Artificial intelligence isn't taking away your profession, it's doing something more dangerous. It's taking away the entry into the profession. Take programmers as an example. For years people were told: "Learn programming, it's the profession of the future." And we said it so many times too. And now the market says: "AI can write the code." And a strange illusion begins, as if now everyone will become a programmer. But that's not true. Generating code doesn't mean creating software. Writing a screen doesn't mean understanding the business.
Putting together an app, what's called vibe coding, doesn't mean building a system that works in a real company with customers, money, errors, security and responsibility. And this is where the main turning point begins. AI doesn't just lay people off, it pulls the simple tasks out of the profession. And if you pull the entry-level tasks out of a profession, the question arises: how is a person ever going to become a great, serious specialist in five years? Today the debate is too preemptive. Some say: "AI will lay everyone off." Others say: "AI will turn everyone into super-specialists." But the reality is more complicated.
The winners won't be those who simply know how to use ChatGPT, Codex or Claude from Anthropic. The winners will be those who understand the business, the product, the data, the processes, and can build a real AI-native system around all that. And in this episode we'll break down which professions are really under threat, which will become more expensive, why the labor market is currently fooling itself, and why "everyone will become a programmer" may turn out to be one of the most dangerous half-truths of the AI era.
As of early June 2026, of course, if you look at the labor market and the most serious picture, there's no sign yet that AI is going to lay everyone off. But the market already looks like a story of the entry point into a profession breaking down, right? The most dangerous blow isn't falling on the programming profession as such, obviously, though we'll talk about that in a moment, but on junior roles, various routine tasks and entry-level employees. And the ones who do administrative work, part of the analytics, template content, support, where the result can be checked very quickly.
Just recently I was telling a story, a case I found quite interesting. In general, I'm interested in my employees working for me for a long time, since I have a great many ventures and businesses where I'm personally involved as a professional, more than twenty. And I'm interested, really interested, in having people. And in one company there was this situation where I asked the director to help me find certain data. And this director was using just plain ChatGPT. And then I went further and asked an analyst to provide me with a report.
And while they were all putting it together, I saw that my request to a simple system, just plain ChatGPT, led to me saying right away that I'd simply let these people go, and I no longer wanted to work with them. And when I talk about checking data now, when I come into any business, whether as an owner or as an auditor, or onto the board of directors, I always ask to be given the raw data on that business, so I can run it through AI models and do a very quick assessment of what's going on there.
So right now I don't need ordinary assistants who do and generate routine things that take me a very long time to explain. But does that mean employees will go away? And here I want to bring up a topic, starting first with programming and the concept of vibe coding.
I hear it a lot: when I tell someone I'm doing something in Codex or in Claude, everyone says: "Oh, so you've taken up vibe coding?" I want to explain a bit how I see vibe coding. After all, vibe coding is working with code. And this term originally, if I'm not mistaken, it was actually Karpathy who first talked about it, about this term, and then it became an all-encompassing official term that absolutely everyone uses. But still, this term was about systems like Claude Code, for example, or Codex in ChatGPT, helping you write code and create code for software development.
For me today, the reason I say that I, for example, don't do vibe coding, or that Codex is fundamentally not about vibe coding, is that fundamentally it's about starting to develop systems that are called AI-native. Software that is a system developed purely by artificial intelligence. And when I develop them, I never ever open the code, I don't see it, I don't look at it, and I don't want to touch it. And even when Codex or ChatGPT suggests that I run some check or launch something in the DBMS, well, anything to do with databases, or in some specific one, or on some server, or in GitHub with the source code.
I say: "I'm not going to do that, you have to do it yourself." So we understand that today the market perceives vibe coding, at least I understand this for sure, it perceives it differently. And vibe coders themselves see themselves a little, a little, a little differently, right? And probably every day I come across people who say they're making some software, some bots, some systems to help a business, some systems that help their employees, some systems that connect people with each other, some with different models plugged in inside, that run on phones, without phones, people check their customer acquisition cost and so on.
If ten people start discussing it all among themselves, it might seem everyone is doing the same thing, but in fact everyone is doing it completely differently. And it's very important for the people who are interested in what's going to happen in terms of earnings, and which professions will truly die, which will be rebuilt, what will remain at all, to be in the loop on what's really going on in AI. Because nine out of ten people I talk to keep saying: "Well, we understand all this, we know all this, we've got it all figured out.
We can see that AI, I don't know, the AI market is overheated, or AI doesn't actually do the things you all talk about, or somebody is missing something. Or yes, we've got great systems in our own country too. I don't know, we've got great systems in Russia, or there are great systems in Europe, or why not use the Chinese ones? Or why not set up local models?" And at the same time, usually maybe one person in fifty understands the difference between ChatGPT, say, 5 and 5 Pro, and the ChatGPT version on a Plus subscription or the free version.
Somewhere someone starts naming, well, there are some other systems. Claude gets called "Cloud." When you ask what the difference is between 4.8, 4.8 Max and 4.7, and what the difference is in how they handle different tasks, people have no idea. So there's an abstraction, a notion of AI, and everyone wants to be very smart about it, knowledgeable, recommending various spaces, software and everything else. So our task, and that's what we do on the ToTheMoon channel, is to understand all of this a bit better and get a bit more of a grip on it.
It's very important, because the topic of layoffs due to AI is very relevant in the world today. Why?
Because big American companies, corporations, keep talking about it. And bloggers and economists pick up this trend. And most importantly, people who very often don't understand anything about it at all. If we see a trend of layoffs, then it's very important here to look at it head-on. I like Jensen Huang at NVIDIA, the head of NVIDIA. He says people often talk about layoffs linked to AI, but in reality those layoffs are linked to the company's financial performance. So they justify the layoffs by saying they'll switch their employees over to AI. Justify, exactly.
And if you look, there's this Stanford report, the AI Index 2026, and the layoffs, of course, aren't mass layoffs yet, but what you can see is that the market for junior specialists is shrinking. That's what I was saying. So how will great specialists emerge then, if junior specialists are going away? And who are junior specialists anyway? Again, I'll continue about this index in a moment. If you look at junior specialists, in the world of AI today a huge number of the world's coolest startups, which are even worth tens of millions of dollars, or hundreds of millions of dollars, or billions of dollars, many of them were made by people aged twenty, give or take, very young.
So these people, they were the junior specialists. And they became great in AI because they created something. I've always said there's a huge difference between junior specialists and top-level specialists. And I've always urged people not to raise salaries for so-called work experience, based on how many years someone has worked, because ten years on they may start working worse, not better. Or because someone has worked as an accountant for twenty years, does that make them a professional accountant? Not always, not at all. Often people who've worked as an accountant for six months are far more professional than someone who's worked for twenty years.
It's a question of something else entirely: the structure of their thinking, their knowledge, the adequacy of their perception, their brain, their ability to communicate depending on what role they hold. And when people become presidents, they're always junior presidents, right? Whenever people have created some very cool, serious businesses, they were always juniors when they created those serious businesses. And there are actually not many people in the world who've created some great businesses, and then take them, run them in parallel and create a lot more great businesses.
Or for whom all these great businesses work out. I've always said: ask any oligarch in the Russian-speaking world, or some super-successful businessman, to open a business from scratch somewhere, only truly from scratch, where nobody knows him. Well, various bloggers run experiments like that from time to time, again, but usually it doesn't lead to anything good. So, the Stanford AI Index 2026, one of the most important studies on this topic overall.
It records that the generative AI model, meaning the AI that exists now, has reached 53 percent adoption in three years, which is faster than the personal computer or the internet. So AI adoption is growing phenomenally, at a phenomenal speed. Let me remind you that OpenAI's weekly users are already around a billion. Gemini's monthly users are approaching a billion. Google AI Mode, by the way, has already reached a billion regular monthly users. So, well, a large number of people use AI, but the effect on labor so far, I mean, what does Stanford say? 53 percent growth, but the effect on labor is uneven.
And the main pain point isn't a general collapse of employment, it's young workers. What's happening with young workers? And, for example, among developers aged 20 to 25, employment has fallen by almost 20 percent over the last two years. But again, it depends how it was measured, by whom, and what was done. And it's clear that in recent years people have been expecting crazy cuts. All the studies, by the way, that big organizations like Stanford often do, they are, to be honest, old, meaning they go out of date, and they're always done on a specific sample, a specific sample of companies.
Who do you measure yourself against? Should companies located in Belarus measure themselves against America, against American tech corporations? Of course not. Should companies located in Russia measure themselves against Indian companies? Of course not. These are completely different markets. What else can you see in the studies? A very important point, if we're talking about professions: investing millions of dollars in developing and implementing AI in companies, in corporations, hasn't increased the profits of those companies, those corporations.
And even where companies… I know a lot of corporations and people, by the way, who don't want to lay people off in order to implement AI. That is, they want to implement AI and increase, strengthen their quality and the efficiency of their companies. And sometimes, from the market, it may seem to you that these are people or businessmen who, on the contrary, are aggressive and want to fire everyone. In reality, that's not the case. I'm fairly well known for easily firing large numbers of people, but I'm not interested in firing people.
The point is that I'm interested in adaptation, in how people use AI and how much we're keeping up with the times, and whether we, or my employees, or my partners, are thinking right now not with two-year-old thinking, with brains from two years ago or five years ago. Usually on the market it's five or ten years out of date. And how much do they understand what's really going on right now, and how well do they truly grasp it? By the way, once again: to understand AI, of course, you have to work with it constantly, constantly work with it.
You have to have these systems set up, you have to have paid subscriptions, and you have to be doing something. Even doing things that may not seem super important or super necessary. But you have to try to create something. Not create software, not create something to make money from it, but create even something for yourself and try to put yourself in new situations in your own life, that is, to find new use cases for yourself. Recently, interestingly, my sister came to me and said: "Sasha, listen, I've decided to download a full export from ChatGPT, of all my chats," — well, she has hundreds of them, maybe thousands, I probably have tens of thousands of chats by now, — "and analyze them using Codex, and it'll be able to recognize a lot of things for me and help with what's going on in my life."
And it hit me. I thought, why haven't I done this? Because I really do have dozens of projects created in ChatGPT, and I have a lot organized there. But at the same time I have chats going back several years and a huge amount of unstructured data. And it would be really cool for me to hand this structuring over to Codex and write the core files. What would I like to do, for example? This is for you as a very interesting extra case on top of today's topic that we're discussing.
For example, have it collect all my data, everything I've ever said about medicine, and structure it into a single knowledge system. That is, create a knowledge system, some file, or structure some MD file, or PDF, or DOC, or HTML, it doesn't matter, or even some PRO, or structure, for example, my data about my family's life, or my individual trips, or some of my hobbies. For example, I don't know, I have a big garden, I have hundreds of plants, I have dozens of different tools and products and fertilizers, or hundreds by now, I don't know what I have in my garden.
It's all tied to the climate, the location and everything else. So have it structure that. Have it structure everything to do with administering the household side of my life, everything to do with cleaning, staff, people. Have it administer everything to do with… and collect my data concerning, for example, my finances or taxes, or my business and companies. So, structure all the many pieces of data I have scattered across different chats, because ChatGPT can't do that on its own. But what can Codex do for me?
Codex can create files for me, and then I'll create a project in ChatGPT and attach that file as a source. And that will be structured, normalized information. By the way, interestingly, you have to understand that ChatGPT can't process your thousands of files, even if it has memory, right? It uses some data that it considers important, whereas here you'll show it what you consider important. Many people think there are lots of systems that can do this. Yesterday the transcription of my Zoom recording got triggered by accident. I was holding a private session there on people's spiritual development, and the transcript of that recording came out, got sent to me.
Not even a transcript, it's more like a summary, what's called the AI summary from Zoom. That's where I got a summarized breakdown of what I'd been discussing. So, I read that breakdown. I realized, of course, that the system had completely misunderstood what we were talking about. And why? Because Zoom used some very cheap model. They're not going to run an expensive model to understand what's going on there. And that system also doesn't know who I am. It doesn't know my style, doesn't know my approaches. It has no data on what I mean. And it was totally wrong there.
So where I said I like red, it wrote: "Alexander said he doesn't like red." Where I said I believe in Santa Claus, it wrote that I don't believe in Santa Claus. Well, I'm just giving you a hypothetical example. So, the mass market is wrong to discuss a profession as a single whole, if you look at it.
And in reality a profession decomposes into tasks. And AI first takes away chunks of the work, then changes the team structure, then changes hiring, and only after that do we see the various layoffs. And the market in many countries reacts even more. The economy especially reacts even later. The law reacts even more, politics reacts even more. Not to mention that education reacts even more. And of course, I have four children, and right now deciding where a child should study and what to do, how they'll develop, is incredibly hard.
I mean, there are more or less simple things. My daughter, she's fifteen now, she's learning to drive, because in America you can drive at sixteen, and she's already clocking up hours, and I'll buy her a car by October, even though we have enough cars, so she has her own. And I could buy her a Tesla with full autopilot, which is already pretty decent. And I realize she'd never actually learn to drive. I think, well, let her learn after all, since she wanted to, and I'll buy her a car where it's not about the autopilot, more something fashionable, interesting, beautiful.
But my three other children, I realize they won't be learning to drive anymore, unless they show very strong enthusiasm. Because we all understand that in about a year or two, there'll be what Tesla calls the Cybercab here. These are cars without rear-view mirrors, cars without a steering wheel, that is, fully self-driving cars. And for this car, obviously, you won't need a driver's license. Obviously it'll take time for a law like that to come out, but they won't be needed for it, once they're no longer needed. Why should my children learn to drive a car?
So, it's the same thing right now when it comes to education. Or when we talk about a profession, just imagine, it turns out every person has to see this kind of trend for themselves. What's going to happen? And that's very hard to see. I mean, with cars it's more or less clear to me. And even then I understand it may get pushed back in time. In many countries it won't happen. For example, the Cybercab or a trend like that won't appear quickly in Belarus or in Ukraine, or Russia, won't appear quickly in Kazakhstan, won't appear quickly in Italy, won't appear quickly, it won't show up quickly on the roads in Vietnam, but it may happen on the roads in China, it may happen on the roads in the USA, right? Very quickly.
And how do you, being in a particular country, make a decision? That's what matters, it matters a great deal. I mean, what should I do? In one of the episodes I gave an example. My nephew lives in Poland, he's studying at a polytechnic institute or university, something related to engineering, science, science and data, and everything connected with that, including programming, of course.
I told him that in order to keep up with the times right now, you need to create something yourself every day on the trendiest, in the trendiest systems connected with AI. And for that you need to follow ten or twenty bloggers, oh, not bloggers, well-known people in the AI field, like Karpathy, or like Dario Amodei at DeepMind, or those at Anthropic, or like the people at DeepMind, at Anthropic, the people at OpenAI. So you need to follow people, the engineers inside, the engineers who are building xAI, including following various Chinese engineers, and watching the cases they post, what they're doing, what interesting things they suggest doing, and constantly trying it, endlessly creating things yourself.
That's the only way to survive right now, in terms of, well, in order to be in a certain top layer, right? If you just go to lectures at university, that's unlikely to help. And now it'll help even less, even less. If you just solve the tasks a business gives you. That, by the way, is one of the big deceptions. That's unlikely to help either, because right now everyone endlessly says they're using AI. Eight out of ten businesses tell me they're using AI. You start going deeper into the details, and you see just how little people actually understand about AI.
That's a big problem, of course. And by the way, it's probably one of the biggest problems there is when we talk about professions, which professions will die out and which will remain.
It's that there's a huge amount of disinformation on the market. You keep listening to everyone, and it seems to you that everyone in every company understands AI, or that all companies apply and use AI. But if you dig in and look at even just some individual area, you'll see how illiterate people are and how they actually don't fully understand what AI is at all. This is very important. As I said, today people don't understand the difference between plans at just 20 dollars and at 200. Although what are 20 and 200?
Well, imagine, heads of American corporations get up and say: "If our programmer earns more than 500,000 dollars, and he doesn't burn more than 500,000 dollars' worth of tokens. He earns 500,000 dollars a year now, and if he burns 500,000 dollars' worth of tokens a year, meaning AI for writing code, for building systems and so on, then we don't need him." That's a somewhat different way of reasoning about what AI is, and about what can be done in these systems, and about what the systems do and create.
It's completely different from people who've opened up some little system and developed something somewhere. Or people simply don't want to invest 200 dollars, or invest 100 dollars, in their own development. Now you can write in the comments: how can I do that if my salary is 300 dollars, or, I don't know, 50,000 rubles in Russia, or 1,000 dollars in Belarus. What do you do if a person earns that much money? How am I going to invest and buy myself a ChatGPT Pro plan for 100 dollars, right?
And my sister told me that my nephew, so, he has an inexpensive subscription, he chipped in with a friend, they bought Claude from Anthropic, and, but the tokens are very expensive, well, that's because you didn't buy it. Buy yourself their Opus Max version for 200 dollars, and you'll practically be unable to use up your tokens. Here the question comes up: how is a student going to buy himself software for 200 dollars? But that's what will let you earn tens of thousands or millions of dollars in the future. That's the whole point.
I can roughly imagine what bloggers, or lots of people in the info-business, are going to write now. Recently I saw this nonsense they write: subscribe, and I'll tell you how to sell your own training, your own personal mentoring, and earn 10,000 dollars per sale in the first month. The point here isn't to earn 1,000 dollars or 10,000 dollars right now. That's not what I'm calling for at all. In this video you won't hear how to earn 1,000 dollars or 5,000 dollars in your very first month. That's ridiculous, that's ridiculous reasoning. I'm not talking about that at all.
Can you earn 1,000 dollars with AI? Easily. Can you earn ten? Seems easy too. But the question is what does "easy" mean, and for whom? For what kind of person is it easy? And the point is that right now you really need to build your knowledge for your whole life. I always focus on people. Maybe this information isn't for everyone, and YouTube doesn't like it. Give quick tips, a quick recommendation, that's what's required. But I'm talking about a person, gradually figuring things out, figuring it out, figuring it out, laying a phenomenal foundation for their whole life, and being able to use it throughout their life, and never being afraid of whether they'll have problems with work or not.
That's never happened to me. I've spent a lot of time as an employee, and in senior positions, and as an employee in senior positions while owning a company, and simply as the owner of my own business. And that's actually how things are for me. Or on some boards of directors. I've always said: I'm absolutely not against going to work somewhere, I'm absolutely not against running my own business. I consider myself blessed in this, that I'm not afraid and don't worry about doing either one or the other.
And so, to form this strength and this quality in you in terms of approaches and choosing your professions. To see these trends, understand these trends and make the right decision for yourself. What can I do? And the more I talk today, by the way, in this short episode of ours, the more I see that the only thing a person can do to strengthen their profession is to understand AI. And to understand it, they mustn't be afraid to invest in it, mustn't treat an AI subscription like a Netflix subscription.
Don't treat it like that, treat it differently, don't treat AI, buying it, like going to a restaurant. Even if you go to a restaurant once a month for 20 dollars, it's better to spend 20 dollars on AI now and skip the restaurant. That will give you far more opportunities now. But I have a huge number of friends who go to restaurants and spend hundreds, thousands and tens of thousands of dollars, but don't buy themselves the Pro version. I even have partners who don't buy the Pro version for 200 dollars. And it's astonishing! It's just astonishing.
A person can easily drink a glass of champagne for 200 or 300 dollars, but about the Pro version they say: "What for? It's expensive." That's a misunderstanding of the market and a misunderstanding of what's going on. One of the interesting stories, trends happening now, is that big companies like Google, like Microsoft, can do analysis on large amounts of anonymized data.
And so, or fun trends. Google started doing tests like this. They set up cameras to see what their employees do, or Meta, Meta does this too, to see what their employees do during the day, in order to automate and optimize their work. I'd do that now too, by the way. Not to replace employees, but to boost efficiency. To increase efficiency. So when, for example, I set the task of optimizing manual labor by 90 or 95 percent, the main point isn't about cutting employees, it's about multiplying the business, or surviving in business at all, or getting a different business of a different scale or a different volume.
It's not about cutting some number of people, especially in companies where payroll makes up, say, 20 percent or 15 percent. So I save 10 percent. So what? My task is how to 2x or 3x the business. Or how to get the business's efficiency in terms of net profit to 50-plus percent, for example. Now that's really interesting. Microsoft did a cool study within its Bing browser, and they analyzed 200,000 anonymized Bing Copilot conversations and found that AI is most often used for work like gathering information, writing, learning, advice, communication.
So it's very simple office work that's going on right now. Does that mean office workers will disappear? Not at all. I mean, will we be coming into offices where there's no secretary sitting there? Will people set up meetings themselves, do it through assistants? That's a funny thing too, by the way. How many startups I've heard about here from well-known people in the Valley, who said the coolest thing would be: if you build an AI assistant that will schedule my meetings, that will read my email, that will keep track of my expenses, then that's a super cool, essential thing.
I don't see a single person at all who seriously works on this and uses it as something extraordinary. It's just some very strange paranoia that scheduling meetings is difficult — it's an absolutely trivially solvable task. I know people whose schedule is thirty days a week, fourteen hours a day, with meetings running fifteen, twenty minutes each, and their schedules are fully set up. It all depends… And at the same time I know executives, like the story they always told about the chairman of the board of the Russian bank VTB, that he has a vice president who earns 100-plus thousand dollars a year, or a million, oh, a year, earns 100-plus thousand dollars a month, and he can just sit waiting in line for him for six or seven hours in his reception room and do nothing.
The approaches are different everywhere, it all depends on the person. By the way, in terms of programming, with what's happening now, a whole new class of people is going to win.
And this new class of people are the people who understand how a business works, who understand it, understand it from the point of view of who the customers are. And these people are going to take off incredibly fast. These are the people who used to be severely limited by rank-and-file staff, severely limited by analysts, severely limited by programmers. These people can break away incredibly far ahead, just amazingly far ahead. The people who, as I see it, are usually the trendy ones, like marketers for example, they won't get even basic success.
Only if it's a marketer with strong skills. As in, he came into marketing with strong analytics skills, strong business analysis skills, and can create fundamental solutions himself rather than spending other people's money.
Another interesting story about professions. There were a lot of claims, you heard this especially two or three years ago, that AI wouldn't destroy jobs that require manual labor, for example plumbers, electricians and so on. Personally, I think the work of plumbers and electricians will be destroyed more and more. First, the work of working through these issues as a whole will be done to a higher and higher standard, because people will be applying AI in their work, and a huge number of ordinary people will be able to do this work without plumbers and without electricians.
And plus, automated robots will appear, specialized robots, new specialized devices will appear that will do this work better and better, and better, and better. So this part is going to get covered, including among builders, including, and this is very important to hear, including builders. I'm deeply convinced that precisely where work was hard to automate, that's where there'll be a phenomenal restructuring of the market, just an incredible restructuring of the market, and we'll get a completely different effect.
This automated, hard-to-automate work is close to cars like Elon Musk's Cybercab, which has no, which has no driver, and which costs 200 dollars a month to lease. I mean, he says this car will cost 200 dollars. That's a fundamental revolution, it changes absolutely everything. And here you have to look at the use case. People start looking at use cases where it could work for you. I hear people, and how short-sightedly people look at things. They say: you'll buy the car, it'll be able to work for you, it might be able to work as a taxi driver and so on. They discuss these things.
But the market itself will be restructured at its core. Because if I have a car like that at home, and my kids, for example, can ride in it, or the car can drive on its own, then I'll give it jobs to do that are connected with my home. Overall, everything will be restructured. So it drops one kid off at school, then the second, then the third, then the fourth, then, I don't know, it dropped off, all but walked my dog. But I'm joking. Hypothetically. It went and picked up some things from someone. I mean, it's a different, a completely different life.
Obviously, what I've just said isn't a two, three, four-year horizon, because a great many laws will have to be modified. And still, there are politicians who aren't interested in such cars appearing. The market isn't fully interested in this, in this happening. But it will definitely be heading that way, and maybe it'll get there very quickly. Could it be that such cars appear everywhere? Here, where we are, you step outside in Silicon Valley and a Tesla is the people's car, it's everywhere.
Although Tesla isn't super focused on producing cars right now, they're focused on making, on producing robots and doing the combination of robots, AI, satellites. So they're creating a different future. Tesla is interested in putting out a billion Tesla robots. That's a market about something else entirely. The point isn't the cars, the point is what data they have, how they operate across all these systems, how, where the processes even take place, and who will even need to travel and when, how these systems will be synchronized with each other.
We can spend a long time discussing the driving rules for these systems, or the right road markings for these systems, or thinking about roads. But essentially, if such cars exist, they change everything that's going on, the whole reality around us. And in San Francisco, San Francisco, I think, is a city, well, it's certainly the incredible leading edge of the whole of Silicon Valley, ahead of the whole world. When there are companies here worth trillions of dollars, and the total is, well, just trillions, the very richest companies, when there are people here who earn tens of millions of dollars a month, when there are cars without a steering wheel driving around the city here, like Zoox — Amazon's car is driving around now, or even Waymo.
The whole city of San Francisco is a super complex city, oversaturated with this, and at the same time you've got buildings being rented out that are 100-plus years old, some half falling apart, some just odd, and at the same time you, you look out the window, or you often come into the city and see the same underdeveloped coffee shop, or the same very strange service in a restaurant that's 150 years old, and at the same time there's something new. So there's a very strong concentration of the new and the old at the same time. And that's the paradox of how incredibly advanced everything here is. It's incredibly cool.
My wife and I were rewatching the first film "The Devil Wears Prada" a few days ago, and somehow I ended up watching it too. And it's such an interesting film. The vibe of the 2000s. And there they, the main characters bring the lead, the other lead character, a Starbucks coffee. And you look at that cup. And I, as someone who lives in the USA now, I live in that same culture. I see that same Starbucks. And you realize that twenty years have passed, exactly twenty years, and Starbucks has exactly the same presence.
So it's not something new that has appeared and is something ultra-modern for people. Restaurants, you can see this very strongly online in Africa, or you can even see it in Belarus, in Ukraine, in Russia. A huge number of companies are very new for people, very fresh. They weren't trends. There are no Starbucks, no Home Depot, no trucks that move your furniture from house to house. They haven't been around for that long. Even when it comes to the mail, even these mail trucks, they're organized differently than even such a long-standing company as the one called Russian Post.
Because this company is still different from the Russian Post that was the postal service in the Soviet Union. They differ, they're different. This life in Silicon Valley and life in San Francisco, in terms of having the old and the new, is very fundamental to how AI is being born here. In AI, of course, in professions, there's also a dark story.
The dark story is that there's a probability, I remember Elon Musk said about two years ago that there's a probability that AGI will be the dark side of the world, and that's it, it will take over everything. And he put that probability at 20 percent. That probability, I think, is like a weather forecast right now. So, there is a probability, of course, that with professions, for a huge number of people, there will be problems with professions and problems with work for a huge number of people.
And right now a huge number of people are losing their jobs. But that has been happening for the last twenty years in various countries. All these various reshuffles. Is there a probability of a very severe crisis? There is. There's the opposite probability. There's a probability that AI will give people the chance to live better while, for example, working less. There's a lot of talk about that too. By the way, now, in the last two years, the last year, and the 2026 trend is not to talk about that. That was the trend, I think, of 2023, 2024.
There was a lot of talk about AI leading to the state, to people no longer working, and the state paying every person some money. That's just some futuristic speculation, right? Because what could happen, this range of probability, is very wide, very big, but you have to allow, within whatever you do, that your life really could be completely restructured. Just as it, just as twenty years ago my children, my wife weren't a factor for me, I had a completely different life, so I understand that in twenty years I may have a completely different profession, completely different hobbies, completely different views, completely different approaches in terms of professional activity and in terms of earning money.
I definitely urge each of you to be ready for that, okay? And the most important thing I urge you to do is to stay on top of the AI trends, which is what we do here on the ToTheMoon channel. Don't forget to subscribe and write your questions and comments. Bye.