial intelligence, it not only takes a job, but solves big, huge human problems. Helping people introduce all these tools, yes, because there may be an incredible problem of large mental, mental, psychological, psychophysical trauma, and we can get very serious backs and backs. political, economic. What do you think about this? Write it up, recommend our channel to your friends. Don't forget to sign and support the channel, put the likis, write the comments. It helps us to develop very seriously. Before we meet!