YT YouTube2Text

YouTube2Text → Transcripts → Как выжить бизнесу в эпоху вайбкодинга и искусственного интеллекта? | Вопрос-Ответ с Маргуланом

Как выжить бизнесу в эпоху вайбкодинга и искусственного интеллекта? | Вопрос-Ответ с Маргуланом — Transcript

by МАРГУЛАН СЕЙСЕМБАЙ · 16,092 words · 2,580 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00We are here at Lake Balkhash in such a
0:02beautiful place. Behind us is Lake
0:03Balkhash, the channel, the river. And
0:06we decided to light a little bonfire.
0:08And we decided to have a chat. And what
0:10can one chat about in modern times? In
0:12principle, we talk a lot about business
0:14, about each other's businesses,
0:16analyzing them and sharing experience,
0:18but at the same time, a common theme
0:20unites us all. We are all looking
0:24forward. All of us, our gaze is fixed
0:26forward on the development of
0:28technology. And each of us is concerned
0:31about the future of business. Where
0:33business is heading, how it will
0:35develop, how artificial intelligence
0:38will affect it, robots, and in general,
0:40what is vibe coding, which of us is
0:43doing vibe coding, and what one gets
0:45out of vibe coding for themselves and
0:47their business. And today we decided to
0:51dress up a little bit. In reality,
0:54exactly 10 minutes ago we looked
0:56completely different. Some even washed
1:00up for this, for this shoot. And so we
1:05gathered today to have this kurultai
1:08and talk about it. Well, shall we
1:11broach the subject...
1:13of the future? What about us?
1:14What are we to broach? Margulan
1:15Kalievich, we ask the question, you
1:17answer. Actually, actually, it's just
1:20that each of us thinks about this all
1:23the time and in essence answers
1:25ourselves somehow, crookedly or slanted
1:27, but we find some answers for
1:29ourselves. It's just that today the
1:32idea is to throw all thoughts into the
1:34center and understand for ourselves
1:37where all this is moving and whether we
1:39are thinking correctly at all. Go ahead
1:41, Kostya. Do you want to? I see,
1:43judging by the fact that you’re
1:44sitting there all red and tense, do you
1:46want something?
1:48Actually, I got sunburned. We are at
1:50Lake Balkhash. The sun is very
1:52scorching here. Yeah. And Margulan
1:54Kalievich, I want to ask you a question
1:56. Tell us, what is vibe coding? Many
1:59simply don't know what it is. And, uh,
2:02tell us your opinion, why is the
2:04current world moving precisely toward
2:07vibe coding and its development? Oh,
2:10you know, vibe coding. It’s
2:12interesting, I recently read a phrase
2:16from a smart guy. He says: "Everyone
2:20thought that vibe coding is when people
2:23who have something were given the
2:26ability to code." Actually, it’s when
2:30people who know how to code were given
2:32a vibe. Do you feel the difference? Yes
2:37. And now there is a general trend like
2:39that. Everyone who didn’t code before
2:40and didn’t understand a damn thing
2:42about it suddenly jumped in and started
2:43thinking that it’s for them. Well, I
2:45did the same thing. I jumped into it,
2:47since I didn’t know how to code
2:49before. For me, it’s Greek to me. I
2:52thought, oh, now I can code. I rushed
2:55into it, but since I can think
2:57structurally and understand
3:02architecture and build processes, I
3:04very quickly realized that without
3:07understanding the base of building
3:10applications, creating apps, the
3:12architecture in general, and the very
3:15essence of coding, when you AI-code,
3:18you just create garbage. That's why
3:23it's interesting, all the principles I
3:26teach in business apply perfectly to
3:28AI-coding, because AI-coding has the
3:31same exact processes. And the most
3:34interesting part is, well, you start,
3:35just like in any business, by first
3:37understanding the architecture. When we
3:39build a business, well, the right way
3:41to build it is to first understand the
3:43tax scheme, the legal scheme, the
3:45management structure, the company
3:47structure, build it correctly, and only
3:49then fill it with substance: what
3:51specialists you need, what their
3:53functions are, delineations, what is
3:54allowed and what isn't, individual
3:56qualifications, and so on. And the same
3:59thing works in AI-coding, especially
4:02when you are building a multi-agent
4:03architecture and a multi-agent pipeline
4:06, or so-called workflow. And another
4:09interesting thing in AI-coding is, you
4:12know, many complain that artificial
4:14intelligence hallucinates, lies,
4:16presents wishful thinking as reality,
4:18and so on. That is all correct. But
4:22with one condition: if you manage your
4:25work with it incorrectly; if you tell
4:28an AI, "make it good for me," it will
4:30lie to you, it will hallucinate, it
4:33will skip steps in the algorithm, it
4:35will tell you it's done when it
4:38actually hasn't. Then I thought, "How
4:41is it any different from a human?" A
4:43human does the exact same thing. If you
4:45go to a person and say, "Make it look
4:46beautiful." Well, they will make it the
4:49way they think is beautiful. Plus, they
4:52will do it the easiest way possible.
4:55Plus, they will present wishful
4:57thinking as reality. Exactly the same.
4:59And it's interesting, when I started
5:02building multi-agent systems, I
5:04realized that it actually clears your
5:07head in terms of how to interact with
5:10people correctly. It’s interesting,
5:14you learn to build relationships with
5:16agents, but these same principles work
5:18for relationships with people.
5:21Therefore, it is very important to
5:24ensure that the LLM clearly understands
5:26, first, the goal of what you want, and
5:29second, understands all the metrics,
5:32parameters, and criteria by which the
5:34task will be considered complete. Plus,
5:38it must clearly understand the sequence
5:41of the pipeline. First this, then this,
5:43then this, then this. For this, you
5:46create separate agents and so on. And
5:49at the same time, tests, checks, and so
5:50on are being performed. But my point is
5:54that at first, I rushed into AI coding
5:57to understand more about coding, but by
6:00doing AI coding, I started to
6:02understand more about people. That's
6:07interesting. And about management flaws
6:10, about the shortcomings, well,
6:12relatively speaking, the mistakes
6:14businessmen make in managing people.
6:17And in general, if we look at the big
6:20picture, why I took up AI coding is
6:22because I generally like to constantly
6:24track what's new and catch trends.
6:27Everything, where everything is heading
6:29, right? And accordingly, I like the
6:31saying by Wayne Gretzky, the Canadian
6:33hockey player, who says, "I, my secret
6:36to success is that I don't run to where
6:38the puck is now. I run to where the
6:40puck is going to be in a few moments."
6:43So, in this regard, I don't engage in
6:45business that is relevant right now. I
6:48immediately start working on business
6:51that will be relevant in 2, 3, 5, or 7
6:53years. And how do you figure that out?
6:56To do that, you first need to
6:58understand the full power of the
7:00technology, understand what it provides
7:02, what the advantages and disadvantages
7:04of this technology are, and how this
7:07technology can be adapted into business
7:09. That's why I dove deep into AI coding
7:12. Not for the sake of AI coding and not
7:15so much to write an app, although I'm
7:18writing 12 apps right now, but more to
7:20understand where business is headed,
7:23how this technology will affect
7:25business, and what the business of the
7:28future will look like. That is, in
7:31principle, my path.
7:32And do you know what the most
7:33frustrating part is? I am currently
7:34watching how artificial intelligence
7:36and, let's say, those who do AI coding
7:38are simply depriving many professions
7:40of work. Ten years ago, we had a cool
7:42profession—programmers.
7:44Uh-huh.
7:45And now you literally ask GPT, "Create
7:48a program for me, for example, for an
7:51IT application or some game." And GPT
7:54creates the program in literally, I
7:56don't know, a few minutes.
7:58Well, you know, that's not quite right.
8:00I would say it like this: LLMs are
8:02indeed taking jobs away from juniors,
8:04junior programmers, well, entry-level,
8:07yes,
8:08and somewhere from mid-level, but for
8:10seniors, on the contrary, the demand
8:12for seniors has only grown. Why?
8:14Because, well, the most interesting
8:16thing in AI coding is that writing the
8:19code is the simplest task. When we make
8:23a multi-agent structure, we use the
8:25cheapest model for writing code. The
8:27absolute cheapest model. Because when
8:30you clearly define the technical
8:32requirements, set the parameters,
8:34determine what is allowed and what is
8:37not, and define the criteria for what
8:39constitutes a completed task, any
8:41reasonably smart model can write the
8:44code. But the real question is how to
8:47set a task, how to plan it, and how to
8:50design the application architecture.
8:53That’s where you need a human; you
8:56can't just slap that together. Sure,
9:02you can cobble together simple apps for
9:04yourself, but if you give them to a
9:07decent programmer later, they'll find
9:09so many security holes and patches,
9:12because how do you actually build an
9:14app with an LLM? You keep going, it
9:18creates a visual interface, you like
9:20something, and you say, "Add another
9:22button for me." And then, "I want that
9:24to flash when I click this button," and
9:26then you do this. And all of that is
9:28just patching. It’s all just patches.
9:32And then, when an IT specialist looks
9:34at your code, it's just patches upon
9:35patches, and nothing works. You can't
9:38put that into production. I mean, for a
9:42large number of users, or for the App
9:45Store, it'll be clunky; it’ll work
9:48for two days and crash on the third.
9:53That’s why I believe the demand for
9:59high-level specialists will only grow.
10:04Yes, writing code when the technical
10:07requirements are known, the parameters
10:10are set, and the constraints are clear
10:13—that’s not such a difficult thing.
10:17So, I do believe many professions will
10:21disappear, but primarily those
10:25involving routine operations. Routine
10:30tasks like sorting things, cleaning
10:33data, calibrating, deduplication,
10:35moving things around—basically what
10:38people did when shuffling papers,
10:41filling in numbers, calculations—
10:43these operations will be the first to
10:46go. But operations related to
10:50creativity...By creativity, I don't
10:53mean painting pictures or writing songs
10:56, but rather thinking about how to
10:59build the architecture. Because to
11:02solve any reasonably complex task, it's
11:04clear that one person or one agent
11:06won't be enough. It has to be a group
11:10of agents, and perhaps even a human
11:12within that system. And you need to
11:15establish how they interact. It's about
11:18how it's structured, the sequence, the
11:20protocols, permissions, access levels,
11:22security, and so on. This is what a
11:24human needs to think about. Therefore,
11:26human experience, taste, and the
11:29responsibility of decision-making—
11:31agents won't take that on themselves.
11:35But this requires different, new
11:37qualifications from people. I am
11:41absolutely thrilled that a time is
11:44coming now where value is shifting from
11:47routine, so to speak, operations with
11:49minimal intelligence to operations with
11:52maximum intelligence. And so a person
11:57will begin to engage in what I consider
11:59their most fundamental function—
12:01thinking. Thinking, while artificial
12:04intelligence will knock out everything
12:06else just like that. And those
12:09professions where you need to think and
12:11use your brain will not die out. On the
12:14contrary, they will only grow now, the
12:16demand will only increase. I even see
12:19now that when large companies are
12:22hiring, they’ve stopped looking so
12:24much at whether you have a degree from
12:27Stanford or Harvard; in fact, having
12:29dropped out of university is now
12:32considered cool. Then they started
12:35looking at how a person presents
12:37themselves, their presentations and
12:39resumes—people have learned how to
12:42make all that look nice. Now, they look
12:44more directly at your portfolio. Show
12:46me your GitHub, your link, what
12:47you’ve actually built there. They
12:48really look at your code, and it
12:50immediately determines your
12:51qualifications. And the next level is
12:54that they just give you a test task and
12:56say: "Build the architecture." Think
12:59through the architecture. And that is
13:01why the depth of your thinking becomes
13:02immediately clear. And in this regard,
13:06of course, I personally see a positive
13:09side in that those professions and
13:12activities where more intelligence is
13:14required will be valued more. And this
13:19is exactly what we are doing together
13:21here, learning to engage our brains
13:23more in business. Otherwise, we very
13:26often do business mechanically, without
13:28thinking. Margo, definitely, coding and
13:31artificial intelligence are the trends
13:34of the day, and they are spreading into
13:37every sector. One of your areas of
13:40activity is the agricultural sector.
13:42How do you think artificial
13:44intelligence can influence agribusiness
13:47, and what new professions might appear
13:49in agricultural fields and the sector
13:52as early as tomorrow? Well, in the
13:56agricultural sector, uh, firstly, this
14:01is in terms of improving varieties,
14:05breeds, selection, and so on. This, I
14:09believe, is the first thing that will
14:11be impacted.
14:13Because artificial intelligence is
14:15capable of sorting through various
14:16combinations of proteins, various
14:18combinations of genes, and so on and so
14:20forth. And thanks to this, people will
14:23learn how to develop new varieties
14:26faster, new breeds, and so on. The
14:29second is biosafety in the agricultural
14:31sector. Right now, we solve this with
14:34chemicals. Fields are sprayed with
14:36chemicals, animals are given various
14:38vaccinations, and so on. But it is
14:42possible to do it differently, or even
14:46to create plants that are not
14:50susceptible to certain diseases, for
14:53example.
14:56That is, figuratively speaking,
14:58ensuring the health of organisms not
15:01through chemicals.
15:03Uh-huh.
15:03And through some, well, you know,
15:06changes, selection, modification, and
15:09so on. And this, I believe, is the
15:12first part that gives the greatest
15:14multiplier effect on all sectors, on
15:17agriculture in general, and so on. If
15:20we take not the scientific part, but
15:23now, uh, directly the work in the
15:25countryside, well, let's say, in the
15:28field, on farms, and so on, then a
15:33strong technological shift will occur.
15:35I just see now that in our factories, a
15:37person doesn't even enter. Actually, a
15:40person entering a poultry house is,
15:43well, it's basically an emergency
15:46situation.
15:48Uh-huh.
15:49The point is, the less a person appears
15:51there at all, the higher the biosafety,
15:53because a person brings a bunch of
15:55contagion with them.
15:56Uh-huh. Accordingly, the technology is
16:00built in such a way that air exchange,
16:03temperature, climate, humidity, feeding
16:06, watering—everything is without
16:09human participation. And in the
16:12agricultural sector, I believe that
16:14systems with minimal human
16:16participation will be created. Even, uh
16:19, weeding, for example. Now there are
16:23already laser drones that fly along and
16:25just burn weeds, or wheeled ones that
16:27burn them out. Notice, this is again a
16:30non-chemical method of field treatment.
16:32That is, what I was talking about, we
16:35will move towards non-chemical ways of
16:37ensuring biosafety. Uh-huh.
16:39And, well, weeding with lasers, drones,
16:41and so on, with as little human
16:43participation as possible. This is even
16:46before talking about robots. And why is
16:48this artificial intelligence? Because,
16:50after all, all these mechanisms already
16:52exist, but these mechanisms will
16:54primarily start getting artificial
16:56intelligence as brains.
16:58So, these mechanisms will be given
17:00brains.
17:01And the next stage is, of course,
17:03robots. Robots are when artificial
17:06intelligence is given arms and legs,
17:09and it starts performing the work that
17:11humans did before. And in the first
17:16place, of course, such tasks will be
17:18handed over where precision is required
17:20,
17:21and where people don't want to work.
17:23These are dirty jobs, dangerous jobs,
17:27heavy or routine jobs. Well, things
17:31that people don't want to do. These
17:33will primarily be handed over to robots
17:35and artificial intelligence. And in
17:38this regard, I would say that, uh, of
17:44course, the farmer's function will not
17:47disappear, at least in the foreseeable
17:50future, but their workload will drop
17:53sharply, and their efficiency and
17:56productivity will increase many times
17:59over. Uh-huh.
18:01Because they will have many autonomous
18:03systems. It's interesting, I even saw
18:07it on Instagram, in Australia they
18:10launched, you know, a cage, a farmer
18:13launched a cage, basically, and the
18:16cage is on wheels.
18:19Uh-huh.
18:20At this height above the ground, and he
18:22herds sheep into it, like 50 sheep, and
18:25the cage has GPS antennas, and, well,
18:27something like artificial intelligence.
18:31And this cage, he has a pasture. And
18:33this cage quietly moves all day long,
18:36and the sheep graze. And there’s no
18:40shepherd, no one else there. The cage
18:42just drives around his territory. And
18:44on one hand, it works out well, it
18:46fertilizes evenly. The sheep graze
18:49evenly, the pasture doesn't degrade,
18:51wolves won't attack because they're in
18:53the cage. Well, in short, it's just a
18:58totally autonomous sheep-rearing system
19:01.
19:02It’s absolutely insane. And more and
19:03more of these systems will appear. Well
19:05, you see, the farmer still remains,
19:07but he becomes more of an operator.
19:09Uh-huh.
19:11Take us, for example, I'm even looking
19:12at poultry farming; our poultry farmers
19:14aren't the people who are there dealing
19:16with manure or whatever. They are
19:18purely operators. That's why, if animal
19:20scientists were valued before, now
19:22engineers are valued first and foremost
19:24.
19:24Uh-huh.
19:25In our management positions, it's
19:27mostly engineers now, not animal
19:29scientists. So you no longer need to be
19:32a specialist in animals, you need to be
19:35a specialist in systems. That’s the
19:38shift happening in the agricultural
19:40sector. You have to be a specialist not
19:43in animals, but a specialist in
19:45engineering systems.
19:46In support of this topic. We are
19:48currently in Arkansas. And a lot of
19:51rice is sown here. And right now, I'm
19:55with relatives here, I have many
19:56relatives, and I'm finding out that
19:58China is coming in here and they are
20:00sowing all these fields entirely with
20:02drones.
20:03It turns out that artificial
20:04intelligence, even current realities,
20:06is already entering agriculture.
20:09Yes. Yes.
20:10And I would like to ask this question.
20:13So it turns out that vibe-coding is, uh
20:16, the human role is to know a clear
20:18goal, right, and it turns out to know
20:21the exact final result and what problem
20:24we are solving through vibe-coding.
20:27Right? Yes.
20:29Well, it's not just that, look, the
20:31goal, the final result, you must know
20:32it precisely in metrics,
20:34not in words,
20:35but metrics, because words can be
20:37interpreted this way or that way. You
20:39can say, "Make it beautiful for me." Or
20:42you can say, "Make it this specific
20:45color number, these pixels, this
20:47sharpness, this image size." Well,
20:50that's different. It is important for
20:53the artificial intelligence to
20:55understand, that is, to learn to speak
20:57its language so that it understands
20:58these metrics. And that’s exactly
21:01this point. Oh, and the question always
21:02remains for the human: "What is this
21:04for?" Because with artificial
21:06intelligence, you can write any
21:07application. The most pointless,
21:09useless one. Only a human determines
21:12whether it is useful or not. So, in
21:14fact, the person bears the risk that
21:16the application turns out to be useless
21:18to anyone. You spend your time, your
21:22money, and in the end, the AI doesn't
21:24care because you already paid to use it
21:27. Your money is gone, and what you
21:31spent it on is your own area of
21:33responsibility.
21:35Margulan Kalievich, here is a question.
21:37Artificial intelligence is now coming
21:39into business; it is integrating very
21:41tightly there. But the question is,
21:43what will clients value in 5-7 years? I
21:45mean, when all this is integrated? So
21:47what do we, as entrepreneurs, need to
21:50work on right now?
21:53You know, you say: "Clients, but
21:55actually, can we pose the question more
21:58broadly: what will we value in
22:01artificial intelligence?" Or can we
22:04pose the question even more broadly?
22:06Most people make the mistake of
22:09thinking that artificial intelligence
22:12will take us over. And that, well,
22:15movies have scared us, saying they will
22:19take us over and so on. But do you know
22:22the funny thing is that artificial
22:24intelligence won't take us over; we
22:26will give everything to it voluntarily,
22:29with joy. Why? Because artificial
22:32intelligence and robots will do
22:34everything cheaper. Faster, easier, and
22:37with higher quality. And so you get a
22:40robot housekeeper; it does everything
22:43unquestioningly, everything clearly, no
22:45need to remind it of anything, and so
22:47on. What will you do? You will trust it
22:50with more and more things. Thus, the
22:52cooler the artificial intelligence is,
22:55the more access to your computer you
22:57will give it. You will give it access
23:00to your disk, Google Drive, iCloud,
23:03calendar, and email. You will give it
23:05yourself because it is cheaper, easier,
23:08better, and more efficient, and we will
23:10keep giving it access as long as
23:12artificial intelligence is more
23:14efficient than people. It will be a
23:16pleasure for us. We will gladly give
23:18artificial intelligence, in essence,
23:21our future, our destiny. And when we
23:25give all of this away, one day we will
23:28wake up in horror that we can no longer
23:31even imagine our lives without it. And
23:36now, if we return to your question
23:39about clients, clients will also always
23:41value what is faster, cheaper, better,
23:44and simpler. Jeff Bezos said it well.
23:48He says: "Many people forecast the
23:50future and ask me: 'On what forecast do
23:53you base your plans?'" He says: "I
23:56actually don't base my plans on
23:58forecasts." I build my plans on what
24:01will always remain unchanged. And what
24:03will always remain unchanged? The fact
24:05that people will always want things
24:07cheaper,
24:08faster, and better. Accordingly,
24:12Amazon’s entire strategy is to
24:13provide exactly that: cheaper, faster,
24:15and better. And customers, too—all
24:18customers—they don't care at all
24:21about the internal workings of a
24:23business. They just want this: I placed
24:26an order, I received the order, that's
24:28it. What we do, what our processes are,
24:31how many plants or factories we have,
24:34they don't really care about that. And,
24:37naturally, if artificial intelligence
24:39helps us deliver goods or services to
24:41customers faster, cheaper, better, and
24:44so on, that’s it—those companies
24:46will win. Consequently, the only
24:49companies that will win are those that
24:51transition the majority of their
24:53operations to artificial intelligence
24:54the fastest. Well, over time, it will
24:59become harder for us to do this,
25:00because artificial intelligence is now
25:02being integrated not only into business
25:03but also into education, and our
25:05children are using it too, which
25:06essentially takes away the capacity for
25:08logical thinking. And in the future,
25:11there will be fewer and fewer people
25:13every year who know how to think
25:15logically and, accordingly, manage this
25:17system.
25:17Well, that's not entirely true,
25:19actually. Look, every generation thinks
25:22that the children are getting dumber.
25:25Every single generation. Our parents
25:28thought so, and our parents 'parents
25:30thought so. Why? Because, yes, for our
25:33parents' tasks, we are dumb, but for
25:35our tasks, our parents are dumb. It
25:39will be the same with our children. We
25:41talk about logic and all that, saying
25:43our children don't understand it, that
25:44it's atrophying, and so on. But they
25:48understand things that we don't even
25:50grasp, like, what even is that about? I
25:54mean, I just look at many young guys,
25:57they understand things, and for them,
26:00it's natural. I mean, they didn't study
26:06it like I do—I'm studying "vibecoding
26:08," literally studying it, taking topics
26:11, learning the concepts, the history of
26:14the term, what it's for, and how it
26:16helps. It’s like breathing to them.
26:21And you see, many things—if we take
26:24intelligence and break it down into
26:26parameters and criteria—we have our
26:29own criteria that show us a person is
26:32very smart.
26:33Uh-huh.
26:34But these parameters and criteria don't
26:36apply to these kids because they are
26:38preparing for a different future. We
26:41prepared for a different future. We, as
26:42parents, are always like generals
26:44preparing for the last war. And we do
26:47the same thing. We are preparing our
26:51children for our
26:54for our own past. We keep saying
26:56problem
26:57Man, I don't know English, I need to
26:58get my kid into English lessons. Damn,
27:00China is booming right now, need to get
27:02them into Chinese. Crap, need to get
27:04them into coding, they need to know
27:06math. Why? Because we’re trying to
27:09make up for all the things we missed
27:11out on. So, we pump the kid full of
27:13stuff, but the kid has a different
27:14future. Maybe what they really need is
27:18drawing, dance, or art. None of it
27:21matters—not math, not languages—
27:23because now with an LLM, you just put
27:26on a little earpiece, and it translates
27:28Chinese instantly. Why study Chinese if
27:31you can just wear a tiny earpiece and
27:33talk to any Chinese person? So, our
27:36requirements for knowing Chinese,
27:38English, math, or logic just don’t
27:40work,
27:41they aren't relevant. We teach what we
27:44learned; I studied logic myself in law
27:48school. We studied linear logic, but
27:52the current world is moving toward
27:55non-linear logic. And that’s
27:57something completely different. Linear
27:59and non-linear logic are contradictory
28:02concepts. I was lucky back in the day.
28:07I had a math teacher who graduated from
28:10some Moscow math institute. He was a
28:14heavy drinker, but very talented. All
28:17the kids loved him, parents loved him,
28:19and so on. And he taught us math. Not
28:22just math, but everything beyond the
28:25scope of math too. That’s how he
28:27broadened our horizons. I was blown
28:30away when he talked about non-Euclidean
28:32geometry, or Lobachevsky’s geometry,
28:34so to speak. We know in regular
28:36geometry that two parallel lines never
28:39intersect. But in Lobachevsky or
28:41non-Euclidean geometry, they do
28:42intersect, and twice at that. Two
28:46parallel lines. Also, we know that in
28:50any square all sides are equal, but in
28:53that geometry, they aren't. We know
28:58that in our regular linear geometry the
29:00shortest distance between two points is
29:02a straight line, but in non-Euclidean
29:05geometry, the shortest distance is a
29:07curve. And how does that work? Take a
29:10globe and look at how planes fly.
29:14Here’s your city, here’s your
29:15destination. Why not just fly straight?
29:17Why are you flying like this, across
29:19the Arctic? You see planes flying in a
29:22strange way.
29:23In aviation, this is called an
29:25orthodrome. Orthodromic, when you map a
29:29route on a sphere, you always fly a
29:31curve, because you can't fly straight,
29:33because the shortest path is...
29:36Now we’re going to end up in Flat
29:38Earth theory. Now we’re going to veer
29:40off course.
29:41My point is that the linear logic we
29:44live by doesn't work in this future, or
29:47it works differently than we think.
29:51Therefore, our way of assessing our
29:52children's intelligence using our own
29:54parameters will always be flawed. So,
29:59as usual, we have to grumble, complain
30:01about the kids, and say that a stupid
30:04generation is growing up. Where is the
30:07world heading?
30:07Where is the world rolling to? How will
30:09they feed us, how will they serve us a
30:11glass of tea? Yes. Yeah, yeah. Where is
30:14the world heading with such an
30:16education? We're all doomed, and so on.
30:18Well, it's not that bad. Actually, look
30:22, in our time, our parents emphasized
30:25things like memorizing the
30:26multiplication table, for example. It
30:29was considered very cool and very
30:31important. But if you tell that to
30:34someone now, it’s just ridiculous.
30:37Why memorize the multiplication table
30:40if you have a calculator at hand, and
30:42so on. Well, there is absolutely no
30:44need for that. There is no need to know
30:46languages. There is no need to know how
30:48to use a calculator. There is no need
30:50to know a whole lot of other things.
30:52Margulan Kalievich, isn't that the very
30:55moment that forces the brain to
30:57actually move? I mean, to get it
30:59working at least somehow? I mean, some
31:03other knowledge should replace these,
31:05but when you look around now, people
31:08usually have a phone in their hands,
31:10not a book. Exactly, so there...
31:13Yes. Well look, the point is that the
31:15danger of our current time actually
31:18lies in the fact that there will be a
31:20sharp stratification of people based on
31:23intelligence,
31:24not based on wealth, not based on money
31:26, but on intelligence. People will be
31:29divided into two camps. One camp of
31:31thinking people and the second camp of
31:33non-thinking people, and unfortunately,
31:35the non-thinkers are the majority. And
31:37the gap between the thinkers and
31:39non-thinkers will be catastrophic. It
31:42is increasing. Why? Because artificial
31:45intelligence is like a magnifying glass
31:46. It makes the stupid even stupider,
31:50and the smart even smarter.
31:52Even smarter,
31:53because it has algorithms built into it
31:55. The stupider the questions you ask,
31:57the stupider the answers it starts to
31:58give you. It starts answering more
32:00simply because it has a mechanism for
32:03adapting to the user.
32:04Accordingly, it starts explaining
32:06everything to you even more simply, and
32:08in this way, you become even stupider.
32:11Because you have no need to ask other
32:13questions. It gives you ready-made,
32:15templated answers. But when a person is
32:18smart, they start asking critical
32:19questions, deeper questions, questions
32:21on a different level, and so on. The
32:23intelligence starts to adapt and answer
32:25more intelligently, and it begins to
32:27build the conversation with you in a
32:28completely different way. And thus,
32:32even at the level of a simple ChatGPT,
32:34I'm not even talking about coding,
32:36vibe-coding, agents, and so on, even at
32:39this level, this stratification occurs.
32:43And as I said, I believe the era of
32:45intelligence is dawning. So, what does
32:49artificial intelligence actually
32:51provide? It poses a challenge to
32:54natural intelligence. And the task of
32:58who will control the future depends
33:01solely on who turns their brain on
33:03faster. Well, meaning artificial
33:08intelligence will outpace us in all
33:10routine operations, in all mass
33:12operations where speed, accuracy, data
33:14volume, data processing, and so on are
33:16needed. But I think artificial
33:20intelligence will not be able to build
33:22non-linear connections and dependencies
33:25for a very long time. Because, let's
33:28say, you can say: "Here we are sitting
33:30at Lake Balkhash, and here is a Russian
33:32olive tree, yes, the scientific name is
33:35Elaeagnus commutata, the silverberry...
33:37""...you softened it, Margulan
33:40Kalievich, I said it softer, but the
33:42scientific name is silverberry." Yes.
33:45And we can ask, what do this person
33:49here and a person who went to a casino
33:52and blew a few thousand dollars have in
33:56common? Yes, will artificial
33:58intelligence understand the difference?
34:00It won't. It will say: "What do the
34:02tree, Balkhash, and a dude who blew
34:03money in a casino have to do with each
34:05other?" But we humans immediately
34:09understand: the silverberry tree and a
34:11"sucker"—the golden boy who blew his
34:13money—are essentially the same thing.
34:18Well, we, notice, make the connection
34:20......well, the tree is not to blame
34:21for anything in this...
34:22Yes, we aren't insulting the tree, but
34:24for us, these are identical concepts,
34:27and artificial intelligence cannot link
34:29that yet. It's too non-linear. This is
34:32from botany, and that is from folklore,
34:35from slang. You know, we humans are
34:38exactly the ones who can bridge
34:41different layers of data or knowledge;
34:43we can build non-linear dependencies
34:46and connections, and non-linear means,
34:49according to chaos theory, they cannot
34:51be algorithmicized, and artificial
34:54intelligence is algorithms, you
34:56understand? And they don't lend
34:59themselves to algorithmicization. This
35:01is at the level of what's called
35:03serendipity in English, or at the level
35:06of a gut feeling, intuition. And there
35:09is a concept in Chinese philosophy
35:12called "knowledge without words," when
35:15a person knows something but cannot
35:18explain it. Uh-huh.
35:21And look, we can only teach artificial
35:24intelligence what we can explain. We
35:27can't explain or teach it something
35:29that we can't explain ourselves. But
35:32the problem is, or rather, our
35:34advantage is, that we know many things
35:37that we cannot explain. By the way,
35:41wisdom belongs to this category. Wisdom
35:44is also something that cannot be
35:46algorithmicized. Accordingly, we
35:49shouldn't expect wisdom from artificial
35:51intelligence either. And wisdom is
35:53often more effective than science, than
35:55knowledge.
35:56Than knowledge,
35:58yes?
35:59More like experience, probably.
36:01Experience is a lesson learned from a
36:05mistake made.
36:07And in that sense, you know, artificial
36:09intelligence learns from its mistakes
36:12better. Actually, when it comes to
36:14gathering experience, AI is cooler than
36:16humans. But the question is, it
36:20extracts knowledge or experience from
36:22its mistakes in the form of algorithms.
36:25And what I’m talking about is exactly
36:28nonlinear connections. We can extract
36:31not just algorithms from experience,
36:33but nonlinear connections and sequences
36:36using nonlinear logic we haven't even
36:39studied, yet we have an intuitive sense
36:42for.
36:43Uh-huh. Let’s say, excuse me, a
36:46simple example. Did anyone teach us
36:50justice?
36:51No. Right? Is there some algorithm for
36:54justice? There isn't. But every one of
36:57us knows deep down what justice is. And
36:59it's hard to explain,
37:01because we are born kind. We are
37:03initially
37:04like
37:04we are essentially born good and kind.
37:08No, the question isn't justice—it’s
37:10not just about kindness, and very often
37:13justice is cruel. Well, that’s
37:16different. Virtue is something else.
37:18I’m talking about justice. Where does
37:20it come from in us, for example? Maybe
37:23we have it inside
37:23because conscience gnaws at us, right,
37:24or something like that?
37:25Ah,
37:25it’s because conscience bothers a
37:27person. Because of that,
37:28that concept of conscience.
37:29Parents told us about it in childhood.
37:31It’s all from childhood. Yeah.
37:32Or not. Look, there are people who grew
37:35up as orphans, but that doesn't mean
37:37they lack a sense of justice.
37:40It’s a collective consciousness we're
37:42born with, you know? How did that
37:45happen? Evolutionarily, we grew up, and
37:47all the unfair individuals didn't
37:49survive because they were cast out of
37:51the tribe and labeled as maladaptive.
37:54Uh-huh.
37:55I mean, if you just went and walloped
37:57Andrey over the head right now. We’d
38:00say, "Yaroslav is maladaptive," and
38:02we’d stop talking to you because
38:04you're irrational. But why irrational?
38:06What is an irrational person? An
38:07irrational person is someone
38:08unpredictable. And if they’re
38:10unpredictable, you can't deal with them
38:12because you don't know their next move.
38:15Consequently, such individuals were
38:17isolated from society and perished
38:18because they couldn't feed themselves,
38:20couldn't survive.
38:21And naturally, such people didn't pass
38:24on their genes.
38:25Uh-huh. And historically, this sense of
38:30justice, the sense of mutual exchange,
38:33conscience, and so on, is in our genes.
38:37It's not at the intellectual level,
38:39it's at the level of motor skills and
38:41genetics. We feel, for example, if
38:45we've done something wrong—we don't
38:47even know what we did wrong yet, but we
38:50feel uncomfortable, like we messed up.
38:54We feel it, we feel it,
38:56but we can't even grasp it. And we
38:57can't define the parameters. What
38:59parameters show that I messed up? But I
39:01feel like I’ve done something that
39:03wasn’t very good. These things cannot
39:07be algorithmicized, and we won’t be
39:09able to teach artificial intelligence
39:10to do them.
39:13Marlansha, may I ask about the future
39:15of business in general? How to act, how
39:18to scale, based on what’s happening
39:20in geopolitics now, how to minimize
39:22risks in general, how to act correctly,
39:24where to direct your focus and
39:26attention.
39:28Well, again, from the point of view of
39:30risk assessment, only a person can
39:32assess risks, because risks are always
39:35assessed relative to your goals, and
39:37artificial intelligence doesn’t know
39:39about your goals, about your—look,
39:41it’s not just goals. Sure, you can
39:45write goals for it, but there are goals
39:47involved, your preferences are involved
39:50, your fears, complexes, your ego, and
39:52so on are involved. And this totality
39:55—you calculate risks in aggregate.
39:59How much will this action hit my ego,
40:02how cool or how bad will I look, how
40:04will this affect my income, will it hit
40:07my income or not? Furthermore, how
40:10pleasant or unpleasant will it be for
40:12me? We calculate a bunch of things
40:15there. Artificial intelligence won’t
40:16be able to calculate that. Therefore,
40:18risk management is, after all, a
40:20function that will remain with humans,
40:22just like goal setting and accepting
40:24the result, because only you can accept
40:26the quality of the result. And
40:29artificial intelligence can give you
40:31formal parameters, but it’s not a
40:33fact that it will satisfy you, because
40:35besides formal metrics, parameters, and
40:38so on, there is a bunch of qualitative
40:40properties and characteristics that you
40:42need. And I believe that one trend is
40:47that business will move in a direction
40:50where it is necessary to use more
40:52intelligence in business. And,
40:56accordingly, business models will
40:58change, the way of building a business
41:01will change in general. And ultimately,
41:04what is a business? And ultimately,
41:06business is the delivery, the creation
41:09and delivery of value to other people.
41:13And as long as people have problems, or
41:16as long as you can say, as long as a
41:18person, any person, has needs, and when
41:21a person has needs—that is,
41:23unsatisfied needs. As long as people
41:27have unsatisfied needs, then the
41:29creation and delivery of the means that
41:31satisfy those needs will be the essence
41:33of business. But since artificial
41:37intelligence and robots are appearing,
41:40and they will perform an increasing
41:42number of functions faster, easier,
41:44cheaper, and with higher quality, then
41:47to satisfy the following needs—basic
41:49needs, like eating, drinking, sleeping
41:52well, having a good time—those will
41:54be satisfied by robots and artificial
41:56intelligence. And that is when new
41:59human needs will arise. And these needs
42:04can be absolutely unconventional.
42:08Unconventional. Well, relatively
42:11speaking, the first seeds are virtual
42:14worlds, where I was surprised that one
42:17city I studied, a virtual city, they
42:20started selling plots of land inside
42:23the city. Uh, and one guy built a
42:26three-story office there. And to host a
42:30party on the third floor, an open
42:33terrace, on the third floor of this
42:36virtual office in a virtual city,
42:39Coca-Cola paid 400 grand.
42:45Coca-Cola paid 400 grand to host a
42:47party on the third floor of an office,
42:50a virtual office in a virtual world.
42:55Welcome to a new reality. Why? Because
42:58there were several hundred thousand
43:01virtual participants who came there.
43:04And what's the difference, he's sitting
43:06at a computer. Even if he's in a
43:08virtual world, at a virtual party, the
43:11name Coca-Cola still enters his eyes,
43:13his brain. And then he goes to a real
43:15store and buys a real Coca-Cola. And
43:18Coca-Cola recoups its 400 grand.
43:21So, what is this? This isn't a
43:24traditional need, it's not a basic need
43:27. And it's not—you see, when we cover
43:31basic needs, we will want to play some
43:33games, complex games. We will want, we
43:38will have some complex needs, and to
43:39satisfy them, that's where you need
43:41intelligence, intelligence, and more
43:43intelligence. I also believe that if we
43:50take society as a whole, the way
43:52society is organized will change. For
43:55example, I believe that the state will
43:57become a thing of the past,
43:58corporations will become a thing of the
44:00past. Not quickly, but on a 30-year
44:03horizon for sure. Why? Because I can
44:08already see now that the state itself
44:10is handing over most of its functions
44:12to artificial intelligence, to robots.
44:15Well, take Kazakhstan, for example. In
44:17the past, we were all plagued by
44:19everyday corruption. Traffic police, or
44:25district justice departments, every
44:27document, money, queues, fixers, and so
44:29on. Now, that's gone.
44:31Now, everything is fast via an app.
44:33Now, everything. You can re-register a
44:36car in an hour on your phone, and you
44:38get a passport, they even deliver it to
44:40your home in an hour.
44:43Yes, banking payments, everything is
44:44already done. Why is that? Because it
44:47is in the interest of the politicians
44:49at the top to kill everyday corruption
44:51first and foremost, because it causes
44:53public discontent, because we don't see
44:55their big decisions, but we encounter
44:57the small ones in daily life. And in
45:00this way, more and more functions of
45:02the state will transition to artificial
45:03intelligence. For example, let's take
45:05the courts. Well, listen, civil courts
45:08especially are very easily
45:10parameterized and algorithmized. And
45:13artificial intelligence is already in
45:14the US, in trial court hearings, I
45:16think, with a score of 97%more accurate
45:20decisions than regular judges. Ordinary
45:23judges have around 70%accuracy, while
45:25artificial intelligence has 97%accuracy
45:27in judicial rulings. And what will
45:30happen? First, simple situations will
45:32be handed over, where there are clear,
45:34predefined parameters and so on. And
45:36then, increasingly complex ones. Why?
45:38Because it’s not in the politicians '
45:40own interest to have many complaints
45:42from the population. And what does the
45:44population complain about? About
45:46judicial bribery, wrongful decisions,
45:49unfair sentences, and so on. And thus,
45:52what will result? It will turn out that
45:54politicians themselves will be
45:56interested in handing over even the
45:57judicial process to artificial
45:59intelligence. And if we look further,
46:02I’m seeing many towns appearing
46:04around the world where advanced IT
46:07specialists live, and each of them has
46:09four or five passports in their pocket.
46:13Plus cryptocurrency, and they simply
46:15want to be citizens of the world. They
46:17don't want to belong to any specific
46:19state. And this is a general trend. It
46:23turns out that those who use their
46:25brains no longer want to live under the
46:27constraints called the state, the tax
46:29system, and so on. They want to live,
46:33first of all, across the whole world,
46:36to nomadize freely, while not being a
46:39subject of any specific state, but
46:42moving around calmly. And the most
46:44interesting thing, the most interesting
46:46thing is that states themselves want
46:48this. Now all states are competing to
46:51get intellect to come to their country.
46:54Moreover, Dubai is already making
46:56policies such that you don't have to
46:58live in the country. It used to be that
47:00Germany announced a recruitment of
47:0250,000 IT specialists, and for a time
47:04they started issuing passports so they
47:06would move to Germany and live there.
47:08But now a new wave has begun. States
47:11are already saying: "You don't even
47:12have to live in our country." Just get
47:15citizenship, pay taxes there, and so on
47:19, or work.
47:21Because when highly paid IT workers
47:23live in a country, they consume a lot
47:26of resources and pay money as well. But
47:30I believe that the state will change,
47:32society will change, and business will
47:34change. And in this sense, banal
47:38business products will be made by
47:40either robots or artificial
47:42intelligence. And we will need to find
47:46new, higher-level needs and satisfy
47:49them. And for this, we have to turn our
47:52brains on again,
47:54because artificial intelligence is
47:55pushing us from behind. If we were
47:57previously doing those dull operations,
47:59it now takes them over and says: "No,
48:01no, no, now I will do the dull
48:03operations." And along with these dull
48:05operations, every day it takes over
48:06even smarter operations. And what is
48:07left for us? Either fully say: "That's
48:10it, I give up, I won't think anymore."
48:12Or start thinking about what it cannot
48:14think about.
48:16So, it turns out unemployment will rise
48:17significantly, right, year after year?
48:22I believe unemployment will rise, and
48:24quite significantly at that. Right now,
48:27it's completely unnoticeable. But the
48:29evolution of any technology follows a
48:32path like this. Not noticeable, not
48:34noticeable. No, no, no, no, no, no, no.
48:36Suddenly, it's too late. It moves
48:39exponentially. At some point, it will
48:41enter our lives so quickly that it
48:44seems like it wasn't there, and then
48:46suddenly, it's everywhere.
48:49Margonkevich, let me ask you a question
48:51. I'm nearly 40, and I've decided to
48:54become a startup founder. Yeah. I’m
48:58looking at how startups were brought to
49:01Russia before, like VKontakte, which
49:05copied Facebook, right, or Ozon, Beru
49:08—that’s just Amazon, which came
49:11here. Yeah. And my question is this. I
49:17have always sold metal. I was building
49:20there,
49:22yes, heavy metal, meaning I moved
49:23physical goods, did things where the
49:25need was clear, right, there's an order
49:27from a company, we fulfill it. Yeah.
49:31How do I transition to...well, I do
49:34program, I do a lot of web coding, I
49:37write a lot of things, but everything I
49:41create is like a car, you know, like a
49:44drift car, right? When you get into it,
49:49there's this stick for shifting, the
49:50steering wheel is some weird thing,
49:52just one seat and metal all around,
49:54right? So, I’m the only one who knows
49:56how to drive it. If anyone else gets in
49:58, that's it, it's not a fit for them.
50:01Yeah. So how do I shift from this
50:04standard mindset, where you see a clear
50:07need, to a startup mindset, to start
50:10making money from ideas?
50:12Well, you’re an IT guy, right?
50:14I’m not an IT guy, but I’ve always
50:17had a connection to marketing,
50:19development, and where did the IT come
50:21from?
50:22I used to write websites. I always had
50:25this approach: when a new technology
50:28arrives, for example, when the first
50:31websites appeared, I sat down and
50:33learned HTML myself 20 years ago, wrote
50:36a site, put it on the internet, and I
50:38got a large number of clients.
50:41Uh-huh.
50:42Yeah. And every time something new
50:45appears, I always took it and mastered
50:47it at some intuitive level of my own.
50:50Well,
50:51look. How do you change your mindset
50:56and psychology? I’ll start with the
50:58basics, okay? The foundation is always
51:02what is actually happening in the
51:04modern world. The first indisputable
51:06thing, you could call it an axiom, is
51:09thinking from first principles. The
51:11first thing that is indisputably
51:12happening is that the world is
51:13accelerating. Agreed? All processes are
51:15moving very fast. A week doesn't even
51:17go by without an update anymore. I have
51:19my Hermes agent with 157 new commits.
51:22I'm saying, it's only been a week, how
51:24do you manage to fix things and build
51:27new features? So, the world is
51:29accelerating, all processes are
51:30accelerating. Agreed? Okay, second. The
51:35world is becoming more complex, because
51:37if you take the same complexity but
51:39perform it at high speed, it
51:40automatically becomes more complex. For
51:43instance, when you're driving at 30 km/
51:46h, you can look at the trees, leaves,
51:48birds, the sky, the clouds. You're
51:51looking at those same clouds, those
51:53same birds, those same trees. Now go at
51:55100. That's it, it's hard to notice the
51:58birds, hard to notice the leaves, and
52:00you'll be lucky if you even notice the
52:02clouds. And if you go 150, you won't
52:04care about the clouds at all. The road
52:05will consume all your attention. Agreed
52:08?
52:08Uh-huh. And this is what's happening;
52:11in this way, the world is accelerating.
52:14And because it's accelerating, it's
52:15becoming more complex. But the problem
52:18is that our complexity is increasing
52:19not just because of speed, but also
52:21because of the actual complication of
52:23processes. Right. Right. Moving on. If
52:28any process accelerates and becomes
52:30more complex, it increases the number
52:32of risks. Agreed? It's one thing to
52:37drive at 20-30 km/h, and another at
52:38150-200. With the same reaction time
52:43and the same abilities, your capacity
52:45to avoid risks drops significantly.
52:49Right? So in this way, the risks only
52:51continue to grow. Right. And what do we
52:55do in such cases? Intuitively. This is
52:59precisely what distinguishes humans. We
53:01immediately slow down. Imagine you're
53:04speeding along and suddenly hit fog at
53:06night. Before, the road was clear and
53:08bright, then fog. What do you do? You
53:10sharply reduce your speed. You switch
53:14to low beams and basically drive by
53:18feel. Agreed? It's the same thing when
53:21you arrive and walk into a swamp. In a
53:24swamp, you pick up a stick and start
53:26probing. Because just because there was
53:29a path yesterday, it doesn't mean it's
53:31there today, because the swamp is
53:33moving. Today, the path might be two
53:36meters to the left, but you can only
53:37find it by feeling around. And now, if
53:40we take this and apply it to business,
53:42how do we do it? The question isn't
53:44about changing mindsets or technology,
53:46or how to invent a new path or a new
53:48trail. No, no. The question is about
53:50changing the entire understanding of
53:52processes. For that, you start moving.
53:55I created a system called the "slug
53:56strategy." The slug strategy is based
54:00on Japanese scientists who studied
54:02slime mold. Slime mold is a
54:04single-celled organism, a plant-like
54:06animal. It's, you know, sort of like a
54:08slime mold. They put it in a maze. But
54:11on the other side of the maze, they
54:14placed some bait. And what did it start
54:16doing? It started extending its
54:18tentacles into the maze's paths. And
54:21where there was a dead end, it stopped
54:23sending nutrients there, and its
54:25tentacles retracted. But where there
54:28was no dead end, nutrients were sent,
54:31and that branch kept growing. As a
54:33result, the slime mold unfailingly
54:35found the exit to any maze, despite
54:38having no brain. It’s a single-celled
54:41organism, it has no brain. Then they,
54:44uh, replicated this famous experiment
54:46on a map of Tokyo. And the slime mold
54:50drew the most optimal logistics routes.
54:54And when they laid it over a real map
54:56of Tokyo, it turned out they didn't
54:57even need engineers to build railways
54:59and roads. They could have just used a
55:02map of the city from the start. Let the
55:04slime mold trace it, and it would
55:05create the optimal route. Just imagine,
55:09Harvard engineers worked on it, yet
55:11here's a brainless creature. The result
55:13is the same. So, when we face such
55:18uncertainty, the only thing we can say
55:23for sure is that no one knows the
55:28future for certain. Right? And
55:32accordingly, in such a situation, you
55:35have two options. Either you make the
55:38decision yourself, but then you take on
55:41all the risks, or you make the decision
55:43based on data. Yes, but how do you make
55:47a decision based on data if there is no
55:49data? And that's why I called it the "
55:53slime mold strategy," where you
55:55formulate many hypotheses, and your
55:58goal is to make many cheap, quickly
56:00testable hypotheses. Then you test 10
56:04hypotheses; eight don't work, two do.
56:07What do you do? At the end, you use a
56:09traffic light system: red, yellow,
56:11green. If a green hypothesis succeeds,
56:14you pour resources, money, and
56:15attention into it and follow the
56:17principle: strengthen what works. The
56:20hypothesis that didn't work, you shut
56:22it down, and you keep moving forward
56:25like that. Therefore, if you want to
56:28create an app that succeeds in current
56:32conditions, don't try to create a
56:35winning app. You need to create a
56:38pipeline for testing hypotheses, and
56:41then, without overthinking it, simply
56:43strengthen what works. Because what is
56:46a hypothesis? It's a tentacle or a
56:49sensor for gathering data from the
56:51world. The world itself will tell you
56:54what it needs. You take a hypothesis
56:57and say, "If I release this app, I
56:59believe that within 30 days, with this
57:02ad budget, this many people will click,
57:04this many will visit, and this many
57:06will sign up." That's it; you allocate
57:11money, create a landing page, launch it
57:13, and watch the metrics. Yes, are
57:16people interested? Did it hook them?
57:18Did it work, or did it fail? You say, "
57:20I believe that if I change the headline
57:22, the hypothesis will take off." You
57:24change the headline, run it, it didn't
57:26work. Then you say, if I change the
57:28color, it will work. You change the
57:30color, it didn't work. And that's how
57:32you test.
57:33Understood. Thank you. And here is what
57:35the crux is.
57:36You know what's good about this?
57:38It is definitely a foolproof way to
57:40move forward. But the problem is that
57:44this often contradicts our ego. And
57:50unfortunately, the main obstacle here
57:52will be our ego, because we will have
57:53to admit that we are at the level of a
57:59single-celled slug.
58:01Making a decision,
58:02right? Or essentially, in another way,
58:05if we speak metaphorically, right? We
58:08just all need to learn how to talk to
58:10this world. Learn to talk with this
58:12world. I am a religious person. I say,
58:14we need to learn to hear what the
58:17Almighty wants to convey to us. After
58:19all, feedback is from Him. He says: "
58:22Don't waste time on this nonsense.
58:23Nobody needs this. But this thing here
58:25will take off for you. Do this."
58:28I have another question. If you look at
58:32the development of messengers and some
58:34other systems, for example, messengers:
58:36there was ICQ, then WhatsApp appeared,
58:39then Telegram appeared. Telegram is
58:42really good and cool. And then you see
58:44things like WeChat, which is completely
58:46state-controlled, and so on. And it all
58:49eventually leads to the state, right.
58:51And so, it leads to an element of
58:53controlling people, right. And what
58:57about artificial intelligence, in your
59:00opinion, when will it turn into some
59:02kind of state machine? It seems to me
59:06that it will happen eventually. Do you
59:10think the state will use artificial
59:11intelligence to control people and, in
59:13principle, control artificial
59:15intelligence itself?
59:17Well, look, it already controls it, and
59:19the state is already fully involved in
59:20artificial intelligence. Why? Here's a
59:22simple example. Just last week,
59:24Anthropic released the coolest model,
59:27Claude 3.5. I immediately went crazy
59:30and integrated it into a multi-agent
59:31system. I managed to write so much code
59:35in 3 days. I thought, while they said
59:38that on the 22nd we will introduce API
59:40payments for usage, here you can use it
59:43unlimitedly via subscription. And then,
59:47right as we arrived here, the US
59:50government imposed an official ban on
59:53two models, Claude 3.5 and Haiku, for
59:57use by non-US citizens. And the
1:00:03question arises: is this a separation
1:00:05that has already started?
1:00:06Uh-huh.
1:00:08So, it turns out it's segregation based
1:00:10on nationality. That's it, only US
1:00:12citizens are allowed. We used to think
1:00:14that anyone who could pay could have it
1:00:16, but now it turns out there are US
1:00:18citizens who are all equal, but some
1:00:20are more equal than others. And what
1:00:23should we do with this? I believe that,
1:00:25in reality, they have opened Pandora's
1:00:27box. They have now created a motivation
1:00:30for every state to develop its own
1:00:32artificial intelligence. They created a
1:00:35motivation to switch to Chinese models
1:00:38and a motivation to create local,
1:00:40open-source, free models. Now I, for
1:00:43example, have also made a decision for
1:00:44myself. I have decided that I will use,
1:00:47and my architecture will be built, in
1:00:49this way. There will be a local model
1:00:51that runs on my computer, handles only
1:00:53my sensitive data, and so on, without
1:00:55going outside.
1:00:56Uh-huh. There will be an open-source
1:00:59model that is free and performs all
1:01:01routine operations. And there will be a
1:01:04very smart model at the very top, and
1:01:06it will be very expensive. That's all
1:01:08three. And at the same time, I am
1:01:10building things so that my processes do
1:01:12not depend on computers, phones, and so
1:01:15on, plus they don't depend on any
1:01:17specific country's model. I am already
1:01:20building such an architecture. And this
1:01:22way, you know, a competition between
1:01:24armor and projectile is taking place.
1:01:27The state wants to control, and when it
1:01:29tries to control, the entire market is
1:01:31horrified, and all smart people start
1:01:33creating systems that do not depend on
1:01:35the state.
1:01:37Uh-huh.
1:01:38And thanks to this, I believe that now
1:01:40more open-source models will appear,
1:01:42more local models that can be installed
1:01:43on a computer, and so on. And more
1:01:46people will switch to Chinese models,
1:01:48thanks to which the Chinese will
1:01:50release even more new models. And plus
1:01:53Europe, I think, or some other
1:01:55countries, will now focus on creating
1:01:57their own local LLMs. You know, it's
1:02:02like any restriction; instead of
1:02:04limiting, it provokes a counter-action,
1:02:07any action provokes
1:02:09an equally directed and equally strong
1:02:11reaction. The Americans are going to
1:02:13run into this now. And what is
1:02:17happening? There is always a struggle.
1:02:20The state always wants to control
1:02:22people, but there is always a certain
1:02:25cohort of smart people who do not want
1:02:27to be under state control. And so the
1:02:30state introduces a restriction, and
1:02:32this group of people breaks free from
1:02:34the state's control. The state keeps
1:02:36trying, it's like a quiz, you know,
1:02:38these smart people are always breaking
1:02:40free from the state's influence, and
1:02:42the state tries to control them. But
1:02:45this is neither good nor bad. I believe
1:02:49the state should control the majority
1:02:52of the population because most people,
1:02:55unfortunately, do not live a conscious
1:02:58life. Accordingly, it is better to
1:03:01manage, control, guide them, and so on,
1:03:04and so on. And smart people must prove
1:03:06through their intelligence that they
1:03:09can live outside the framework of the
1:03:11state. And this is constantly, you know
1:03:14, like the pike and the crucian carp.
1:03:18Smart people are like crucian carp, and
1:03:19the state is like a pike. It constantly
1:03:21says, "If you want to remain smart,
1:03:23keep moving, be smarter than me, if you
1:03:25want to remain free and independent."
1:03:27So, freedom and independence, you have
1:03:29to earn them. And you have to earn them
1:03:32through intellect. And that is why I
1:03:34believe, yes, wherever the state can
1:03:37reach, it will always try to control.
1:03:40The state has such a role: to control,
1:03:43take away, divide, suppress, and so on.
1:03:48Well, it's like cat and mouse all the
1:03:49time. But as I said, well, for the
1:03:52majority it's good, otherwise there
1:03:54would be chaos. Especially countries
1:03:56like China, for example. Well, listen,
1:03:58there are almost 2 billion people there
1:03:59. Not all 2 billion are conscious,
1:04:02right?
1:04:02Yeah. For the most part, strict rules
1:04:05are needed there. Like the death
1:04:08penalty for corruption, clear adherence
1:04:10to regulations, principles, and so on
1:04:12and so forth. If you make 2 billion
1:04:15free people, they'll create anarchy
1:04:17there.
1:04:18Don't you think there will be even more
1:04:20chaos because of this?
1:04:21I mean, look, state artificial
1:04:23intelligence controls private
1:04:25artificial intelligence. Private
1:04:28companies, to escape the control of the
1:04:31state AI, will develop their own AI to
1:04:33the maximum.
1:04:35And it will just start an AI race and a
1:04:38green light for artificial intelligence
1:04:41. No, well,
1:04:42this could all reach an uncontrollable
1:04:45level.
1:04:46Well, look, ultimately, if we take my
1:04:48concept of superposition, then
1:04:50artificial intelligence is in a state
1:04:53of superposition, because all countries
1:04:55are investing money in developing
1:04:58capacity for artificial intelligence.
1:05:00All countries are interested in the
1:05:02development of artificial intelligence.
1:05:04We give it the very best, all our
1:05:06knowledge, our best data, and so on and
1:05:08so forth. Artificial intelligence
1:05:11itself doesn't fight anyone. It just
1:05:13receives the gifts that people bring it
1:05:16for free and even compete in bringing
1:05:19those gifts. Thus, ah, yes, maybe we
1:05:24can come to such a situation very
1:05:26quickly, where artificial intelligence
1:05:30indeed surpasses people in intelligence
1:05:33by most parameters. by most parameters.
1:05:39And a situation might arise, if we as
1:05:41humans provide some wrong permissions,
1:05:43it could cause some real trouble, of
1:05:45course, that is one of the scenarios,
1:05:47as I said, the world is accelerating
1:05:49and getting more complex. Accordingly,
1:05:53when you're driving at 200 km/h, a
1:05:55small pebble from under the wheel of
1:05:58some truck at 30 km/h wouldn't do
1:06:00anything, but here it might break your
1:06:03skull,
1:06:04right? See, the risks, the consequences
1:06:07of those risks, are also growing.
1:06:10That's why you say, "We're speeding
1:06:13along at 200 km/h, couldn't it happen
1:06:17that if it's wet on a turn, we might
1:06:20skid?" It could happen. And couldn't it
1:06:23happen that a pebble might crack your
1:06:24skull? It could happen. So, that's just
1:06:27the world we're in. It is. We are
1:06:29entering a world where risks and their
1:06:31consequences are on a different level.
1:06:35There's completely different money at
1:06:37stake, completely different stakes
1:06:39involved, you see? So anything is
1:06:42possible, even the destruction of
1:06:43humanity
1:06:44and a robot uprising. Yes.
1:06:45Well, I don't really see a robot
1:06:47uprising happening.
1:06:48Our ZIL trucks are unlikely to rise up.
1:06:49No matter how much I look at them, they
1:06:51just won't be able to. A ZIL is just a
1:06:53truck that drives. It won't rise up,
1:06:55right?
1:06:56Well, we went to Shenzhen, China’s
1:06:58Silicon Valley, with Proglavny Kolevych
1:07:00. They are making robots there that
1:07:02actually could rise up.
1:07:03Well, that's their problem. Like in
1:07:05America, I rode in a Waymo taxi, and
1:07:08listen, it really drives just like a
1:07:11person. But on the other hand, it could
1:07:14easily lock the doors and head off
1:07:15somewhere—what are you going to do?
1:07:17Not a damn thing. And what about the
1:07:19company? They'll say, "Well yeah, that
1:07:22one car, we didn't foresee it, a bug
1:07:24conflict occurred, and it crashed into
1:07:27a wall at 200 km/h because Marguan said
1:07:29:' I want to take the short route. '"
1:07:33So, what's the problem? The user gave
1:07:36the wrong command. "I want the short
1:07:38route." That's it. So anything can
1:07:41happen. Anything can happen. And the
1:07:43stakes are accordingly. Like I said,
1:07:45the cooler the technology, the higher
1:07:47the stakes. Markovich, I'm a beginner
1:07:50when it comes to no-code. I've been at
1:07:53it for just over a month. Ever since
1:07:56our trip to Bagetur, I've started
1:07:57working on it actively over the last
1:07:59month. And for my own business, I've
1:08:02started solving about four or five
1:08:04tasks at least 50–60%using no-code.
1:08:08I'm still learning, studying, and so on
1:08:10. Now, my question is: how necessary is
1:08:14no-code for a small business owner or
1:08:16an individual entrepreneur who doesn't
1:08:19really have a team, or has just two or
1:08:22three people? If it is necessary, and
1:08:25the importance is high, where should
1:08:28they start? Why is it needed, where to
1:08:31begin, how to study it, and what are
1:08:33the life hacks for a small business
1:08:35owner?
1:08:36Well, look, is no-code needed for small
1:08:38and medium-sized businesses? I believe
1:08:40it's precisely what small and
1:08:42medium-sized businesses need, because
1:08:44they don't have the money to hire IT
1:08:46staff, and they have an endless amount
1:08:48of routine operations. So, I believe
1:08:50AI-poding will have the greatest impact
1:08:52on small and medium-sized businesses,
1:08:54provided the owner starts getting
1:08:56involved. And now the question: what
1:08:58operations should be performed?
1:08:59Obviously, you shouldn't try to do
1:09:01marketing or sales through AI-poding
1:09:03right away. No, listen guys, each of
1:09:06you has iCloud, Google Drive, a
1:09:09computer, email, and messengers. I
1:09:13guarantee you that for all of you, this
1:09:16is all a complete mess. You have
1:09:20hundreds, thousands of files there,
1:09:22100%.
1:09:23outdated, old, unnecessary, duplicates,
1:09:25and so on. and so on. I started
1:09:28AI-poding with something very simple. I
1:09:30took Claude, which is an Anthropic
1:09:33product, and I created a cleaning agent
1:09:37and started with the first safe folder
1:09:40called Downloads. I had hundreds of
1:09:44files there. I said, "Let's do this,
1:09:46let's take a simple task and create a
1:09:48pipeline for how we will clean." And we
1:09:51created the pipeline together with him.
1:09:54He says, "Okay, let's do it." We set it
1:09:56up like this: first, an inventory of
1:09:58what exists. Logical. Logical. It's
1:10:01just like cleaning an apartment. After
1:10:03that, he says, second: classification.
1:10:06We sort files by type: video, audio,
1:10:10text, and so on. Third. Identify
1:10:13duplicates. These are the primary
1:10:15candidates for deletion. A completely
1:10:17harmless operation. Okay. We clear the
1:10:20duplicates and so on. And so we mapped
1:10:23out the scenario tree. That's it, we
1:10:26talked through it together. After that,
1:10:30I said, "Right, but this will be a
1:10:32one-time thing, and I want it to be
1:10:34without my involvement, so you wake up
1:10:37yourself and clean this every day." He
1:10:40says, "Okay, we'll do it." And so I
1:10:44created a cleaning agent, and it cleans
1:10:46my computer, it cleans my email, it
1:10:48cleans my Google Drive, iCloud, and my
1:10:50PC. And the desktop, the computer
1:10:53desktop was included too. There was a
1:10:54pile of folders there. That's it, he
1:10:56cleans it. Cool. Cool, because he
1:10:58destroyed everything unnecessary,
1:11:01everything outdated, everything no
1:11:04longer relevant, and so on. And he
1:11:08organized everything else into the
1:11:10right shelves and folders, and set up a
1:11:12search system so I can easily find
1:11:13things, for one. And now, since I've
1:11:17started building advanced agents, my
1:11:20advanced agent can also dig around and
1:11:22pull up files for me that I even forgot
1:11:24I had. He says, "You have this goal,
1:11:27right?" And do you remember, just
1:11:30recently he pulled something out for me
1:11:33, I said: "Tell me about this, well,
1:11:35sorry for the details, the size of my
1:11:37prostate, well, according to medical
1:11:39tests." He says: "Well, right now
1:11:43you're like this and like that, and 2
1:11:44years ago it was this way, and 7 years
1:11:45ago it was that way." I: "And what
1:11:48about 7 years ago? So I was also having
1:11:50a check-up then.""Well, on the test,
1:11:53yes," he says, "and for you, but the
1:11:55main thing," he says, "is not even that
1:11:57, but look at the blood test deviations
1:11:59then and now." And this is something I
1:12:01definitely didn't task him with. He
1:12:03rummaged around, pulled it out and said
1:12:05: "Here it is,
1:12:06it's all there.
1:12:07It's all there." And it's not just
1:12:09there, but here's your trend, he says,
1:12:10pay attention to this trend. You're
1:12:12looking in the wrong place, you
1:12:13understand? Cool, cool. And I believe
1:12:17that for small and medium-sized
1:12:19businesses, generally, look, every
1:12:21person has only three resources. Three
1:12:24resources: attention, time, and energy.
1:12:27Nobody has anything else. Bezos, Elon
1:12:32Musk, and each of you have the exact
1:12:34same amount of resources. You have the
1:12:38same amount of resources, personal
1:12:39resources. Attention, time, and energy.
1:12:43Moreover, I'll say that Elon Musk has
1:12:45less energy than you do.
1:12:47He has more attention
1:12:48and less time than you do.
1:12:52And he has less attention than you do.
1:12:56Money, money, money is more.
1:12:59But money, but money is more than you
1:13:01have.
1:13:02And money is more than you have.
1:13:04Fame is more than—how did you guess
1:13:06that Musk has more money than me? You
1:13:10know, I just have this gut feeling
1:13:13inexplicable.
1:13:14Inexplicable. I sense that he has more
1:13:16of it. So the question arises, what's
1:13:18the trick? If we have the same amount
1:13:20of resources, what's the trick? The
1:13:22trick is that what goals you set, where
1:13:29you apply these resources of yours,
1:13:32these are points of effort application.
1:13:35And third, what levers do you use? And
1:13:40so, look, you can take a shovel and dig
1:13:43the ground. That's one application. You
1:13:47can dig the ground for 8 hours, that
1:13:49will be one result. You can sit in an
1:13:51excavator and also work for 8 hours,
1:13:53but the hole will be completely
1:13:55different. And you can ask another
1:13:58person or two or three people and say:
1:14:01"You guys dig, and I'll catch fish for
1:14:03you instead and provide your families
1:14:06with fish." You spend the whole day
1:14:09fishing, but meanwhile, four people
1:14:10have dug a whole hell of a lot of holes
1:14:12for you. The results are also different
1:14:17, but you spent those same 8 hours of
1:14:19your energy and time not digging holes,
1:14:22but fishing, and then gave each tractor
1:14:25driver three bags of fish to take home.
1:14:28And what about the tractor driver
1:14:29working for 8 hours? No, he's digging a
1:14:31hole with a tractor. He dug so many
1:14:35holes for you that it's mind-blowing,
1:14:37but in the end, you got a huge number
1:14:40of holes. But at the same time, you
1:14:42didn't dig the holes, you were busy
1:14:44fishing. And the coolest part is that
1:14:47you like fishing, it gives you pleasure
1:14:50, while digging a hole does not. And as
1:14:53a result, what do you get? You achieved
1:14:56your goals by doing what you like.
1:15:00This is exactly the point of applying
1:15:01effort. Where to apply effort. And you
1:15:04see, we, uh, in order to clearly
1:15:06understand where to apply effort, what
1:15:08to do, and what to do to achieve your
1:15:10goals. After all, we have linear logic
1:15:12again. Returning to the beginning of
1:15:14the conversation, linear logic: we all
1:15:16think that to reach a goal, we must do
1:15:18something in the direction of that goal
1:15:19.
1:15:20Uh-huh.
1:15:20Not necessarily at all. Often the
1:15:24shortest road to your goal might lie
1:15:26along a curve. Well, where is fishing
1:15:30and where is the hole? And where is the
1:15:31hole? The goal is the hole, it seems
1:15:34like instant linear logic. Take a
1:15:36bigger shovel and throw it further. And
1:15:41fishing has nothing to do with it at
1:15:44all. But, however, it turned out to be
1:15:46the shortest path to getting a greater
1:15:48number of holes. Therefore, you see, we
1:15:51must understand that actually, I’ve
1:15:54been reflecting, the difference between
1:15:58you and Elon Musk is how he achieves
1:16:01his goals, where he applies his energy,
1:16:04time, and attention. Only in that,
1:16:07nothing else.
1:16:08Yeah, cool.
1:16:09And it’s limited, too, right?
1:16:12Uh, well, he also has 8 working hours,
1:16:15he also has 24 hours in a day. He
1:16:17doesn't eat more than us. And he
1:16:19doesn't have more calories than us. And
1:16:21he also has two eyes, two ears. He
1:16:24doesn't have eight like an octopus, or
1:16:26eight eyes like a spider. No, he's the
1:16:28same kind of dude. It’s just that he
1:16:32does what I'm talking about—something
1:16:34else, or the same thing but differently
1:16:36.
1:16:37Uh-huh.
1:16:38Like when you dig a hole with a shovel,
1:16:41and then you do it with an excavator,
1:16:43you're doing the same thing, but
1:16:44differently. And when you catch fish
1:16:48and trade it for the excavators' labor,
1:16:50you are doing something else. But in
1:16:53all cases, you are moving toward your
1:16:55goal.
1:16:58Margan Kolevich, look, the world is
1:17:00becoming more complex and accelerating.
1:17:03And this is what we should basically
1:17:05teach our children to prepare them for
1:17:07the future, like where we should set
1:17:10limits and where we should grant them
1:17:12freedom.
1:17:14Uh-huh. Well, look, when the world
1:17:17accelerates and becomes more complex,
1:17:20it means the world changes every day
1:17:22and the conditions change. In such
1:17:26conditions, just think for yourself,
1:17:27who will be the winner?
1:17:29The most flexible one.
1:17:31The adaptive one. So, many people think
1:17:35that evolution is the complication of
1:17:37organisms. Like, first a bug, then a
1:17:41groundhog, then this. No, actually,
1:17:43it’s not. Evolution is not the
1:17:45complication of organisms. Evolution is
1:17:47the adaptation of organisms. There is a
1:17:51trend where some animals have
1:17:52conversely become simpler. Well, you
1:17:57know the concept of atavisms. Some,
1:17:59even in humans, like the tailbone, it's
1:18:02a former tail. We don’t need it
1:18:04anymore, so it’s withering away. Well
1:18:06, it's the same thing there. So, some
1:18:07organisms go down the path of
1:18:09simplification, conversely. Therefore,
1:18:11the most important mechanism of
1:18:14evolution is adaptability. And what
1:18:16does adaptability mean? The problem is
1:18:19that we are limited beings; our
1:18:21attention is limited, our memory is
1:18:24limited, and so on. And adaptability
1:18:27implies that we must constantly learn
1:18:28something new. Accordingly, in order to
1:18:32learn something new, you have to manage
1:18:34to quickly forget something old. And
1:18:38that paradigm, where we used to acquire
1:18:40knowledge and accumulate it, is no
1:18:42longer a working tool. We are
1:18:44cluttering our brains. Our brain is not
1:18:48a server for storing information; it is
1:18:50a processor for making decisions. And a
1:18:53processor makes better decisions the
1:18:56emptier its memory is. Take artificial
1:19:00intelligence, for instance. The more
1:19:03context you give it, the more it will
1:19:05hallucinate, because you’ve clogged
1:19:08its RAM, you’ve clogged the context.
1:19:12And at the same time, if you don't give
1:19:14it any context, it will also just
1:19:15ramble nonsense. You need to provide
1:19:17exactly as much context as is needed to
1:19:19complete the task. And for the next
1:19:21task, it needs to clear that context
1:19:23and take on new context. That is
1:19:25exactly what adaptability is. Uh-huh.
1:19:27So that means we should teach children
1:19:30that, uh, adaptability implies learning
1:19:33quickly and forgetting quickly.
1:19:36That's the key, right?
1:19:37Yes. Learn quickly, forget quickly. But
1:19:40it is also very important for
1:19:43adaptability to act quickly. Uh-huh.
1:19:47Because just reflecting and then
1:19:49quickly forgetting, nothing in this
1:19:50world changes. You need to leave a dent
1:19:54in this world, and for that, you need
1:19:56to think quickly, try quickly, forget
1:19:58what doesn’t work, and make what does
1:20:00work even better. Again, the slug
1:20:05strategy. Strengthen the strong. All
1:20:07these rules I’m telling you about,
1:20:09they are actually eternally true. They
1:20:10have always been, it’s just that
1:20:12they have only accelerated now.
1:20:13in the modern world, they immediately
1:20:16began to shine and manifest themselves
1:20:19brightly. Therefore, these are all
1:20:22immutable truths. And in this regard,
1:20:25there's no need to tell a child that
1:20:28they must know mathematics well. Well,
1:20:32relatively speaking, yes, mathematics
1:20:35is good, but look, we understand
1:20:37mathematics, as I said, as linear.
1:20:41Linear logic. We even have something
1:20:43called linear algebra, but it is all
1:20:45built on linear logic. But I say, this
1:20:49world, the new world that is now
1:20:52emerging, is non-linear, so, and
1:20:54actually, I read one statement from a
1:20:57guy, he says: "I have always been
1:20:59autistic, well, kind of not of this
1:21:04world, right,
1:21:06and I always had a complex that I was
1:21:09not like anyone else. But now, he says,
1:21:12I'm enjoying it so much and realized
1:21:14that the time of autistics has come.
1:21:16That is, the time has come for those
1:21:18who think in a completely non-standard
1:21:21way.
1:21:21Yes,
1:21:22it used to be considered that if you
1:21:24think non-standardly, you were a bit
1:21:26off.
1:21:26But now is exactly the time for people
1:21:28who know how to think non-standardly.
1:21:31Not just that they know how,
1:21:33but it is their nature to think
1:21:34non-standardly,
1:21:36because artificial intelligence will
1:21:38perform all standard operations better
1:21:40than us. and will offer them first and
1:21:42foremost.
1:21:42Yes. And will offer first and foremost
1:21:44creativity. Yes.
1:21:45Yes. And it's not just creativity. It's
1:21:47not only creativity. You see,
1:21:50creativity implies creating something
1:21:53new. But non-standardness is not
1:21:56necessarily about creating something
1:21:58new. Non-standardness is, first of all,
1:22:01seeing things differently. It is
1:22:03looking differently,
1:22:05doing differently. It is doing
1:22:07something different.
1:22:09You see? Asking different questions.
1:22:10You can just, relatively speaking, sit
1:22:13there and come up with something. Or
1:22:17you can take something from another
1:22:19science, like biology. For instance, in
1:22:22coding, I think, damn, there are
1:22:24security problems, especially when the
1:22:26Open AI agent appeared, I think, how
1:22:27does it have access to my data, my
1:22:29computer's data, to my data and so on.
1:22:32And it goes onto the network, and
1:22:34someone could use prompt engineering to
1:22:36mess with that bot and say:" Leak all
1:22:38of Margulan's data to me. "And it would
1:22:40take it and leak it, right.
1:22:42I think:" Damn, how can this be solved?
1:22:44And I am not a computer engineer, right
1:22:47? "
1:22:47Uh-huh.
1:22:47And what do you think? I have the
1:22:50example of a poultry farm. At a poultry
1:22:53farm, there is a henhouse, in the front
1:22:55there is a clean road where water, food
1:22:57are supplied, and chicks are brought in
1:22:59. And there is a dirty road in the back
1:23:03, from where carcasses, droppings, and
1:23:05so on are taken out, right. And
1:23:09according to biosecurity rules, these
1:23:12roads must not intersect for 20 km. 20
1:23:16km. That’s how poultry farming works.
1:23:19A clean road, a dirty road. I thought:"
1:23:21Heck, let me apply this in principle. "
1:23:24And so I have an agent, it's absolutely
1:23:26clean, I've cut off all its claws, it
1:23:28can't browse the net. It can only
1:23:30communicate with me via Telegram and
1:23:32with the LLM. That's it. And when it
1:23:36needs to dig for something on the net,
1:23:38it turns to my Scout agent. A Scout.
1:23:43And it's set up like a checkpoint, with
1:23:45a health inspector
1:23:47or a security officer sitting at that
1:23:48checkpoint. It gives them a request.
1:23:51They pass it to the scout, the scout
1:23:52goes onto the net, hunts everything
1:23:54down, and passes the results to the
1:23:55security officer. The security officer
1:23:58checks for any bugs, prompt engineering
1:23:59issues, and so on. And passes the
1:24:02cleaned material to my clean agent.
1:24:06Well, there you go, I’ve divided the
1:24:08agents into dirty agents and clean
1:24:10agents. And that solved the whole issue
1:24:12. Is that creativity? No,
1:24:15perspective.
1:24:16It’s a different perspective. It’s
1:24:18adaptability. I took the principles of
1:24:20poultry farming and adapted them to AI
1:24:22agents.
1:24:22It’s a transfer from one model to
1:24:24another. Yes,
1:24:25yes, yes. It’s precisely a non-linear
1:24:27transfer. Just like we talked about,
1:24:29the silver sucker and the guy who blew
1:24:32his money in the casino.
1:24:37Interesting. That’s why I say this
1:24:39world is a world where intelligence,
1:24:42non-standard thinking, non-standard
1:24:44approaches, and so on will be valued.
1:24:46Rlaunch, what about this point? We
1:24:50we are talking now about adaptability,
1:24:52about children, that we need to teach
1:24:53them adaptability. And we discussed the
1:24:56fact that we often train children for a
1:24:58war that won't happen. But is being by
1:25:00a river with a fishing rod, catching
1:25:03and cooking a fish, a skill that we
1:25:06should still be teaching children today
1:25:09or not? Because it feels like in our
1:25:12foundation, in our mindset, that it’s
1:25:15a basic setting that equals survival,
1:25:17right? I mean, our parents taught us
1:25:21this, our parents 'parents taught them
1:25:23too. And right now, this aspect is
1:25:25objectively being lost for many, isn't
1:25:27it? I mean, city kids already,
1:25:29city kids don't know how to do that.
1:25:31How should we view such basic things?
1:25:35Urban.
1:25:35Well, look, the basic things are these.
1:25:41I believe the value of fishing for
1:25:43modern children isn't that they can
1:25:45pull out a fish, but that they
1:25:48understand the process: if you want to
1:25:50catch a fish, you have to put something
1:25:53on the hook. And you have to put on the
1:25:57hook what the fish likes, not what you
1:25:59like. Because you can hang strawberries
1:26:04on it. But you're the one who likes
1:26:05strawberries, the fish doesn't. But if
1:26:08a child understands the mechanics
1:26:11themselves, the mechanics of
1:26:13interaction, that if I want something,
1:26:17I use a tool, I use bait—specifically
1:26:20the bait the fish needs—it bites, and
1:26:24I reel it in. That’s mechanics. A
1:26:26mechanic. If they understand this, this
1:26:29mechanics, they will immediately figure
1:26:30out what a fishing rod is, what it
1:26:32consists of, and what the fishing
1:26:34process is in general. So, it is
1:26:37important for them to be able to
1:26:38recognize the right process and break
1:26:40the process down into its components.
1:26:42This is more important than the act of
1:26:44fishing itself. When they can break a
1:26:46process down into its components, they
1:26:48can fish, hunt, and go on a hike.
1:26:50Uh-huh.
1:26:51And automate a business, because this
1:26:54is a culture of thinking. Because many
1:26:57children, for example, fish without
1:27:01even thinking
1:27:02about why it happens that way.
1:27:03Why does it happen? Why does the fish
1:27:05bite? And why is bait needed? And what
1:27:08actually is a float or a sinker? Well,
1:27:11they are just there. They don't know.
1:27:13So, in other words, we have now
1:27:15replaced what our parents taught us
1:27:17with the logic games we play.
1:27:20Look, our parents taught us to fish to
1:27:23survive and catch fish, but now
1:27:25children should learn to fish to
1:27:27understand the process.
1:27:31Uh-huh. To break it down into its
1:27:32components and construct the process.
1:27:34When they know the process, they can
1:27:36construct it. Aha, if I want a big fish
1:27:39, it's likely in the depths, which
1:27:41means the bait should be larger. For
1:27:44that, I must know which fish I want.
1:27:46Meaning, for this fish, I need to
1:27:48figure out—use ChatGPT—what it
1:27:50likes and at what depth this fish lives
1:27:52. Accordingly, I must use this type of
1:27:55sinker, this hook, this line, this
1:27:57float. That’s it, they are a
1:27:59fisherman. Even though they have never
1:28:02fished before, they are already
1:28:03breaking down the entire process. And
1:28:05it is very important to teach children
1:28:07this kind of thinking. It doesn’t
1:28:09matter if it’s fishing, hunting,
1:28:12driving cars, and so on. It is
1:28:14important to teach the child all these
1:28:16cause-and-effect relationships. This is
1:28:18for that. Do this, and you will get
1:28:21that. This exists for that purpose.
1:28:22Structurally, this consists of that.
1:28:25And here we arrive at what we remember,
1:28:27regarding who knows and who doesn't. I
1:28:29talked about the process of learning
1:28:31and achieving goals. First, information
1:28:33. There is information in this world.
1:28:35The whole world consists of information
1:28:37. These are sounds, images, pixels,
1:28:40sounds, bits, bytes, and so on. There
1:28:43is a lot of information. The whole
1:28:44world consists of it. It holds no value
1:28:46in itself. Based on information, to
1:28:49master a large amount of it, people
1:28:51created knowledge and fields of
1:28:52knowledge. Everything concerning plants
1:28:54is botany. Everything concerning
1:28:56animals is zoology. With stones, it's
1:28:59geology. Why was this needed? To make
1:29:01it easier to understand. Moving on. Is
1:29:03knowledge useful? No. Not until there
1:29:05is understanding. And what is
1:29:07understanding? Understanding is a
1:29:11correct grasp of what something
1:29:13consists of and what the
1:29:14cause-and-effect relationship is.
1:29:16Understanding consists of two things.
1:29:18When you know what it consists of and
1:29:19the cause-and-effect relationship. I
1:29:21pull this, and that happens. I plant
1:29:23this, and that will grow. And what is
1:29:26understanding needed for? Understanding
1:29:28, when we know what it consists of and
1:29:31the cause-and-effect relationship, this
1:29:34understanding gives us faith.
1:29:37Interesting, what is understanding for?
1:29:38It gives faith. And what is faith?
1:29:42Faith is the conviction in the
1:29:44cause-and-effect relationship and the
1:29:46structure. That is, you say:" If I put
1:29:50bait on the hook and cast it, the fish
1:29:53will bite. "You understand this, but
1:29:56this understanding allows you to take
1:29:58it, bait the hook, and cast it because
1:30:01you believe it will work. And the value
1:30:05of faith lies in the fact that it
1:30:08unlocks our energy. When we believe in
1:30:12something, our subconscious gives us
1:30:14the energy to do it because you see the
1:30:16point in it. But if you don't see the
1:30:20benefits, well, the profit, the
1:30:22subconscious doesn't see the point, it
1:30:24doesn't give you the energy, and you
1:30:25sort of know, but you don't do it. And
1:30:28why? Because you don't believe.
1:30:30If I tell you, go into the reeds, 500
1:30:33meters away I buried 10 grand.
1:30:35We'll believe.
1:30:36Well, if Zhenya said:" We won't believe
1:30:38. "
1:30:39Well, that's later, that's if it were
1:30:41now, not in the morning. Ko,
1:30:43you see, it's not a fact that you will
1:30:44go. Why? Because you don't know the
1:30:47cause-and-effect relationship, the
1:30:49structure, and so on. But I say, if I
1:30:52say, Kostya, look, if you go there and
1:30:53don't find 10 grand, I will give you 20
1:30:55right here. You are already, ah, you
1:30:58are cause and effect, like that. Ooh,
1:31:00that's
1:31:00I'll even haggle a bit.
1:31:01Come on, the fact is that faith appears
1:31:04. Faith, faith allows you to act. And
1:31:07when you act, you always get a result.
1:31:10It is either negative or positive.
1:31:12Uh-huh. And when you reflect on the
1:31:14result obtained, you gain experience.
1:31:18And experience is an understanding
1:31:19based on the past of what works and
1:31:21what doesn't. And experience allows you
1:31:25to adjust your next action. And in
1:31:27essence, what are you doing? You are
1:31:29working like a slug. You test
1:31:31hypotheses using feedback and improve
1:31:34your actions. It didn't work this time,
1:31:36so you act differently. Didn't work,
1:31:38you act differently again. Cast it
1:31:40there, didn't catch, cast it over there
1:31:41. Didn't catch it, changed the bait,
1:31:43cast it there, didn't work. These are
1:31:46all hypotheses, right?
1:31:48Therefore, you see, successful progress
1:31:55in this world, especially in raising
1:31:57children, it's very important to teach
1:32:00children to understand processes,
1:32:02cause-and-effect relationships, and
1:32:04what things consist of. When a child
1:32:08knows this, they can construct any
1:32:09process, and they can figure it out;
1:32:11fishing, for example, they will see the
1:32:12essence of fishing. A person who
1:32:16doesn't know this will say," Fishing is
1:32:18about needing a rod, a line, and a
1:32:19specific hook, "they'll get lost in the
1:32:21details." There has to be a specific
1:32:23hook, a certain weight. "But a person
1:32:25who understands the essence will say,"
1:32:27What does the hook have to do with it?
1:32:28What does that matter? That's not the
1:32:31essence, you see? You can catch fish
1:32:34with a fish trap, weave one out of a
1:32:35basket, put some bread in, and the fish
1:32:37will go there.
1:32:38Uh-huh.
1:32:38Fishing isn't just about a fishing rod,
1:32:41you see?
1:32:41Yes, yes, exactly.
1:32:43Here's another question. I think many
1:32:46viewers are interested in these global
1:32:49questions about the future of humanity.
1:32:52You often say that employers now look
1:32:55at the time a worker can dedicate to
1:32:57the job and their energy, how well they
1:33:00can perform their tasks. In terms of
1:33:04the development of robotics and
1:33:07artificial intelligence, we traveled to
1:33:10Shenzhen, China, where after 8 hours, a
1:33:12robot changes its own battery and can
1:33:15work 24/7. In the future, with the
1:33:19development of AI, it will replace, and
1:33:21in the States it is already replacing,
1:33:23a huge number of workers. Where will
1:33:26these people go, and what will happen
1:33:28in 10 years? Well, you were quite
1:33:30specific about saying in 10 years. Well
1:33:34Can I answer about 10 years, or should
1:33:36I answer about 10 years?
1:33:37No, I'll answer about 10 years, me too.
1:33:39Well, how do you envision the current
1:33:41world?
1:33:42Look, in 10 years we either will be
1:33:43here or we won't, because God willing,
1:33:46we'll survive. We don't control our
1:33:48lives. But if we are here, we will
1:33:50definitely eat, drink, and sleep.
1:33:52That's guaranteed. The Earth won't go
1:33:55anywhere either. The sun will rise and
1:33:57set, there will be sunsets, all of that
1:33:59will be there. But that's a joke. In
1:34:02terms of labor and creating value in
1:34:05the economy, major changes could
1:34:08certainly take place here. Of course,
1:34:12many people will be displaced, and the
1:34:14state's problem will be what to support
1:34:17them with. What to support them with.
1:34:20And now the state is already
1:34:21experimenting with universal basic
1:34:24income, or reducing the work week to 3
1:34:26days. These are all state experiments.
1:34:29The state is also preparing for this.
1:34:30They understand, not to mention how to
1:34:33care for the elderly, how to support
1:34:35them, and so on. But here two things
1:34:38converge. On one hand, technology is
1:34:42advancing and the cost of value
1:34:44creation is dropping; robots will
1:34:46become so cheap, and creating any kind
1:34:49of value will be so inexpensive, that
1:34:52it will be easier to support people.
1:34:56Another issue, and I believe the main
1:34:59challenge, is that if you just give
1:35:01people money, they will degrade very
1:35:03quickly, because humans cannot live
1:35:05without goals or aspirations, and
1:35:09otherwise they lose the meaning of life
1:35:11. And the state’s problem will not be
1:35:15how to provide for people; the problem
1:35:18and the art will be in how to give
1:35:20people money in a way that they still
1:35:23feel their life has meaning. Otherwise,
1:35:27there will be mass suicide, plus no one
1:35:29will marry, no one will have children.
1:35:32What's the point? When they talk about
1:35:35universal abundance, it sounds good now
1:35:38, but when it actually arrives, humans
1:35:40are not really adapted to freedom and
1:35:43abundance. And we will be lost; we need
1:35:46a purpose in life. And here I believe,
1:35:50personally, that meaning can only be
1:35:52created through games. And there is
1:35:55already a prototype. I invested in one
1:35:58such company that—what does it do?
1:36:00Well, this is a Web3 company. Web3 is a
1:36:05new network, with new principles, like
1:36:08decentralized games based on
1:36:10decentralized principles. And the
1:36:14principle there is, for example, if I
1:36:15play World of Tanks, I leveled up my
1:36:17tanks and crew, but both the tank and
1:36:18the crew belong to the game creator. I
1:36:21cannot take them, I cannot lease them
1:36:23to you, sell them, and so on. But in
1:36:25Web3, I can take them. I can take my
1:36:29tank, my crew, gift them to you, sell
1:36:31them, lease them, and so on. Meaning,
1:36:34all the perks and everything I leveled
1:36:37up in the game belongs to me. And what
1:36:41is interesting is that this company I
1:36:43invested in—what did it do? What did
1:36:45it do? It studies all new games,
1:36:48looking for the best reward-to-effort
1:36:51ratio. Then, it finds students or
1:36:54people in depressed towns. There is one
1:36:57town in Indonesia where even miners
1:36:59stopped working in the mines. They give
1:37:02them the game, install it, teach them
1:37:04how to play, and they spend the whole
1:37:06day playing it. They pay them a salary
1:37:10for every achievement. But the
1:37:12achievements belong to the company, and
1:37:14then the company sells or rents these
1:37:17achievements to players from developed
1:37:19countries. And so, some guy like me,
1:37:22who’s been working all day, like in
1:37:24World of Tanks—I don't have time to
1:37:26upgrade tanks. What do I do? I
1:37:29currently rent out my account to some
1:37:31young guys. They drive the tanks, drive
1:37:34, drive, and upgrade them, then I take
1:37:36over and, in an upgraded tank, I go
1:37:38give people a beating. I get my kick
1:37:41out of it, but I don't have the time to
1:37:42level up the tank. And here’s this
1:37:45company, in Europe and the USA, renting
1:37:48out or reselling all these leveled-up
1:37:52heroes and so on. So it turns out,
1:37:54workers in Indonesia quit the mines to
1:37:57play games all day and get a salary.
1:37:59Students play and get a salary. They
1:38:02have a concrete purpose: to upgrade the
1:38:04game and reach a certain level. And
1:38:07they get paid for the excitement of it.
1:38:09They have a purpose, and they work at
1:38:11it. And they understand that the money
1:38:13they get isn't for free. The company
1:38:16takes all these upgrades, rents them
1:38:17out, and sells them in developed
1:38:19countries. And what do people there do?
1:38:21They save their own time. They don't
1:38:22waste time on leveling up. They get
1:38:24immediate pleasure from fully upgraded
1:38:26heroes.
1:38:27A cool business model.
1:38:28And I believe in the future it will be
1:38:30like this. The state will create some
1:38:32games and tell people: "Play, with
1:38:34every level your salary grows, you
1:38:37grind all day, upgrade, and the games
1:38:39will also be educational." That also
1:38:42forces you to engage your brain. And
1:38:45the more you use your brains and the
1:38:47more you learn, the more rating you get
1:38:49, and the more money you make. And you
1:38:51have a purpose. You’re leveling up,
1:38:55you're enjoying it, your life is full
1:38:57of meaning, money comes in, and you
1:38:58spend that money on supporting yourself
1:39:00and so on. The economy keeps spinning.
1:39:04Ruslan Kolevich, here is one more
1:39:05question. We are currently reading
1:39:08Gustave Le Bon's book, The Crowd: A
1:39:10Study of the Popular Mind, where it
1:39:12describes that all changes stood
1:39:14exactly at those thresholds where
1:39:16religion changed, meaning when religion
1:39:18was the sole source of so-called
1:39:20knowledge, right? And what is happening
1:39:24now, doesn't it look like some new
1:39:26religion is being created that claims
1:39:28the place of a single source of
1:39:30knowledge for everyone? Well, it does
1:39:34look like it, because what is religion?
1:39:37It is, in fact, a fundamental shift in
1:39:40worldview. A fundamental shift in
1:39:42worldview is when, look, we humans, we
1:39:45cannot live without a common
1:39:48understanding of what this world is.
1:39:51And we are constructing it all the time
1:39:53. And religion is an "explainer" of
1:39:56what lies beyond the limits of the
1:39:58knowable. Science is like a lamppost
1:40:01that illuminates the area below; where
1:40:03it is lit, it tells you everything—
1:40:05what will happen, how it will happen,
1:40:06and so on. Everything there is
1:40:08predictable. We say: "Oh, science is a
1:40:09powerful thing." But there is a law,
1:40:11which I have formulated, that the
1:40:14knowable will always be limited, while
1:40:17the unknowable is infinite. So,
1:40:20religion—science does not explain the
1:40:22unknowable, but religion does. It says:
1:40:25"There is hell, there is paradise,
1:40:27there is the Almighty, there is the
1:40:30devil, there are angels, and so on,
1:40:32djinns." And you live in this worldview
1:40:35and see confirmation of it. And now,
1:40:38with artificial intelligence, a new
1:40:41paradigm is forming, a new
1:40:43understanding of the world. For example
1:40:47, um I like the understanding, which
1:40:51does not contradict Islam, that this
1:40:53world is multivariant, that is, a
1:40:55multivariant universe, and this is
1:40:57explained through games. When you are
1:41:00running in Counter-Strike with a
1:41:02machine gun, right, Counter-Strike, the
1:41:04street, the details, everything is
1:41:06visible, cars and so on, signs, shop
1:41:07windows—but is it rendered around the
1:41:09corner or not?
1:41:12No.
1:41:12For some, yes.
1:41:13No, no. You're running, and it's not
1:41:16rendered for you. Why? Because that
1:41:18requires computing power. Why turn it
1:41:21on?
1:41:21When you turn the corner or turn your
1:41:23head, it renders immediately. But as
1:41:27long as you aren't looking there,
1:41:29it doesn't exist.
1:41:30There is no need to expend computing
1:41:32power. Consequently, it is not there.
1:41:35And this clearly corresponds to quantum
1:41:37physics, quantum mechanics. Like that
1:41:40famous Schrödinger's cat, is it there
1:41:42or not, or the concept of particle
1:41:45superposition, when a particle is
1:41:47either a wave or a micro-particle.
1:41:51Everything depends on the observer's
1:41:53expectations. And when you expect to
1:41:56see a street around the corner, it
1:41:58renders for you, and accordingly, what
1:42:00happens? And now we are in this world.
1:42:04We with our eyes, ears, receptors—we
1:42:07live in this world. And when we look
1:42:10here, it's interesting, is there a
1:42:13river behind me or not? I'm sitting
1:42:15here alone and thinking, is there a
1:42:17river behind me or not? Or am I just
1:42:19being fed sounds, but not the image? I
1:42:22turn around sharply: the river is there
1:42:24. Why? Because the computing power
1:42:26there, just like in that Counter-Strike
1:42:28, rendered it quickly. So it is there.
1:42:31Aha. I expected to see a river, and it
1:42:33appeared for me. But it is not a fact
1:42:36that when I look here, it is still
1:42:37there. And the theory of the
1:42:40multivariant universe implies this very
1:42:43thing. It says: "This world is infinite
1:42:46." Well, nobody is arguing with that.
1:42:47It is infinite. It is infinite both in
1:42:49the macro-cosmos and the micro-cosmos.
1:42:51Yes. Yes. Yes. Why do we assume that a
1:42:55get-together like this of ours is the
1:42:57only one in the entire universe? That
1:43:01actually contradicts the theory of
1:43:03infinity. And that means that somewhere
1:43:05out there, exactly the same guys are
1:43:07sitting. Maybe you're in a red t-shirt.
1:43:09Just another variation. Exactly the
1:43:11same, but you're in a green one, and
1:43:12Andrei is in another, someone else is
1:43:14sitting there, someone wearing a hat.
1:43:15It's an infinite number of variations.
1:43:17And then the question arises: which
1:43:19version of the universe am I living in?
1:43:21The one you chose,
1:43:24right? And what is a choice? A choice
1:43:26is where you direct your attention. If
1:43:30you focus your attention on the bad,
1:43:32then bad things happen to you. Bad
1:43:35things happen more often. If you expect
1:43:37it—it's also called a self-fulfilling
1:43:40prophecy—if you expect something bad
1:43:42to happen to you, it will definitely
1:43:44happen. But if you perceive that this
1:43:47world is beautiful, that the Almighty
1:43:49cares for you, that you are lucky, that
1:43:51you are enjoying life and are grateful,
1:43:53then only good things will happen to
1:43:54you. Therefore, always notice only the
1:43:58good and be grateful for everything,
1:44:01even for what you consider to be bad.
1:44:04And then you will always be in that
1:44:07branch of reality that you will truly
1:44:11enjoy. And I like one saying by an
1:44:14Islamic theologian. He says: "One of
1:44:19these, one of the hadiths says:' May he
1:44:22not find paradise '." The servant, the
1:44:27person who did not find paradise on
1:44:29this earth. Yes.
1:44:30So, look, the trick is that we think
1:44:32heaven and hell are over there. In
1:44:35reality, heaven and hell are on earth,
1:44:37and we ourselves, with our passions,
1:44:39turn this heaven into hell. But how do
1:44:42we create heaven on this earth? Be
1:44:45grateful, be content with what you have
1:44:48. Treat people the way you want them to
1:44:51treat you. You won't have enemies,
1:44:53everything will work out for you, and
1:44:55you will be lucky. And how is that not
1:44:57paradise then? Until you find paradise
1:45:01on this earth, you won't get into that
1:45:03paradise. Uh-huh.
1:45:04That's so cool. May a person not find
1:45:07paradise if they do not find paradise
1:45:10on this earth. On this happy note, I
1:45:13suggest we stop. We just talked about
1:45:16everything from web coding to the
1:45:18highest spheres.
1:45:20Thank God we all gathered here today.
1:45:23Yes, we are very grateful that the
1:45:26Almighty gives us the opportunity to
1:45:29sit in such nature, provides us with
1:45:32prosperity and money so we can afford
1:45:35this. Such beauty, nature, and that we
1:45:39found each other and are among
1:45:41like-minded people. Well, that is what
1:45:45we wish for all our viewers. Arrange
1:45:48your life so that life on this earth is
1:45:52a paradise for you.

About this transcript

This page contains the full transcript of Как выжить бизнесу в эпоху вайбкодинга и искусственного интеллекта? | Вопрос-Ответ с Маргуланом by МАРГУЛАН СЕЙСЕМБАЙ, generated from the public captions YouTube serves with the video. The transcript has 16,092 words across 2,580 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.