Про жизнь и искусственный интеллект. Большой разговор с профессором Иваном Ямщиковым

[music]

So how much time do we have?

Guys, this is how a quantum computer will be: one continuous temptation, intelligence which for everything will be decided by us.

And let us do it; we won't build either because something it's scary.

Let it be better like that.

Now I have a will of bondage going on, analogies with a nuclear bomb, for example.

Yes, when the war started, I decided that I would want to call a spade a spade.

Laws do not allow this to be done without risk for one's own freedom.

Fascism is such a disease of growing up; different nations carry it in different forms, but almost everyone brings it.

So that you have it, there are many Norsk laureates needed: 400 years of simply funding the university, and then somehow you start to accumulate results.

Must be remembered that we are territorial monkeys, after all, quite aggressive ones that have cool signal bioacoustic.

Your what do you have?

Sims, more topics you have, more of this free will itself is a more modest attitude of the QC to my own intellect.

It seems good.

[music]

We are in Würzburg; this is an old little town in Bavaria.

Seems like this town has been here for more than 1,000 years.

Besides everything, there is a very old university, one of the oldest in Germany, about 600 years old.

And that's it.

It is symptomatic that we will be here to talk about one of the most fashionable modern themes that are only on our planet: artificial intelligence.

Still, by the way, they say that Würzburg is the center of German winemaking.

I noticed that right here on the slope, the castle is such a fundamental castle, so to speak, in this town.

Right on grapes grow on the slope of the castle.

Apparently, this is what Karabas's possessions are like.

Barabas once were here, I meeting Russian in the past.

And now, apparently, German and international scientist Ivan Yamshchikov.

He worked in Yandex, worked at Skill Factory, studied in Germany.

In general, we will talk with him about the most advanced science in the field of the most advanced.

[music]

Topics.

Hello!

Hello!

How are you?

Here, what do we have?

You know how, like this, we have two pikes.

And this is in France?

Yes, yes, yes.

And this, what is this?

A gift?

Riesling.

Riesling?

Yes, we in Germany around Loza.

I'm already on this for now.

This is the biggest one behind us.

Germany's largest vineyard is called wow stone wine single vineyard.

There are some fierce number of hectares; it goes along the dolman.

Well, yes, I listen.

I just didn't notice that at first.

Everything is really so cool there.

There is a local tradition here.

It is possible to participate in the Queen of the Vine competition.

It's like a beauty queen.

But to participate in families, girls must be at least 100 Loza in ownership.

Is it a lot or a little?

Well, 100 of these pieces of grapes are not a lot of space, much more.

Yes, and here they grow accordingly, make white wine.

That is, these are famous German Riesling.

Listen, well, we are in the region which is called Franconia.

Gio, it's called a frak.

Are there any bottoms?

Franconia or Barbarians?

Franks, they then let's go a little further west.

Did some stay here, decided that it's already good here?

And actually, here half the region Franconia, that is like this.

A little 100 km to the south is Limes; there was a border of the Roman Empire.

There was a wall.

And we are as if on the side of the barbarians who after le Rayat.

[applause]

[music]

Here, listen, explain to me, we are in Würzburg.

Why are you here?

I work here.

I am a professor at the Higher Technical School in the city of Würzburg.

There are three universities; there are three higher education institutions here.

The University of Würzburg is here, the conservatory, the highest technical school.

And I am a professor in one of them.

There is a conservatory center here.

I mean music centers.

Yes, they play there organs and all that.

Wow, that's it.

In general, this is what a university town is like.

There is a castle on the mountain called Marienberg.

Bishops lived here.

Here, the secular was combined with the religious.

It was such a religious city in which was one of the first universities in RSU.

University appeared in Germany.

Actually, this is the wine that I give you; it was called from the cellar, which is called Julius' cellar or Julius Hospital.

It is named after him, Bishop Julius himself, the university founded at which they became and founded a hospital here.

There further, which is still like that.

It's called the city hospital, and the restaurant at this hospital still exists.

It's called that because, well, that's what it's called: feed patients.

So it needs to be done somewhere, cook, and then you also need to somehow for medicinal purposes.

And in general, somehow it's all so overgrown.

Listen, I saw that the last Nobel Prize related to this university is in medicine.

Yes, yes, yes.

They have, in general, many are the most famous here.

Nobel laureate from Vers Buur is X-ray.

Everyone who has had an X-ray knows that as if they have some relation to this city.

Well, this is a university city; there's something here.

Order of children of Nobel laureates: 14.

14 here are 14 Nobel laureates, 120,000 residents.

All over Russia, specially watched 30 Nobel laureates, two in the world.

Zamir and also in literature, five.

Yes, first of all, we are for the Nobel Prize.

This is the Nobel Prize for Literature.

This is a big deal; she didn't do anything about it.

No, she, I mean, there is no here, here there is no literature here, only physics, chemistry, medicine, some applied things people do.

But here it just works out, which is slightly more than 10,000 inhabitants, one Nobel laureate for the history of the city.

And it is connected with the fact that it has been here for 500 years.

University, and how would you like it?

There are many laureates; we need 400 years just to finance the university.

Then somehow you start to accumulate the results.

Here are very strong society biased towards science.

Science is a social lift here.

If you become a doctor, that is, you receive a PhD, it's printed on tickets.

They call you; people say hello, doctor such and such.

If you come to the hotel and check in, they have a free room, better class.

They often themselves to you they offer because they think that, well, as if scientists.

Well, in short, this country has become industrial.

How much was there 150 years ago?

Industrial countries live off brains; you come up with something, and everyone makes money.

So when you're a doctor or a professor, it's important here, social lift.

And here people on average unemployed in education, at least for now.

Maybe this will change, but for now, they have great respect for people who work with their heads.

University, I understand it relatively young, yes, a little bit.

The youngest, in my opinion, University of Bavaria in the sixties he appeared.

That is, it is so in the 6th century.

Yes, yes, yes.

That is, it is a thing that grew up just like in St. Petersburg, let's say ITMO.

ITMO appeared initially from a vocational school for workers, which were there in Petrograd, worked in St. Petersburg before the revolution at the factories.

I think either Zeiss or Siemens workers had to be taught somewhere.

Then word for word, and as if slowly, but technical professional is true education is becoming higher education.

Here too, there was approximately the same story.

That is, a school appeared in the 19th century; they taught skilled workers.

Then it all somehow developed, developed in the middle of the 20th century, turned into a higher educational institution.

A there is some department here, AI, I understand, an artificial intelligence center has appeared here.

Intelligence, which I have been leading since October.

Adv, thank you, dick professor or what?

Yeah, right, dick professor.

I'm in charge before this since October.

Yes, a center has appeared here, artificial intelligence, because we are in Bavaria.

This is how it is.

Germany, how is it?

League, bes Germany is a Federation.

The Federation is made up of pieces; these pieces are called lands.

This is the Earth; it's called Bavaria, and it has its own National Bavarian Program development of artificial intelligence parallel to German.

That is, Germany for the development of artificial intelligence spends something like 4 billion euros.

Bavaria spends 3.9 billion euros.

How much sow Germany?

Well, to do such a curtsey so that it would be a little less.

But still, how polite.

And accordingly, within the framework of rich Bavaria.

Listen, well, they have long ago guessed that science brings benefits to people.

They are not the first time.

That's what they do in the eighties.

They invested well in all kinds of biotech and pharma; now they want to invest well in artificial intelligence.

Pull thread it with American scale.

Well, it's clear that Europe is like a small pipe, lower, thinner smoke, everything is clear.

There is such a thing; there's a thing in mathematics called power law.

That's almost all related to people by their behavior distributed gradually according to the law, you know, Gaussian normal distribution.

Here, Gauss is a pretty rare thing.

And how everything that is connected with human behavior is distributed gradually.

The law is when, I don't know, here is one track Billie Eilish is listened to by billions of people there.

And how billions of musicians publish tracks that one person hears.

Yes, that's it, as if this is the so-called distribution with heavy tails.

All financial markets are also arranged.

That is, you have there is and you have a winner in some, I don't know what industry occupies there 70-80 about this industry.

And then there is second, third, fourth, fifth player.

And in areas of artificial intelligence, staff is now confidently in first place.

The amount of money there are results and companies in second place, China.

Well, that's all.

[music]

The rest.

Listen, I understand that this city suffered greatly during World War II.

Although it looks, it is now quite old, that for Germany is not typical.

It means strongly suffered is putting it mildly.

Here, 16 March 1945, in 15 minutes, 90% of houses were destroyed in the historical center.

And died, in my opinion, 99,000 civilians in 15 minutes.

The same squadron that bombed Dresden was bombed.

But everyone knows about Dresden thanks to Kurtu Gutu; about Vrs Buur knows much less.

And still, here every year on March 16th in the evening for 15 minutes, the bells ring exactly at the time when it was going.

There is a bombing here; the locals joke that here one church for each degree.

There is a lot of hardness in the church because it is 40 degrees here extra.

And the water is very hard; thanks, this is where good white wine grows.

There are many different minerals in the pond, and that's all.

Churches, 40+ churches here are beating 15 minutes.

Well, yesterday I heard only one within 10 minutes, so it's impressive.

I think, listen, this is some kind of was.

We can try to take n here if they work.

It was some kind of just rage, retribution, some kind of senseless action.

Or here was Kato center; difficult to say.

Well, at some point, the British decided to commit war crimes for which there was nothing.

That's the end of it, stories.

E, you take the mug, and then give it back to you.

They return 3 euros, but no, to you they return the same amount as you pay for the mug.

Just if you want, you can take it away with you as a souvenir.

Here often, well, here we have our own circles, and often here is the official mug of the local fairs.

And it is accepted in any place; you can hand it over somewhere.

Ho, yes, in this.

It's great not to litter with plastic city.

I understand that this is a German tradition.

This is the gluwein in the mugs that they come back.

I have seen this in other places; there's not only gluhwein here.

There's also a bar; you will take beer.

You will have a price for beer and the price for what is called in German fn, that is, for the glass, which you can break it or you can not break it.

This is a fight against theft, so to speak.

Inventory returning to history, Würzburg, I understand that everything is in general what we see here is that it has been restored.

Well, yes, 90% restored from pictures.

Something the structure is made to look old-fashioned; something partially preserved.

And accordingly, it was built back here on the left.

There is a panorama of how the city center is, the city looked like after the bombing.

Well, that's it, quite impressive.

Yes, but drain decided not to restore everything.

Let's go listen; let's stop.

What post?

What does it mean decided?

Dda happened approximately the following situation.

There became part of the GDR, and they were like this.

Well, as far as I know, the cities were destroyed in Warsaw bloc war.

Soviet Union suggested several alternatives.

They said, for example, you want to restore there.

If you want the old, you build the new; we want it.

We'll build you a metro, or do you want a high-rise?

For example, Warsaw, the Warsaw skyscraper, Rock like Stalin.

Stalin like skyscraper; she is literally like that.

At that time after the war, it was built under Stalin, as far as I understand.

Dresden said, "No, we will restore everything."

Well, that's it.

We immediately said that we couldn't restore it; it will work out because there will be no money.

So let's restore some basic key things.

And at the same time, it seems to me some buildings in Dresden have been completed to be restored already in the united Germany.

That is, already the Warsaw bloc fell apart, and that's it.

And they finished everything; they restored the node there.

As it happens, the degree of meaninglessness is also important to understand.

Yes, that is, I kind of come from Petersburg, and it is clear that history there are blockades and so on and so forth.

That everyone it concerns me, and literally there are grandmothers there.

They told me about my grandfather.

That is, I have there great-grandfather returned with a piece of shrapnel in the head and great-grandmother.

Go, as if in the war, it's just me.

Listen, how do you like it?

Well, you, well, I chose Germany myself, and you studied there.

I studied in Sweden; I studied for a Master's degree in Sweden.

Then I returned to Russia, worked in Moscow in Yandex for almost 3 years.

And then I entered here for graduate school because my professor from Sweden moved here.

Wrote to me, "If you want to come to our graduate school."

And then I went observer at the elections in the thirteenth year.

I saw that in the southwest of Moscow, Vladimir Vladimirovich Putin is gaining more than half of what we had honest plot of land in the southwest of Moscow nearby.

I looked at all of this with the university, and well, I guess I should go study.

Apparently, I don't understand something in life; I went to study.

But when did you choose Germany?

At least just because of the professor, that's all.

In general, it is better to choose in science the person you will work for than whatever it is, than just a sign.

Of course, but you don't have no, no, it was moving.

You didn't think about that?

Oh my God, I chose Germany to study.

Then I chose something else for life move.

I don't know; I don't know.

It seems to me, again, there, despite the fact that everything is like that, family situation.

My grandfather always said, "The German man, a serious man."

And I have always treated with great respect and German engineering thought and to German scientific thought.

Fascism is such a disease growing up of nations; different nations carry it in different forms, but almost all of them she is being transferred.

Well, Italy, you are like, if I worked in Italy, you wouldn't help me.

I said, "But didn't you think when you chose?"

Italy, this is Dolcevita; this is, of course, it's good; this is pizza, pasta.

I, in general, I think it is important to know history.

In this sense, I think it's useful knowledge; it gives this very, firstly, it gives context.

Secondly, it gives understanding.

What the hell happens in this world anyway?

At the same time, you need to understand that, well, there I am.

I say Portugal, Spain, Italy; they all went through a stage of fascism in one way or another.

The states could have passed in a different form, and as if in Britain was some kind of motley, extremely popular, could win the election there would be fascism in Britain.

Well, that is, so it happens.

I think Tom, we are like grandchildren and the great-grandchildren of those who saw the war.

We go to Germany and drink calmly mulled wine; we work, we live, we study.

I naturally, I can't help but think about transfer Russia, Ukraine, now war.

It's impossible not to even say this inappropriate.

At what point between people normal relations will be established again?

I don't know; it seems to me that the situation is different, century, another context, another history will show.

Depends greatly on how they will be the participants of this conflict behave after it ends.

Strongly, it depends on how it ends, yes.

There are some in Germany too; that's why Germany is now such a place which other democracies are watching.

And damn, they often bring up Germany as an example because there is culture here.

Repentance is this story that it started with the fact that at some point, the political elite is not just anyone, and the people in power realized that for Germany strategically, it is important to talk about the horrors of what the fascists were forced to admit their guilt for this.

And how it would be better if they took it in money, what is called.

A it was a long internal process.

And if when you start reading about history, there is a great movie called Labyrinth silence about how the Germans found out about it.

They found out what was happening in the concentration camps.

This is not in the forty-fifth and not in the sixth.

I think they found out about it at the end of the fifties when quite young the prosecutor in Bonn decided to make himself a career.

Well, you can say it cynically, tell this story.

Yes, here it is met the prisoner cons lager that him I told him, and he was like, "Oh my God, really, this was happening?"

And he was young; he was like, so to speak, he was a person who was there during Hitler's time.

The maximum he could do was not, I know that it is not possible to participate in the Hitler Youth.

The fact that he reached his age, that is, it's like a blank slate.

With this blank leaf, he can put on a white coat and to come out means to tell more and more.

He moves a huge number of documents, found hundreds of testimonies, and he did.

He sought to make this process public; that this process was covered by journalists.

It was mainly covered by young journalists who also sort of built their own on this name and built recognition.

On the one hand, but on the other hand, he was able to really prove that a number of camp guards exceeded their official duties.

That is, it was not such a process that let's brand it.

Then we have documentary evidence and confirmation of witnesses who can prove that they weren't told to do that.

That the answer is, "I was just following orders," not it can be done.

And, roughly speaking, first of all, it is made the generation that turned out to be this not tainted in many ways.

Because later there, back in the sixties, and then later student protests, that is rude saying this, the children at some point decided we need to take responsibility for our fathers on yourself or vice versa to get rid of burdens.

And these are related cabbage soup.

[music]

[applause]

[music]

Tell me where we are now because that we changed the visuals a bit.

Yes, it became a little while later; we came to the lecture hall.

Here they give lectures to students.

We came in the lecture hall of my faculty.

So I'm in charge of the artificial intelligence center.

The majority of people have intelligence professors of our center; they are professors of the Faculty of Computer Science.

And the faculty of business informatics has its own.

The building at the faculty has several different Bachelor's and Master's programs.

A u our center has a separate building on the other in another part of the city.

And there, accordingly, we are engaged in science.

And in our classes are mainly held there, Master's degree.

And this is where classes are held; there are other bachelors and masters there programs.

That is here, there I have a bachelor's degree in computer science, bachelor digital society, which it's called at the junction sociology and numbers.

That's their occupation in short pass.

Here, what are you reading?

I read introduction to machine learning.

I am reading entrepreneurship for engineers.

I read Natural Language Processing, semantic data.

I taught a course on financial mathematics, but he didn't go in.

I decided not to read it anymore, but it's possible I'll read some demon such as financial literacy and 101.

Well, well, just for everyone who wants it because in the bachelor's degree especially there is little such courses that you know applied finance for life.

And perhaps such a course will be optional; it's also useful.

I read here management innovations for undergraduate studies.

As a general rule, companies are introducing artificial intelligence.

We read this together with another professor; he has this course in which he talks about two blocks: management itself.

And I'm telling about the implementation of artificial intelligence technology on the example of a set of cases.

Many people go to lectures.

Well, it depends from programs; on average, there are from 20 to p a person if some small.

And I'm currently doing a small course about computational creativity.

Right there, few people go.

What is this?

Well, I'm not doing it; other generative models, these are models which can generate something, create.

Yes, as if the inevitable question, how is this related to creativity?

Yes, that is, there is how much the model can do something new, not new.

How to characterize it?

Well, and plus now there are a lot of arts intelligence.

And in general, technologies are starting to grow into art.

In general, as if they always grow, but how do we do it all the time?

They say that this is a new history.

Yes, it's not like technology is art always influenced.

Yes, the Italians have such things: bright colors because they had them paints.

And the Dutch don't have such paints; that's why they had everything muted.

And they didn't have access to such paints, ultramarine, which the Italians had.

And here is the direct impact of technology on arts.

And even now, there are all sorts of things; there are these exhibitions like Art and Science.

Very many, well, and all this is interesting.

Tell students, well, you've always been.

I understand that he is an adherent of such interweaving art.

I remember that you had you generate there is Neuron.

Yes, summer album, which I so I understand that someone who knew Letov.

I actually decided that it was original some text.

Well, no, there was differently; there was Letov's brother who said that we are graphomaniacs, that we are not generating nothing.

And we wrote it by hand; they wrote it by hand.

But like, here we are graphomaniacs.

We, and we at that moment decided that we passed the Turing test in the sense what if brother for this thinks that it is the man wrote such poems.

I try to copy for this; then it means ours, the neural network passed the Turing test.

And we are like, I would like to see such a specialist clearly.

Somehow we were able to achieve Egor's creativity mislead.

Listen, if talk about what is actually represents artificial, I understand that there is intelligence in general several first components.

Secondly, definitions, and it seems to me that we speaking about artificial.

We don't always give ourselves accurate intelligence.

The report is that we are specifically now we mean come on.

That's a good question, so let me try to do it as briefly as possible.

The first one means there is machine learning.

Machine learning - this is clear that this is a section of computer science or is there data science on how to do it algorithms that improve quality his work with the help of some feedback algorithms.

What is Cast work is usually some kind of function which can be measured using the data.

Here, if it is admissible, it is called a function loss or error function.

That is literally did the task and looked how many times did you make a mistake.

Then you got this feedback; somehow aggregate it.

We changed the algorithm a little bit; the problem again, if you made a mistake in the smaller one number of times means we are something we learned.

And that's what machine learning is, a seemingly understandable area.

And people who are involved in computer science in general, it is much more comfortable to use a combination machine learning than artificial intelligence.

Because the phrase artificial intelligence, it appeared in the sixties.

There was such a comrade, Marty, who made the tongue lisp programming and other things.

At some point, I gathered a seminar in the city of Dartmouth in the States.

And they there, they asked the gift for a little money, such an American defense agency.

Give us some money, and there will be people on Mars.

We are now literally taking the apple trees away; summer, we will come up with everything artificial intelligence.

And actually, either in the application or did they learn from this seminar?

Produced this; I liked the phrase it so much.

Journalists and science fiction writers that it took off.

How could such an effect happen, really?

And now this word is used to call everything.

Whatever you really have there is a machine learning algorithm.

Before that, what do you have?

I just don't know the set a system with a set of rules.

She is from the outside looks like she's doing some smart stuff tasks.

Well, I don't know, you enter a card there; she understands something about you there and there tells you to go right or left.

Well, however, it may just be a set the rule that says here are those to the right and these to the left.

But someone in order to increase the round rating says we have artificial intelligence, and it's normal; it works.

Yes, yes, fashion.

That's what it means inside this artificial intelligence or machine learning, as you like.

There is a separate section called deep learning or learning artificial neural networks.

They are a separate type of algorithms that inspired by fragmentary knowledge of how our brain works.

Artificial neural networks are such type of algorithms within this family machine learning.

Among these neural networks, there are, for example, generative ones.

And here big, there is a big topic there related to CH GPT and what is happening now.

Everyone says this is LLM, large language models.

Large language models in Russian; it's called BM.

Good sa good abbreviation.

BM rarely use it in Russian; I think it's very funny.

We, we are doing BYA mommy.

Well, such a sweet word here, T tyam.

Yes, here, and here are big language models are artificial neural networks with billions of parameters trained on everything possible.

Well, if we are talking about leading commercial solutions like there Open or Anopic, they are literally that's all we found in digital.

They tried to use it for learning big breakthrough and big idea.

Behind all these algorithms, she is in that they can be taught without marking.

Here, hosting data is always expensive, time-consuming, and it's difficult.

But when you have a real one, the text written by a person is real written by a human, it's cool in some way.

At that moment, people realized that if part close this text and tell the model, "Restore what I closed," and then compare what the model answers with what is in reality is written.

That you are not as if you just need to place the data with you; there is data that is created historically by people.

But you can teach them model to reproduce human text.

It is a big interesting idea because it is the way all the data is on the internet, including the code of web pages and ending with some chats.

That is, about there's a ton of data, but it turned out to be all of it equally little.

We have now reached the border when all the data we have already not enough to continue scale networks.

But one way or another, it was a big qualitative breakthrough.

And in combination with a certain architecture, this very uneven network architecture is called a transformer.

In combination with that, we came up with a way to teach these the most transformers on the data that are not they demand.

Well, in fact, they do not demand preparations, but not as fierce as I would like it as it was required previously.

On one side, there was this one architecture which is pretty good on this data is learned; it gave this exponential breakthrough in quality.

And then the last cherry on the cake was what turned out to be simple: restore next word.

This is cool, but not big enough the number of applied tasks.

And most importantly, not enough to convince people that there really are a lot of such a model can solve intellectual problems unconvincing.

Because, well, people don't understand how to implement it.

Well, predicts what the next word is predicts.

Well, why?

How would I do this?

In reality, it is necessary in most cases; you need something else from the model.

The last cherry on the cake was that it's called human-centered learning reinforcement.

When from such a big one models who can do closed words restore to training transform such a model in a chatbot who understands which answer will you like to a person.

Well, that is, imagine you are just a model who knows how to next word e, and you have a request.

Some user, you write to him, you, yes, yes.

You write him something, and here he is actually answers you.

But this one, the process is one more step before the model is twisted before the chat model.

It's also like some a big deal and such complicated magic because it is not clear how to do these correctly.

Data collect how to prepare them.

How make the model more, I don't know what's more pleasant about the person there communication.

Here are people saying I'm with I talk to Claude Antro, and he is more pleasant.

It's not clear what the person means, but there are many people who like me.

Chat, I don't like GPT, but I like clot.

Why?

I don't know.

And vice versa, there are those who are there; they say, "But for me, it's the opposite."

Chat GPT, I like it.

That is, it seems that it is some of them are already so subtle personal preferences that can be obtained only by collecting a large amount of data from people.

And here they did Open first GPT-1, GPT-2, GPT-3.

And even when they made GPT-3, it wasn't big an event for most people in the world.

Because there was no such chat version, and there was no opportunity to touch it with my hands and be surprised how much of everything the model trained in such a seemingly simple way.

A set of ideas suddenly, not at a simple level can do.

Here's what happens next when they made Chat GPT; it was just like that a breakthrough for a lot of people.

Because they are so wow, and models can you can ask this way and that way and that way.

Model don't know generate recipe dishes from the products that I have in the refrigerator.

Can I ask her?

Imagine that she is Lincoln's Aram; tell me about the origin of the Northern War south.

And it was for everyone, for the big number of people was surprising.

But the hypothesis that it is simply scalable, you can come to this point.

She is there before this GPT was published.

She was gaining popularity among, say, so people who are engaged in science.

And well, that's where the same Anrok did the same thing, bet, and in parallel somehow with Pi ran to this there through with a little delay.

I understand that there is also a certain theory of the weak artificial intelligence and strong artificial intelligence.

That is, this some looks ahead, yes.

And then there is talk about the limits of this artificial intelligence.

That is, we can generate something that will be ours?

Just like a tool or something that will be above us?

Oh, something load zero you.

Yes, it's a long and complicated conversation.

A so firstly, it's strong again; artificial intelligence is such an idea.

Here is the definition: intelligence which in any by any parameters happens surpasses human here in any area.

That means we can do it, apply it, and it works.

Everything would be fine, well, with this definition if we understood what human is intelligence and how to measure it.

It was awesome; there would be a definition here.

But the problem is that I am convinced that our modern ideas about cognitive in the sciences of brain biology.

Here's an example: yourself, how you perceive what you see when you hear the word Alchemist.

Here you are; you see some kind of person who mixes some questionable test tubes.

Something exploded there; he is sitting in the Tower; he has a cap, but it is poorly washed.

Most likely, he has some kind of divorces there, copper sulfate on face.

I am absolutely sure that, well, it is necessary to understand that there 300-400 years ago, it was a scientist at the forefront of science.

Simply science moves so that in retrospect, well, he looks like a homeless person.

And most likely knows about the same amount about science how many homeless people are at the station.

Yeah, right; it's just us.

Well, how would we?

We try as view, and in this sense, when we talk about intelligence, we so happened that computer science has moved in this direction, digging and digging seems much further.

And to some places where people can reach who are engaged in cognitive science.

And it seems like biology, well, they haven't gotten to the bottom of it yet.

Not yet, yes, maybe.

Well, maybe it is other places; maybe it's not the same intelligence which is ours.

Maybe it is at all some other one; maybe there are several of them.

We don't know, but we don't have the language to say this.

To say because we are talking about our own intelligence is quite poor; we understand everything.

Yes, that is, cognitive sciences as sciences, as I understand that it worked out.

That's what it is, which in the century together with the wife writes that now we will make a computational engine.

She says sooner or later will write music and compose for us poems.

She does it 200 years ago.

People who are cognitive scientists in this time, as if we were about this time found out that if you shock a frog, then she twitches.

And then there is Nau Vanovo own internal; we know, but in scientific terms, well, explain what it is, how it works.

How we can't compare this, so when we say strong artificial intelligence is like some kind of philosophical such a concept which, well, in as a comparison, bad doesn't work well.

And that's why there's talk about the strong and the weak in artificial intelligence; they are difficult to lead.

And you tell me, what is weak?

Weak artificial intelligence, accordingly, they call it when the situation is when you have is there some kind of model or algorithm, a program that solves problems.

Well, you have a certain type of clear quality metrics; one task, and in this task, let's say, the algorithm has surpassed a person.

Yeah, that's a lot of GPT chats, roughly speaking.

Well, not necessarily GPT chat.

Well, you can, well, here, according to the photograph, I don't know, I need fluorography to diagnose cancer or not cancer.

There, comparatively simple binary classification.

Yes, on many tasks, let's say radiology, artificial intelligence shows better results than the average doctors already now.

And or there, well, I don't know, spam is not spam.

Here you have a spam filter.

This is an example of a narrow or weak artificial intelligence.

It is good; it can filter spam from non-spam.

Well, or bad, I don't know, depends on that what is your algorithm good, a specific task that is formed outside.

And the algorithm that with the help of feedback on your own mistakes, how would I learn to do it well.

This feedback can be people; it may be provided by some other algorithm.

It's not that important.

Well, in short, is there some kind of feedback?

Maybe just provided by the environment.

I love to give an example of an IQ test as very bad intelligence metrics.

And the main one, the problem with the IQ test is that everything for some reason they think that it was invented for in order to measure who is smarter in the top quantiles.

Were made so that take those with an IQ of less than 80 into the army USA.

Yes, because they set the task.

They realized that there was a certain amount of conscripts from whom in the army, even with if you give them orders, there will be more harm.

What's the use?

And somehow we should not take them because, well, they might shoot you.

Someone in this sense, test IQ is good; it works and reproduces.

But about this part of the IQ distribution, for some reason, we even somehow it's not even accepted to talk.

That's what he's measuring there; why is that?

Another question, but I, this is one of the most reproducible intelligence tests.

To it didn't mean as if people, and at the same time, he is quite bad.

He is very noisy; it's noisy in the sense that if you and I, we will take the IQ test now, and then, I don't know, in six months, then the probability is that our positions relative to each other, they will change quite a lot.

There, I don't know, 30%.

Well, plus, you should not be tired in general.

Yes, it is like very noisy, very bad line, essentially like a logic puzzle.

Well, that's it, very noisy, bad line.

And further, the question is, does logic intelligence come down to logic?

Yes, probably not, but there is empathy.

Is empathy intelligence or not intelligence?

And how does it work?

Is it possible to tell a person, I don't know, the girl says something like that about her feelings, and he starts to offer her practical solutions?

It is possible whether to call him a smart man or not, or not.

Well, that's a difficult question, I guess.

Sometimes I behave like this; a person tells me something.

He tells me; he complains about life, and I instead of saying, "Buddy, I did this, I understand you, old man, everything is fine, everything will be, hold on, let's go have some tea," just tell me more.

I'm starting something, the solution to offer.

90% of people, in my opinion, make this mistake.

Yes, well, that's it; stupidity, man, this moment ate man.

Sometimes acting stupid doesn't mean that he is generally stupid, right?

And probably too, you can say about the model.

Yes, well, you know, I generally think that one of greatly underestimated qualities artificial intelligence - that's what it is.

And work with artificial intelligence systems, intelligence modern, especially here, language models should inspire.

We will say with some sense of modesty.

So, well, that is to say we have some the basic idea is that we are the only ones here, smartest species on the planet.

Relatively simple algorithm, a design that works well with many tasks that we thought intellectual.

Probably, well, sort of a way to think about how we are actually what do we think intellectual.

Why?

How so it turned out that this is the reason?

Well, I don’t know; there is a word in English humility or there to be humble.

Yes, sort of, I don't know how humble I am.

That's it; there is this, well, humility is a bad word because that humility implies that you I resigned myself to the fact that something was given.

And you, it's not modesty; that's what it is, some more modest attitude intelligence.

I think it's good, and the first thing you should learn from seeing this, what is GPT chat is what?

Well, no, we still need to remember that we, after all, they are territorial monkeys, quite aggressive ones that have cool signal bioacoustic.

Surpasses some number of people at some quantity tasks.

This is actually a surprising fact which, well, well, it just doesn't matter to me.

It stops striking by itself.

I don't know; it seems to me, first of all, inspire because look, we now will get much more.

We have a chance for intellectual freedom; we may not be able to cope, but we have a chance for more intellectual freedom.

Because we haven't grown there for the last 50 years, white-collar productivity of that what is called.

That is, people intellectual work so that it does not mean.

And now we have models who will be able to achieve this productivity grow up to this.

When we grew up physical ru performance, yes, it was associated with enormous growth, well-being, life expectancy.

And so now, first of all, there is hope that it will be the same.

Secondly, there is a big problem associated with the fact that people who spoke wrote all sorts of things, creepy books like Population Bomb once again, they were wrong.

Well, now it seems like all demographers are saying that the peak of the planet's population will be achieved in the 21st century.

And then it will be to decrease.

And it is not at all obvious that will happen to the planet in which the population is declining, especially if, for example, it will be a democratic planet.

Because when you have the majority of voters are more worried about availability, don't know, houses, elderly, and hospices than kindergartens, suddenly it turns out the birth rate will be, of course, of course.

And when does it go down for you?

Birth rate and fewer workers.

Who will pay for those same hospices and homes elderly and what?

And who will actually be think with your head, do science?

We have quite a lot of unsolved problems there.

Here we want to become, maybe we will someday wanted.

Let's do it like that; if we once wanted to become a paid viewer with us right now, there is no ambition left there.

It's like that; there are a lot of things related to biology there, how to build closed ecosystems so that everything works well and so on.

All this requires intelligence.

And we have there will probably be a maximum of 10 billion heads, and then less.

Here are some of the ones that were given; what was given is what will be.

[music]

Listen, you studied at St. Petersburg State University.

How much?

Yamburg State University.

I just listened to the Czech faculty, a couple of clips performed by you with excellent criticism from the rector of St. Petersburg State University, who is now the master of rollback.

Rollback associate professor; I am a Master of rollback.

Master, I was a rollback associate professor like this.

Lyrics of a beautiful song performed by you and authorship.

Yes, I understand; well, it was collective creativity, but it was me who sang.

It was you, the Master of Recoil, sang to the Professors' Team.

Well, I sang this to the audience, professors, professors, and students at the same time.

And for you participated together as if professors also spoke satirically in some way about what's happening at the university.

Just they did it more subtly because they were smarter, and you, so to speak, the truth ma.

We, we as if in Nekrasov's style, you know how without these references and nuances.

Yes, they say you almost expelled for, they say, but they didn't expel me thanks to the dean.

They say thank him very much for this to the dean of the faculty, which one?

Alexander Sergeevich Tsov, who actually it was for a long time leader of a team of professionals, the head of our faculty.

Here, a wonderful person, and I have lectures I once read.

Well, they say, yes, what to the rector?

I didn't like that I respond to the fact that in the city of Petergov supposedly there is a science city, but it is not clear where it is.

And where did all the money go for it then?

There was a big one; then Shnurov made a song Shishkinsky forest.

How was it?

Shishkinsky, what was it?

Khimki forest, like.

And we made a song; there was a blue fence there.

Chorus: blue fence, apply for grants.

I am the source of all innovations.

In the process of us singing it, there some were playing out in the background.

There were people sitting there with an inscription; there, I don't know, advice or something like that.

The students pumped them up; it became thicker and thicker.

Rise up under them; the balls swelled up.

Oh, well, I had some kind of enough.

Ilichevskiy God of war; they shoot unsuitable.

What are we doing now?

Satirically, so to speak, we will point out disadvantages, and they will be fixed that will not be.

Here, look, you are the same said that the southwestern district is voting and chose to do PhD in Germany.

So to speak, what remained unclear to me?

Thirteenth, fourteenth, fifteenth; you are spending your sixteenth year in Germany.

The fourteenth year was a turning point point for many, but you on the contrary decide to come back after Germany.

Well, not in the fourteenth; I decided to come back when COVID happened.

I returned to Russia in the nineteenth year.

I decided to return because COVID happened in 2020, more precisely.

There, as it were, I decided to come back in 2010 when I was actually COVID.

There were lockdowns here.

Yeah, I'm like, "Damn, I came to a free place."

Country, a somehow it's not free.

A I see.

And in Russia, everything is open, and at this moment, Yandex's proposal arose to make a lab for the tower.

And I thought, "Well, okay, I guess."

Well, first of all, I always have, I had this idea since childhood.

Dad said about this that this is the right one career is arranged, so it's good.

There described in British books in the British books of the century, you give birth in, I don't know, about some kind of grav sura, poor, poor house.

No, poor house because you were born in a poor house in grav sura, and you do nothing.

Are you going somewhere there in India or I don't know what's where in Bengal?

Or maybe you need to go to France.

Well, anyway, you somehow travel around the world and learn something.

You get some valuable skills, capital, then you go back to the column and buy there is a house, and you live there and bring benefit to the grave that sura.

Yes, as if it is the correct trajectory for the poor heir good family.

Well, even I would say you were sent a track like this.

Tell to study in Germany so that there get all the knowledge and then take it back to your homeland.

Bring, well, it's not an illusion; that's what I thought that this is right.

I still do it; I think so.

That is, in general, I think that it's quite valuable when people can study somewhere where they teach something differently.

And well, and then these knowledge is returned to the country, and from this, the country didn't seem better to you.

That already in that moment came.

I, this is a question, and I'll say right away that it's not a provocation.

Didn't it seem to you that the country already some irreversible processes are taking place?

Or were you under some kind of illusion that everything is possible somehow?

Well, yes, that's where it is somewhere here is Marco besi.

And there is somewhere Svetloe; it's a great movie that it's called Pray about how LGBT the London community decided during the ter support the miners on a real basis events.

Because the LGBT community in London decided that if Cher is against us and against the miners, so we are for the miners.

And they began to collect humanitarian aid because the mines were closed by natural people sat and starved.

Well, yes, there were serious strikes and protests; they became bring food to the miners so that the miners we ate something.

And as they say, legend has it that the first Pride in London was not agreed, but they didn't disperse it because the tough guys came from Newcastle and defended the pride.

Well, that is normal fight in which always, well, the government never wants to give in.

In power, the fact is that in principle society works when you do.

You think that it is a society that we would like to be in this together.

This society of ours was for me; I honestly don't understand why some things which are not happening there in Russia do it differently.

Based on what is wrong, what does it mean people with bright faces?

And there are other others.

Well, because, well, this is our country; we are all together; we want it to be better.

And yes, there are people who think otherwise; criminals, well, they exist in any country.

Well, somehow we have to live with this, learn and work with it.

I like it; it seemed important on the one hand.

On the other hand, I saw that objectively, with all that, there are a lot of problems there.

And a completely objective trend that a lot of things are getting better.

Grow up as if the next generation, people who behave more freely, who build business more actively, they do some local stuff.

That is, I started making a podcast a long time ago.

I make a podcast and talk to people.

Among other things, I speak in mainly with people who either business, either do science or do something public affairs, something socially useful.

And they do it everywhere; they do it there.

That is, I don't know, I was there series when I went to some science festival in the regions of Russia.

I always made episodes about these cities, and in every city I visited, there was something cool of my own where people are just themselves.

They took it, got together, and did it; that's all right, good things.

Well, and me, it seemed that I would bring pain and benefit if I will be a part of the society I was born into.

And I can bring some of my own competencies in order to, that is, Russia of iPhone vs Russia of Chanson.

So no need, well, that's not right; that's the problem.

There's nothing against it, firstly, because chanson is listened to on iPhone in full growth.

Well, firstly, and secondly, because there is nothing wrong with chanson.

There is nothing; that's what's wrong with the sedentary bandit theory tells us what is needed.

Stop pretending that these are two different things.

Russia Gan RP also seem to be singing like saints in American society; this does not interfere with American society to be a democracy.

Yeah, well, it's like the most popular style there; music - these are people who literally they talk about how they shoot at on the streets.

It's such a chanson.

Well, just musically, it is designed differently, but the narrative there, it is absolutely from chanson.

And nothing special; there is nothing scary or bad about this amazing.

And like, well, these are good understandable songs.

Moreover, if there think about the World Cup, Russian chanson music is mainly the plot of injustice.

It's basically that the very plot that worries you.

And okay, you may like some other music; you may like some other songs.

But I'm completely, I'm sure that if you're serious about it, I don't know what to do with electoral politics, for example.

Then you, well, here there is a minister president of Bavaria.

He about once a month he eats brel camera that he loves so much.

So pretzels, maybe yes, maybe no, but this is it doesn't matter.

He understands that participation in electoral politics, it's just that he's a flexible political animal eating bre.

It's like part of the job, you know.

He wears these leather leggings every October and drinks a liter of beer.

And he probably doesn't like these in these shorts.

And in principle, he probably, well, how I would like to skip it from time to time.

But it's part of the job.

He doesn't say that I cool, I know how to develop Germany.

And you don't understand anything; I tell you, I'll explain everything now.

Well, it's like electorally has no chance.

Well, that's it, of course, definitely a politician's talent talk to everyone like that.

No, here, here, well, as if that is, I am just a stop.

The question you voiced is like this: Russia, chansons of Russia, iPhones.

I hear from very many people who are in one or another would like Russia to be in a different form.

Let's do it this way with electoral democracy.

I'll formulate it broadly, so here's the first step to so that it would be electoral democracy.

Needs to understand that if people I like chanson, I need to learn it somehow as if to be a part of oneself.

People as if it was hard and painful, unpleasant, maybe, but somehow, or you part of our people.

And then we have electoral democracy, or you always, even if we have an electoral democracy, you will think that somehow voted wrong.

Here's a good one example, by the way, what's in America now it happens.

Yes, look at the reaction to Trump's election, so to speak, progressive public, which relatively speaking, for freedom, the right of people vote as they think and so on.

Well, there are just people there who are on the same page can't stand Trump.

Well, of course, well, for the cause, but, but, but, but, but just democracy, it works that way when the dude wins which you don't like.

Anyway, agree.

Listen, well, yes, but they don't go to the Capitol, and no matter how they storm it.

They put up with it for 4 years and don't give a damn.

They go to demonstrations, but like what do?

[music]

If we talk about the future, after all, development of improvement of these models, where resource for improvement?

That is, if the data is not a resource, that is, the idea was something more.

If you feed the maximum amount of data, these models, they will switch to some quality level.

There are several hypotheses; come on, let's start with the simplest one.

There is a series of articles Microsoft is just about their models, which is called.

So there is an article in which the mechanism of the Transformer is offered.

It's called Attention; that's all you need.

Need this is such a meme already because the article has been there thousands of times already quoted; everyone knows her.

And here graduating from UN means such a thing already, such bad manners in the field of artificial intelligence.

And among other things, there is a textbook Crof in which they say, "Let's do it; let's take a small model, small, but we'll teach her using textbooks that are straight good quality and also special are designed for training."

And we will show that a small model gives results comparable to the big model because that we have very good data on it taught correctly.

This is approximately the same thing that we do in Play.

We do there; we show that the models are there in three to four, five times less than comparable some open models with which we compare on those tasks on which we teach these models.

They show results are better or comparable.

Data quality, data preparation, noise filtration, garbage filtration, synthetic data generation in the right way to complement your training data.

Here is a funny example; we taught the model on literary texts, among other things, which came out from under copyright by the statute of limitations.

That is, it is learn the classics.

Or there is another example; our model speaks Latin well because Latin is almost entirely without copyright.

Here, and not on it usually somehow they don't teach; somehow it's not accepted.

And we decided it was cool; we still have a little bit sweat laid out SPK LM, which means this is just so perfect, so small little Latin model too cool side effect.

Overall quality data is the first big history.

The second big story is what says Dario Asmodeus is SEO Anthro.

He says that they don't see a problem with scaling.

Maybe because Anthro has some data which is not open or some approaches to teaching that do not exist open.

Maybe Dario is just lying.

A maybe they have a smaller model for now; they don't have it.

Tom in the joke, and you say the same thing.

There is a third big story, then about the fact that synthetics are in a wide range.

In the sense that synthetic data is very don't do in area.

If we understand that the model can perform some tasks better than a person, then you can do a lot there.

Synthetics nage neri is there interesting; I'm on the subject of the cedar dig further.

Yes, of course, of course, of course, using Self Driving as an example.

Interesting example.

Guys, I know a few startups who make environments for simulation cities to learn self-drawin there.

What?

They just can't come up with anything there, apparently environment, of course.

And there, first of all, in digital Wednesday, you can come up with anything you want.

Secondly, you do everything on the engine; there, I don't know, let's say Unreal Engine or something like that.

That is, you are just a mile eat like levels for a computer game.

There are some artificial intelligences there simulates human behavior, and you spin handles so that they are even more crazy.

People made the country; that is, you can do it there to invent more than what happens in life.

The main thing is cheaper; you can collect some mega, just megatons of data.

I just ride and learn this faster car.

And then I would like to have such an interesting example on hand fingers.

It's clear that everything is happening now; they are digging.

They are exploring these possibilities.

Well, yes, here is a story about what, how, how collect synthetics.

Interesting synthetics, how to cook it and use it now.

Interesting cutting edge; another one, the big history is about that some appear now.

There are visits to some other ones architecture transformer quite a long time ago.

It exists as a transformer model as if the only architecture that what scale can we possibly be?

Let's come up with some other architecture which will scale even better.

Yes, that is, the architecture of the neural network itself, how it is structured in a broad sense.

We are still relying on the idea proposed there 7 years ago in the article Attention All UN need.

Now lately, some results that show that other architectures give similar results.

But it may turn out that so questions related to scaling they somehow specifically connected with this architecture.

And is there any architecture which scales better than some other architecture?

Can you explain for Humanities?

Architecture, analogue architecture is what like neurons connected inside Neuro C.

That's how it is, roughly speaking, where are these grasses?

Yes, here analogy, but better.

Well, yes, here you have it; sequence of how data pass through the neural network.

How it works flow, I don't know, the information in it, if whatever you like.

And what operations are not currently available?

There are an infinite number of such, yes, as many retics as you like, just some kind of chosen as optimal.

Until today days, well, look, she's not like that optimal.

We just know empirically what is this, studying.

Well, yes, this studies well.

Why is it so good?

Learns, firstly, because we teach in a certain way.

We teach with the help of an algorithm which is called algorithm backpropagation of error.

And this is, by the way, some Russian came up with it.

Well, there, like Soviet, yes, but not quite.

Well, let's do it now; let's not bury it any further.

In short, they came up with it there and then improved it.

And Soviet scientists and American scientists, because almost everything was invented in the 20th century at the same time.

Soviet and American scientists sometimes with a lag of several decades because the information is bad spread among themselves.

Yes, especially secret; there is such a great meme that all science was invented in Soviet magazines in the sixties.

A we all live in a simulation because science would end.

And in principle, if you do you want to make a scientific career?

Just find a file of some Soviet magazines for the sixties, translate there in Chinese.

I will publish in Chinese in magazines; you will be a Chinese star.

Yes, or there in English.

Well, sort of like there really are some pieces of Soviet science which completely died together with Soviet science.

This is generally we must understand that it seems to us that science is so straightforward sequence of ideas when we took ideas of Lida.

But in reality, science is this a tree that has a bunch of branches who just died because it worked out.

Yes, there is definitely someone there; he came up with something; then he just didn't have it.

There were students; something happened, and he like we wouldn't know anything about him.

Much m in science, yes.

That is, I have a favorite example; Evariste Galois was a great French mathematician.

Ivari Lou, I know this; he was killed in a duel precisely because I think he was killed when he was 20.

But he was in a duel; he is great because firstly he was a republican.

And at that time, it means in France, royalists means the republicans were divided.

And he believed that his articles are not published because one of the French mathematicians, who is he for?

In my opinion, to whom he gave for review your article with Bolzano porridge aubrey.

Yes, yes, yeah, something like that.

That means he was a piece of cake royalists and did not want to publish it.

Beautiful, perfect character, yes, died at a duel at the age of 20.

It turned out that the mathematics he invented is strong, useful when you need to calculate aerodynamics of wings and in demand.

O remained 100 years after his death.

How many people died, and as if their drawings were not preserved?

Well, we just don't know, yes.

There is another example from my scientific Jurgen's supervisor was a graduate student, which there are several articles.

They published on the archive; archive - this is it, archive.org is a site that you can post any scientific work unreviewed.

You just want it put it out to kind of indicate that you the first one got this result.

That is, you can post any article there, but not the fact that this article is correct.

There may be articles with errors; they may be some kind of pseudoscience.

Well, anything can happen, yes; it's an open system.

And this one has the person; he has several articles in the amount is about two hundred pages of such neat mathematics.

And not a single journal; these articles have not been published in the last 20 years.

He eventually stopped publishing; he went to study science, well, like some kind of industrial things.

Why?

Because none of them are adequate mathematician not ready free review an article of about 70 pages in the magazine.

You that they've gone crazy about something that they understand like three people on the planet.

Dude says that he solved some problems, but as if, yes, is this so or not?

And are needed whether they will be with us or not, we will find out obviously in the next series.

And in science, there is a lot of this.

And well, here comes the artificial one intelligence.

This is just artificial intelligence in this sense.

This is me; I lead it's a very long, long, long segway to another interesting thing that is possible.

Here are some of these ideas; if we know where we have some stumps, there may be something interesting.

It is interesting as if direction, yes.

And here it is as if here additional brain power.

It's useful, but I tell my students that sciences are theoretical, experimental, and mathematics long time was only theoretical.

Yes, you here on the tip of the pen, I found something there.

Here is computer science; you can, well, me, it's probably easier to perceive this way that I am a physicist by education.

Yes, I am; I perceive computer science as I threw in some experimental mathematics numbers to so-so.

Here's what I looked at; I looked at what fell out.

This is what fell out interesting.

So let's write it down like this; throw in such and such a number here; it falls out.

Think about this, dear theorists, what about to do this further?

In a sense, this approach turned out to be extremely productive from an applied point of view.

Because we have there, I don't know, chat GPT, but we still don't have an answer.

Why should it work?

Here is a good one example between experimental science and theoretical science.

And most importantly, that other sciences always fall apart like that.

That is, in physics, this is normal absolute history, or in chemistry, you first synthesized something fist some, and then you try to understand what kind of its properties and what it is in general, why it is needed.

Yes, or maybe I put a prism there.

How is this a great meme that Isaac Newton invents homosexuality?

Well, sort of, so here I put the prism, and it decomposed into a rainbow.

So let's write down that why go figure.

Well, here's Prism Rainbow; here is what Experimental Science wrote.

And for many things we have done, we first set up an experiment and found out that it works.

And then sometimes there, after decades, sometimes even after a century, understood and it behaves like this because light is a wave.

And that means here, well, and with machine learning too, in a broad sense.

I mean, we're not there yet; that's what we are; we understand it very well as intelligence works.

And therefore we understand why those some architectures work, some don't; we can learn such architectures, no, and so on.

Listen, there are many; they say that there are powers today's is a certain barrier to breakthrough development artificial intelligence is so.

Yes, well, it seems like it's not that big.

The problem is if we have controlled thermonuclear reactions, there are definitely small problems.

What if we have them for some time?

It won't happen yet, but we will have enough the number of nuclear reactors.

It's also a small problem; in the States, many companies that are trying to solve exactly this.

This problem is being launched back by nuclear power plants.

That is already, that is, burn the barn, burn the hut.

That is, well, I mean already in pursuit of artificial intelligence, we'll build five nuclear power plants.

Well, five, firstly, it's not enough.

Secondly, I don't understand what's the problem.

Well, that is, if you look at the statistics on the number of human deaths per kilowatt of energy.

No, no, I don't think it's bad to build nuclear power plants.

I just mean the scale is impressive.

Well, wait, wait.

Well, we've always done it this way.

I don't understand what surprises you.

Well, how else?

If so, that's how you imagine it to myself.

What is the rocket on which Yuri flew into space?

Here is any Russian; a person must imagine it.

Well, I saw the picture.

Yes, how the car flies.

Yes, you, I thought that this rocket is this thing roughly similar to a five-story building.

Here, only to our five-story Khrushchev-era building put on its butt.

Here, imagine this is a five-story building that was built on the butt filled with kerosene.

Yes, and there is a compartment like this at the top, and in it, Yuri Gagarin is sitting.

Well, I mean, we are for the price is not worth it.

You understand this; it means a five-story building.

And Sergey Korolev says, "Come on," and he's on this one.

Here is kerosene at home; he says, "Let's go."

And a brave man is flying, by the way, absolutely no hero.

As if that's completely, this is a genuine reason to be proud.

Big story like a monkey flew into space - this is a big deal.

Well, how if this is really a big deal?

We already have made like, why are you surprised?

What in order to make artificial, we need to put intelligence, you know?

I don't, I'm losing the ability to be surprised.

Tell me, please, there is a lot of talk about this; there is a quantum computer.

There now Google reported that I actually already read it.

Some people and asked specifically, am I telling the truth or not?

They think that there might be some errors there.

So in a quantum computer, the main thing, the difficulty is that he can do a lot, solve many problems at once, many times more than usually.

So, and how much do we have time?

Guys, this is a quantum computer in a normal way.

The computer has a zero and a one in it; a quantum computer can have anything from zero to one.

Because quantum system, well, here you have it; there is infinity.

Yes, yes, there is infinity; one qubit can take any value from zero up to one.

It seems intuitive, pretty cool.

The problem is that we know very few algorithms for such cubes which work.

If strongly again simplify, there are some physical systems that, well, roughly speaking, there if you take the ball and let it go, it will fall to the bottom of some thing.

Yes, smart means to minimize the ball; he had his potential energy, some energy on the hill, and it rolled down.

His energy went into the hole, less times he minimizes something.

So this is an optimization problem; that is, we something that means optimized done less.

Yeah, they brought it to a minimum; they found it at least some of this is optimization.

Here, let's say machine learning is almost always the task of optimization.

We introduce error function or we optimize it like this so that there are fewer mistakes.

The less, all the better.

Yes, here is the optimization problem, among other things.

Some physical systems can very quickly decide because the physical system is so arranged.

What do you let go of the ball?

It falls to a minimum.

If the ball is at some point such a small molecular one, as it were properties, then you can somehow build the physical system that gives you parameters of the optimization problem which you want to solve.

And it's just laws; physicists will solve this very quickly.

Interesting; it's a way to sort of use molecules, roughly speaking, or there are such very microstructures for solving complex problems.

Optimization tasks that are several times faster than our algorithms approaches with such zeros and units.

The problem is that it is quantum computers that are built on these the most qubits are technically quite difficult to implement.

For example, because any noise in the system, there, I don't know, electromagnetic fields of some kind can greatly change the state of this qubit.

These noises need to be somehow something with to do them.

Yes, it is necessary somehow so that they were mutually exclusive.

And are still there a lot of games about how to do a lot of, well, we know what to do some cool algorithms.

We need solutions; one qubit is not enough; one bit is not enough; you need a lot.

And I would like to eat a lot of cubes.

Interesting idea about what if there are many of them.

How to place it correctly inside the chip, create the right conditions?

Then these very ones, noises will cancel each other out, and this whole situation will somehow be stabilize.

And there Google

Results, on the one hand, and with the introduction into the industry, on the other hand, the industry must also work with them, interact in any way.

What could be from the fact that here, let's say, it's a completely standard practice? What our master's students write for their diploma within the framework of work in one or another company is that they are given some clear applied task related to using artificial intelligence to do it.

I don't know, there with the founder of startup companies, or students do startup projects, or professors go there as early-stage advisors.

I don't know the situation when the professor somehow combines his position in industry and at the university.

Well, I don't know, here in technical schools it is written that once every five years I can take sabbatical - it's six months - and I can take the sabbatical on research.

Can I take my dog to work in the industry business to do for six months?

Lose qualifications so as not to be in separation from the industry itself because the idea of the technical school is that to become a professor, you need to have at least three years of work in the industry, besides the fact that you should have articles and a dissertation.

Because you prepare people, first of all, for that to work in the industry. You must be close to the industry, don't break away.

What were you doing at Skill Factory?

I understand there were Skill Factory and Yandex. Well, if it was from the last one, then yes, it was.

I came to Skill Factory with a recto strategy. We tried to make it so that the education we gave created added value for the student.

In the sense that, firstly, we tried launching a master’s program. It’s probably the best idea as if the essence of education.

Listen well, in tech there is a large category of tech that it's simply called Ibis.

Aha, there is a piece from Teha which is called Te Aga, and there are those who try to teach hard skills.

I tried to push to the side, and to me it seems that we have examples there.

There are examples of students with whom we succeeded, with which it didn't work out.

It's generally difficult. The most interesting thing I got out of there when I actually started working with the guys is when you listen to Kaz Deva.

People who came to study as IT specialists have absolutely amazing stories.

I don't know, there is a mother of two children, she is single, and she wants to become a tester from a regional city.

She says, "I don't want to work anymore as a policeman. I want to be a designer."

Like that, it happens too.

Yes, there is always a security guard nearby a store that says everything is cool, but it's very inconvenient to study on a mobile phone.

I don't have a computer.

Yeah, I do. Mobile. Can I have something on mobile? I'll learn.

When you talk to these people, first of all, I don't know a lot of things you find out.

Secondly, you see how much the education system is terribly organized.

It's scary, scary because it motivated people who want to do something with their life, and the education system can't do anything to suggest.

Some of them have already spent five years on higher education, years of my life to ensure that what?

Yes, that's a good question. It's a tragedy, and we all pay for it.

Yes, that is, unlike Skill Factory, where a person pays himself, or any tech at all where a person pays himself, and for this we all chip in.

This is, let's chip in.

Well, yeah, kind of not free education for everything.

Someone pays. In fact, they are like the first generation of people who received higher education, have already received a bachelor's degree, and so they came to our master's program.

When you work with these people, you inevitably have, well, first of all, it changes a little picture of the world.

Secondly, this is much, well, a completely different experience.

It's not such an experience when you work with the man who first got into a good mod circle.

Then it means in a good university.

Again, these people with bright faces, PT seventh school, conditionally.

I love all these people very much.

I have many friends from my math circle.

We all seem to be doing great, but the world doesn't end with a math circle, guys.

And very, it is important to keep this in mind.

First of all, secondly, it seems quite important to talk to these people and teach them something and make their lives better.

Because, in general, if we could, I don't know, if only we could do the same, I don't know, billions of people to teach who don't have access to good education now, it is obvious that we would all live better, longer, and richer.

Yes, in Yandex, you worked in addition to this one, the laboratory that you made in St. Petersburg for some reason because I'm from St. Petersburg.

Wow.

I wanted to go back.

I wanted to go home.

I didn't want to go back to Moscow.

Come back? What is this dream?

Come back to Moscow?

What Russian person laughs?

Return to Moscow except Moskvich?

To hell with them.

Also, not so little know us, so little.

I, as I would be a big fan of regional patriotism, among other things, I like that here each region has its own, each city has its own default beer.

That is like if you are here, you need to drink this beer, and that's it.

The rest, yes, and all the rest are like also not bad, but not as good as ours.

And this is the regional identity for me.

I always liked it very much.

That's why, actually, I wanted to go to St. Petersburg, come back.

And in Yandex, you were, besides this project, you were still working, I understand?

Then I worked in Toloka.

After that, how it actually started to happen, the division of the company, Toloka went here.

I helped Da Toloka, worked there with interaction with science itself, led the team that was engaged in all sorts of technical PR, interaction with scientists, interaction with the conference.

Can you explain in a few words what Toloka is doing?

It's a huge, great service for cooking data.

Mark as a dog is a dog, so let's have a word cook.

Good, here when is it possible?

Here, imagine that in the kitchen you can have many different things to do with products so that they turned into a dish.

In one word, the word is cook.

Well, that's Toloka in a broad sense.

This is a service that helps prepare data so that later people who are engaged in machine learning could use them for some of their own tasks.

This is also the markup of this very data if necessary for training with a teacher and some other things.

Exactly like that.

Well, here it is as if there is this famous saying, "New oil, data."

This is new, not already him 10 years ago.

This is a new expression which already 10 years.

Yes, but this expression is about the fact that any machine learning algorithms do not work if there is nothing to study.

Yeah, right.

Toloka chose the direction for itself in which they say, "Let's give something to learn from."

Because you yourself, these are the tasks of collecting markup data, yes, data quality control, data.

There are a million complex problems related to the topic.

Well, when you work with huge arrays of data, some things that seem very simple become complicated right away.

If we want to do it, I don't know, of a car that drives itself, and we need to give her a markup there, like traffic lights recognize what's here, such a light is on.

And here it is like this.

Oh, well, it would seem that let's give it to people.

Mark people will say where what light it's on fire, and we'll teach you how to check it.

What people told you the truth?

Well, maybe we need to cross it somehow so that people double-check.

Yes, and how, and what there, the confidence intervals should be.

What percentage of errors is acceptable then?

Because a person can simply curse just always a, a, a, of all possible options.

And it's simple.

Hired for now, some for now.

Yes, some people who do something for a little money there.

It's not only people; sometimes it's also models now already.

That is now it is some kind of human-machine complex which is generally rebuilt.

Great serious know-how, and actually, Lucky it is.

And I hope that the guys, they here, we entered this big one, NC, which is now headed by Arkady Vosh, and they disconnected Nadya.

In the end, we divided, let's say, one part collects data in Russia and the other part to the global world.

Yeah, and accordingly, that part of the company that remained behind Arkady, she is in Amsterdam.

Well, it's not just in Amsterdam; they have a lot of them offices all over the world, and this is one of them.

Pieces of the company you worked with them understand not yet while you were in Russia, but already when they undressed.

Yes, they left.

I left, and we decided that we needed to interact with each other.

Judging by the latest news from the field of competition for chats, GPT, relatively speaking, well, there are several competing firms, roughly speaking, which develop more or less competing products.

Are clear?

Date you buried.

Well, or is it a question of survival for these corporations?

This is OpenAI.

Yes, this is Google.

Next, continue.

Well, there are OpenAI, which have a long partnership history with Microsoft.

At the same time, Microsoft, its models also does.

They have a family of models which is called Fi.

Further, there is Amazon, who have a big investment in the company Anopic, which has models.

There are Lauda, Haku, there, Sunet, and so on.

And soon there will be the next generation models like A.

And then there is the company META, which has different models.

The META company suddenly changed expectations.

Suddenly, it turned out to be a huge player in the market of open artificial intelligence because at some point they just moved.

It means, and these are the weights, our model, Lama, take and explore.

You twist it yourself with your hands.

What made them do it?

Well, I don’t know, maybe it’s because Mark started surfing.

Maybe something in this direction.

I mean, no, but I'm joking.

Well, I mean, they started losing, and they decided in this way.

It seems that they fight for talent because they understand that they need talents.

A, and they believe in some things around artificial intelligence, but they don't understand directly where in the products directly in Lop it to use.

But at the same time, I want to be on the radar, so to speak.

What it's called in this place?

There, I don't think we should underestimate the role of personalities in history.

There works Yan Kun, one of these founding fathers of artificial intelligence, received the Turing Prize for Da Yan Kun.

He is in France, in my opinion.

There is one like this now, French-speaking, works at META, leads research in artificial intelligence, and quite systematically and for a long time promotes the idea that technology and science should be available to a wide range of researchers.

Further, the company Ral is a startup French which follows as if I started making open-top models of Orny.

They don't understand what will happen to them, but in, I don't know, many of them about resources and companies, models are used.

People like it just so small, comparatively a small, relatively young startup who is also trying to solve this situation.

Then move further to the side of openness because there are two positions related to the safety of artificial intelligence.

There are people who claim that artificial intelligence can be dangerous, and therefore it is very important that it does not was available to anyone, no matter who.

Well, there are people, not just people, like Lisa Austin village, and claim this.

I understand that these are people from the community of scientists who are working on the problem.

Yes, well, actually, everything is like this from communities of scientists, but also communities of businessmen.

Yes, that is, the company OpenAI, although it is called OpenAI, it has its own model; it doesn't open, but they have open ones.

That's just its name.

And among other things, it is motivated by the fact that it will be too much.

It's dangerous if anyone can use their models.

Although here, the META company opened its models.

Who uses it?

Randomly, and everything seems fine.

I mean, nothing has happened yet.

You're what?

Yes, yes, that is, and in this sense, to me, it seems that there are scientists, there are priests, there are priests.

So they keep secret knowledge, and scientists, they tell in any simply on each corner about how this thing works.

Well, it's like this.

This is different scientists from the priest, and important science somehow save.

And this is not me in the country of scientists.

I think that models should be open.

I think that people, well, first of all, if at some point you built some cool model, if it is really so cool that it gives you a huge competitive advantage, the advantage is huge competitive advantage.

It is like reassemble the world around you, but this ethically, it's quite strange because you made this discovery thanks to there thousands of years of human history.

You, the model was trained on all the data in total of humanity throughout its history.

You studied with some people for money, taxpayers, and now how would we let us declare you the emperor of the world because you have a private model?

Well, that's it.

Yes, if you take it off, Hollywood cinema influence, Indiana Jones, there will definitely be some.

Yes, that is, it seems to me that a scientist should be intellectually honest, and one must be intellectually honest to the end.

These people say we are worried about security.

We consider that to anyone, artificial intelligence cannot be trusted because he can be such that he, like, it will change the whole world.

Changer?

Wait, well then, next step, then it turns out that the whole world will you change it?

And who authorized you to do this?

Do what procedures can we do?

Somehow vote?

We are for you changed the world at once.

Questions if you are serious about this proposal.

Otherwise, then we need to somehow talk about how you ensure that you do everything that's right.

You just say we to whom we won't give it to you because you have it means that some faces are not enough bright ones, then I have questions for these people.

Yes, and on the other hand, people who do about Source, they say here is the model.

Well, here we are in a startup talking about this model.

Here is the code, here is the weight, then here are the data from, and to anyone who wants to laugh to burn a million euros on a computer can just go and play it straight.

Sorry, what's the difference then?

Why?

Well, how?

They say, "Will is not mine."

Analogies are drawn, for example, with a nuclear bomb.

Yes, there is, that's why the North can't now Korea take your model.

And well, because it's not a nuclear bomb.

Very simple answer because it's a bad analogy.

Because that it's not a weapon.

Yes, that is, there is still once that is, it is simply besides everything, it's much more complicated than that.

It's not about it's just a matter of what you have specific weapon.

Weapon sde weapon has one goal.

I'm not for that open models, I don't know what they do, they do missile guidance themselves.

I'm not for this.

This is an example of Russian artificial intelligence that makes one goal.

Means if we really believe that there is a wide artificial intelligence that does everything and the reduction, it also makes rockets better than everyone else the rest, then of course he should be open.

Because then yours, artificial North Korea will take artificial intelligence that will self-guided missiles or nuclear weapons bombs.

And we will take artificial intelligence that is fast and at the same time will protect us from that.

Of course, of course, and only openness guarantees the possibility of actual protection.

Because otherwise you get in.

That is, you find yourself in a situation where everyone else has access to technology, and everyone uses it based on their own moral and ethical ideas about what is good and what is bad.

In a situation when you are guaranteed to find yourself in the world in which there are some extra possibilities in a narrow circle of people, when in a narrow circle there are some circles of people, territorial superpowers, monkeys can be sure that they are good.

Don't expect anything from this breed of people, of course.

That is, you kind of change it, well, such a cognitive distortion, well studied when, yes, what, as if a terrible end or horror without end.

Yes, this is a cognitive distortion when you change people are generally bad at it.

The probabilities of interacting and very bad with small probabilities that is rare events - it's very difficult.

Even people who know theory well, probabilities are not good at it, predict rare events.

And in general, somehow imagine them in your head than one in a million event differs from one for a billion.

Well, it's difficult because some big numbers to imagine such small probabilities than they are qualitatively different complex.

And here, when we tell such hypothetical scenarios, what there really could be such an algorithm who is better than a person in all respects, parameters, people, they assess its risks as if this is it.

There is such an algorithm, and the world is the same now, but in reality, as soon as such an algorithm appears, the world changes absolutely.

Well, he becomes different.

Everything like, uh, I don't know about the feedback system there, what is called incentives in English.

Yes, that is, some motivating motives sharply change in those people who in a minute back they said that they are for everything good against everything bad.

And from the point of view, there is no long-term perspective.

Openness is a risky scenario which is less risky than a closed scenario.

Let's get back to the players, Mistral, and all this.

No, there is China.

There are several Chinese series of mods.

There are companies there, 0i.

In general, there are several Chinese producers of various kinds of IT products.

But here are the models of the family n.

Now such widely known ones are IBM.

They have a family of models, granite, not that so that it would be something top-notch, but also there is.

There is still a lot of different kinds of some smaller startup players that are in some places they are trying to do something.

Somehow they claim that they did something cool.

Little this.

Well, I don't know, there are people who do it, democratization, let's say, medicine.

Yes, here I am recently actually went to Plyos.

He became part of the Alliance for Artificial Intelligence.

It's such a big, big Alliance of different companies, among others, from such notable ones there is SES now and IBM that are moving various kinds of rubbish initiatives.

And we participated in the discussion panel together with such Anna Hitley.

She is a professor at PFL and the professor ate, and she made one of the best language models in the field of medicine.

That is, she took as a basis an open-source model and completed their studies in clinics in African countries.

In many African countries, the official language, English, is lacking.

There, doctors and the qualifications of doctors are far from perfect.

Among other things, here is the laboratory that is doing medical technologies for countries in which health care is not their first priority.

And second and third, the country's budget sees a big possibility that language models can greatly improve the quality of diagnostics of medical care in these countries.

Because they are, and a qualified doctor can sometimes replace from a diagnostic point of view.

And already a poorly qualified doctor or the absence of a doctor is such a model directly guaranteed to improve the situation.

Well, right statistically significant.

It may increase there life expectancy in the country if implemented and explained correctly to nurses how to use it.

Yes, he's actually already testing something there.

Yes, it was connected to this panel via video calls from Kenya from the hospital.

And Krutay, what a woman.

Yes, great.

Well, once again, how would return to artificial intelligence and its impact on society.

First, a poor man with an open model can do much more than a poor man without an open model.

And in this sense, how is it how Colt was a great equalizer?

Can be a great equalizer.

Moreover, I am almost sure that he is smart.

A poor man with artificial open-source intelligence, artificial intelligence, almost probably smarter than the stupid rich man.

A person with a private artificial intelligence, even if private artificial intelligence is better, at least for a while, for sure it will be like this.

And then what if we are really seriously considering it, some kind of fantastic scenarios when artificial intelligence takes over for humans completely, decides what to do, how much is a person agent then?

Well, in general, a little bit of freedom is lost.

Yes, who is smarter then?

Somehow, the question is lost.

That is, firstly, open models seem to help here too.

Secondly, it is very important where it is heading, median, and does it move.

Here is the bottom part distribution up.

Yes, that is, we see according to statistics, here are the people living behind the extreme poverty line as the percentage of the planet's population is decreasing.

On average, people get richer.

Yes, they appear super rich.

But it seems to me that it's not a big problem, honestly saying, let them be.

I mean, well, great.

They didn't just do that.

They appear; they did something there.

We thought there, well, in short, to me, it seems like this is a strange discussion to have.

Discuss how to make it so that it doesn't happen.

Poor, as it is.

As in the joke, yes, gentlemen, revolution.

Yes, and what do they want?

So that it wouldn't be like, well, granddaughter of the Decembrists.

And how Barney still wants a revolution?

They want there to be no rich people.

Strange.

My grandfather wanted there to be no poor people.

So I am, in this sense, as if on the side of the Decembrists.

I would like there to be no poor people in this too.

In the sense that artificial intelligence will help.

Recently, in the international media, generally, skepticism is probably recorded.

First of all, investors, what are you laughing at?

Express concern with concern what is the amount of data that is received.

Model increase in the amount of data does not lead to a proportional improvement.

Models ruled me in different directions.

At some point, I noticed a podcast of the founders of the company Andrin Hor, which is in the public domain.

And further, further, I disappeared for a day.

Yes, I, Andri, began to hear well on the topic of everything that is connected with.

They are absolutely great book, by the way.

This is generally one of the best below.

Wow, wretched 20,000 views.

What it just horrified me.

I just didn't, I understand people, what you are watching immediately.

Everyone there, and they too, they confirmed, they confirmed that yes, it doesn't surprise anyone that books exist of different quality.

We know that there are some quality books from which you learned a lot of things, and there are some so-so books.

But with data, any this situation, this analogy is true.

And do you have any data from which the model can learn a lot of things?

And there are some data from which much sense no.

They can be dirty; they may not be very noisy, useful there for those tasks that you, in the end, you want to decide with the model.

Now there is still a problem that now we have generative models, and the internet is clogged with them in such quantities.

What we don't know anymore, what data human and what kind of model are they?

And they are like that produce a lot of data.

Well, I mean, well, I was just talking to someone on Twitter.

I try to follow some people on Twitter.

Don't talk to strangers because many of them may even be, oh yeah, God bless him.

Listen, talking on Twitter with people is generally not an activity.

Thank you.

Wait, well, content on the sites, it is obvious that there is a lot of content generative.

People write there generative news.

People write any notes using artificial intelligence.

There are also these generative, as they are called, reels.

Yes, yes, yes, and you have more and more and more generative content.

And here it is value for the model of further education.

It also decreases because it's like already the second derivative of the original data.

And it seems, well, we have it, there are all sorts of empirical results that show that the quality of the model's work depends, in a broad sense, on the diversity of data.

That is, data there must be different ones.

It seems that they are there.

Why am I interested in creativity computing?

Because for now, generative models are good at outputting some kind of average answer.

Yeah, but bad give out something unexpected, something new.

What to do with it?

Bye, it's not clear.

This is diversity, and the second - this is data quality in a broad sense.

That is, from how much they are in general correspond to reality, before how noisy are they, and how much are they generally useful.

If you have, I don't know, there, do you have different Reddits?

Do you have a subreddit more useful, and there are less useful ones?

The media has already fed us all the information.

We know everyone in everyone.

Well, I mean, when we are talking about commercial solutions, yet one of the big problems, we don't know who and how much data did they steal.

We know exactly what they stole and how much.

Who doesn't know, there are some conversations there.

Paying for data is generally surprising.

Yes, in front of himself.

Yes, that is, the person came to, I stole a book from the library for you.

Later says, "Do you want 5 rubles?"

VS stole the library.

I came to you and stole your entire library.

Then he came and said, "Well, how much do you want for your library?"

Why so much?

Come on, less.

Well, everything has already been stolen.

Why, in principle?

Yes, well, that's interesting.

That is, and what's interesting is that again, yes, that is, there from the point of view of intellectual, I understand the position of honesty.

There is no such thing.

Copyright Koray invented panties means everything from ten to we means everything from ten.

Well, then post the model, post the model, post the code, everything from and to non-profit.

Why only ours?

Why is it only ours?

Who always say that you can steal, of course, that is, well, again, intellectual honesty is important here.

Yes, there I am a small fan of copyright.

I am there, I know that historically, Leonardo Davinci, even Pushkin somehow managed without him.

Managed without him, although he already was printing press and all that.

But somehow this scheme worked, and it is probably possible to build a scheme that will work without copyright.

But then those people who here they close the models.

Well, guys, well then let's bring everything out into the open, everything into the light.

Or if you are like a copier, well then come on, pay, go there.

What does it have to do with you?

Pay first and then learn.

Pom, take it.

This is the university.

No, no, this is the residence of the bishop.

The bishop himself lived in a castle on the mountain.

We had him there looked.

In the 10th century, decided that locks are not fashionable, comfortable to live.

Daru probably, poe vmm.

Anyway, locks are not fashionable.

I need a baron residence.

Be Luz, and actually him, they made a residence for the baron.

It is included in the list, and this is the park at the residence.

Here, but I didn't live there for long because then the unification of Germany happened, Bismarck and so on, and the significance the bishop has become smaller, as it were.

Judging by this, tell the building structure that he is French.

I watched, of course, naturally, how it all is.

Baroque parks are inspired in one way or another.

Versailles has been seen here near Berlin.

How can I time?

We think that it is some very unique.

It's another reason why I really like the history of science, history of technology.

Because when you're in, you look at the historical context, trying to imagine what the world looked like for people there, I don't know, 300, 400, 500 years back.

You understand that some things who are now trying to convince you that this is something unique, new, special, they are not so new and not so unique.

Here, I like to give an example of this startup named Yagan Gutenberg.

Yes, he was, by today's standards, by the standards he invented the printing press.

I came up with ening techy.

Yes, that is technology that makes it possible dissemination of information on a large scale, which were previously impossible.

Then radio later became the same technology, television, then on here rule, then internet.

Yes, that's what he did.

Print, look, but he would have done it himself.

The machine didn't go far.

There are examples of people who are similar printing presses they did, but they didn't become famous like that.

And what about there was a trick?

He found himself a good SEO.

SEO lived nearby in the city.

In Tenge lived the second man whom all they know, his name was Martin Ter.

He came to Tengu, looked at the machine, and said, "We will print the Bible."

Because it's not clear what to print.

And Gutenberg says, "Fire idea, began to print the Bible."

People don't read Latin; it's not clear what it says.

Let's translate, and SEO says, "Beer, the Bible should be printed in German."

Then they will take it.

Well, on the channel marketing, seems like you know, wait, people illiterate.

SEO thinks that pictures are needed.

Needed, let's go to a designer.

As a designer, his name was Durer.

He lived in the same, in a nearby town.

He made engravings.

The engravings in the Bible were a great hit with those who can't read, for those who can read and the actual texts.

And then Luther came in, drove her, sell, well, distribute, sell there.

As here, the story is different, but one way or another, op just an explosion of new technology on top of it.

Then people start to come up with, okay, the Bible is cool, but everyone has already bought the Bible.

What is ours there will be?

And then someone else breaks away and says, "Ho, books! How to recognize a witch?"

This is the topic.

In general, a lot of people died because of that.

The books themselves, how knowledge has spread, how to recognize fake news about that was widely spread.

How does it mean to recognize witches?

I think what was burned here, by the way, there were a lot of them here.

Executions are generally the same, this one Bishop Julius who founded the university and which did what?

This hospital is called Julius Hospital, wine there, and all that stuff.

He is at it arranged some kind of mass brutal executions because the witch must be caught and burned.

Naturally, this is just some kind of degree of savagery and ignorance simultaneously developed and enlightened.

Well, their reality was just like that.

Arranged, well, in the sense of the book, book source of knowledge.

The Bible is a book.

You are in, I have read two books of the Bible in my life.

And how to recognize a witch?

Here, trust the printed os.

Well, of course, you should have inside it will happen that the second one is lying.

And if the second one is lying, maybe the first one is lying too.

There are only two of them.

There is actually another cool example there from a close time.

It's like this very thing, a famous name in the world is the name Ariga.

Because that's what they still call it, Continent.

Although Riga Vis Puci did not discover the continent, but he was a travel blogger.

And apparently, it was checked very hard because he claims that he went there twice.

Historians are not sure, and it seems he, God willing, he went there once.

Claims he was captain of the historians in this, it seems, it is not true.

He said that he knows cartography well, but maps he seems to have something in common.

Judging by, he was a blogger who was like when his friend, the publisher, returned, he said, "Let's give it to them."

This is a cool thing.

It all fell apart and turned out to be a fair story about how he rode to America.

Then, when he died, they started making the first one, the atlas, one of the first atlases of the world.

And the guys who made the atlas are like, "Call the continent."

They are literally there in the preface say that our man recently died, friend Ariga.

Therefore, the continent about which we all learned from his wonderful stories, we call America.

It's unfair.

A could have called com.

I'll find the quote specifically.

I read to students.

Unfortunately, I still, I didn't learn it by heart, especially on Russian, but I will read it completely.

Robert Heinlein, "Every person should be able to change diapers, plan invasion, slaughter pigs, construct buildings, steer a ship, write sonnets, lead accounting, erect walls, set right bones, ease death, carry out orders, give orders, cooperate, act independently, decide equations, new problems, to introduce fertilizers, programming delicious, prepare to fight well, and die with the dignity of specialization in affairs insects."

That's how I think that Robert Heinlein is right, and we are missing people.

As if people, man, adaptive, a creature with a very flexible brain from that, that he finds himself in very different conditions, especially in childhood when the brain is most flexible.

We'd better not learn.

We learn to fix it.

Well, it's a big question, and can we at all?

But it would be great if every person could to reveal as much as possible what he has here.

There are how many 80 billion neurons in their brain.

It would be cool if everyone could to reveal this potential.

I am almost sure that wherever you look, there are not enough people.

And the problem is we instead think about how to make it so that there are 8, there are 10 billion in the top people on the planet, each hit the spot maximum professionalism.

For some reason, we default to self-realization.

We accept that among them there will be those who are not reach this point because they are like bad people or what?

Or not enough quality people or something else?

And let's just do it there, we'll come up with ways to feed them.

Well, how?

I think it's a very strange position, and what kind of things we will have problems because people won't know what should they do?

It seems to me that people need to be treated.

Should I treat you warmer somehow?

Love, and artificial intelligence, maybe it will help us, among other things, make it so that the percentage of people who reach this very thing, professional realization, grew and not fell.

Moreover, the whole history of development technology tells us exactly this.

On average, the percentage of people who achieve self-realization grows.

Yes, that there is, as it were, on average, we move from society in which you find yourself in a very rigid class or estate structure, and you plow the land because your dad plowed the land.

And you have less land because you had six brothers, and Dad's allotment was divided into six pieces.

And then one of you dies of hunger, and someone gets a piece of it, and somehow you continue, and your children will follow in your footsteps.

We have passed into some another world in it.

You have a little more; you still have a lot of chances.

There are restrictions on your social mobility, but we as a society seem to be realized that it would be good to have restrictions there were fewer but more social lifts.

And among other things, here is one of these elevators.

We also found out experimentally.

It's called education.

We know right away that there are people who were born into a family of farmers, got in.

So then in education, and something happened to them, and they suddenly became great scientists.

Then, yes, we have agreed with this too.

Now remained as if a million-dollar problem.

How to make it so that every child in the right time, this is the most education I made it, and in its best form, and not in how did it happen?

I saw it again, the problem we have is we don't have enough people.

We don't have there are enough teachers that we can pay enough just to hire good ones and put them in every school.

Here in each one so that in Kineshma, then here, the next Albert was born in Kineshma.

Einstein, and he looks good teacher of physics and mathematics and history and geography and Russian language.

Well, this doesn't work like that.

Industrial education system, artificial intelligence can definitely make it better.

Maybe change the balance, maybe add something like this mentoring.

Even now, actually, if a person tries to talk to the model so that she could change it, some basic questions, the model works well straight out of the box.

Commercial model or even oron the model that you launched locally and straight prom Tom prescribed.

Listen, come on there, act as if you were a specialist.

I need a credit in quantum mechanics to hand over.

Means here's what?

Ask me, I'll leave some questions.

It will be even better.

And if someone is around, this product will probably build like that.

Very good, I'm really in favor of this.

I think this is very important.

And again, it's amazing that all this is about for some reason, how was all this deliberately set up on the human cross?

I mean, who all?

Because in general, because in average in conversation around artificial intelligence, by default, the main scenario that is discussed in that, yes, that means people will become stupid, that people will not do anything, they won't develop, which means it will be one thing solid television.

Let there be one solid artificial intelligence which everything will be decided for us, and let us do it.

We won't build either because something, it's scary.

Let it be better this way.

Now that's a super plan.

That's just it.

Big trunk scenario.

Here I am, gave a lecture here for school teachers.

I told them how I use artificial intelligence for teaching.

I told them to the students, "Look there.

Here you can throw slides of the model generates questions for you directly in the format of an online test.

You can excellent, as if I don't know how to quickly throw them into the kohut and play with students during lectures.

Game, who will answer the questions faster to prepare this following the lecture educational material.

Now you can, there, five minutes earlier, it was required a few hours of teacher's time.

Now, thanks to the fact that we have several types of software and models that do this, it's much faster.

There, I don't know, it's not there.

Students need to be asked more questions.

Write home essays.

Bring because here I am, you need some essay on such and such mtusi.

It makes absolutely no sense.

Come up with other ones.

Methods of certification such as you suggest.

Don't fight at all with the fact that students, they write an essay in an hour, GPT, and just I can naturally come up with other ways when a creepy technology appeared which was supposed to make everyone stupider called internet search.

Students took abstracts, downloaded, and submitted them and received grades for them.

And there were also websites with all the homework done.

They said that from the search, everyone will become stronger, more stupid.

But in general, a real scientist should go to the library.

Well, wait, wait, wait, wait.

But actually, I don't want to represent a retrograde in this.

The conversation will probably have to be since you, Radical Techno optimist.

I say rather about the concerns that exist.

But really searching for information in compared to the ability to read large volumes of text, search so fast, copy-paste from po compared to the ability to express one's thoughts structurally on paper.

On paper than anything from the ability to read large the volume of text is really, well, sort of feel the difference.

That is, people who are good at expressing their own ideas, thoughts structurally think actually they are quite qualitative in their work from people who don't know how to do this.

If you don't train.

No, you understand, do you understand what the matter is?

Do you understand what the matter is?

The matter is fair for the rich and the poor.

Inevitably, things that make a living, it's easier to have side effects.

You know, there is actually no way to say that sugar and more calories - this is just like that in itself evil.

Evil is irrepressible diseases, and the question is not about ban sugar, but in a reasonable way it is used.

In exactly the same way with artificial intelligence.

Yes, that is how the question is not about not using chat GPT and in order to carry out this edge.

And people periodically don't know how to conduct this line.

Here is what to do with it.

That is, this is, and society in response to this, especially, well, such a society.

How to say it?

First reaction, what to ban?

Brilliant, ban, not let go.

That's it.

We understand that the ban can do this.

All you need to do is make it more attractive.

This is the Armed Forces.

The first reaction is clear.

So, well, well, why?

Because we are like that.

Are designed as a species because we humanoid because that this is a reptile brain.

Well, I think that you have to somehow not necessarily give free rein to the reptile brain every time, especially if you work in the education system.

Well, well, that is, probably, if you methodologist of the city education department, you can do it somehow master the brain ref.

Seems like it seems literally to be so to speak, can cope with the brain reptiles looking for a methodologist GORONO.

Well, this is roughly what I advocate in this regard.

Fear of technology.

Why do I think that you need to teach students?

Well, you have a new tool appears.

This tool becomes part of the labor market.

It becomes part of expectations of the employer.

If a person with chat GPT is 30% more productive than a person without chat GPT, I expect as an employer that you will work with chat GPT.

If the task education of people to life, of course, the task of education is to explain to people how to work with tools, explain what the advantages of these tools, disadvantages.

You are not embarrassed by the fact that we stopped teaching people to count slide rule.

And I remember people who were confused by this, who explained.

They gave many arguments why the slide rule is really important.

She develops spatial awareness, there, thinking connection of numbers and geometry.

And thanks to the slide rule, in general, another breed of people is emerging.

Perhaps if these people are right, maybe this is possible.

Explain the current such techno review rulers like this formed Felix had the same objects.

Yes, and we them at school, they advised to use it.

I saw the abacus in the store, abacus, all.

Well, I mean, I have a favorite example.

In this sense, there is one of the dialogues which is called there.

It means that Plato says that books are not very much, to put it briefly.

He says, "You see, Fyodor, book, she is for herself.

She's not talking, some idiot I will read strongly at I forgive, but read Fyodor.

Magnificent dialogue.

He says, well, roughly speaking, his idea is that whatever you ask about a book, it remains silent.

A fool will read it incorrectly and understand what their troubles will be caused in general.

That is, well, how would be the times of Plato and Aristotle?

People already, they are worried that Kato's technology is write down.

This is all this where is this?

It's good to walk around the garden and talk.

That's when people sing.

They grow up smart and cool.

I think that if the tool appears, it needs to be trained how to use it.

Because this instrument will be part of the child's life.

Moreover, now there are children there, which 10.

We can't even imagine what kind of world will they live in 20 years in the sense of what tools they will be using.

And absolutely definitely chat GPT, there would be no unity with this peak civilizations.

We need to explain to them how to learn new things, how to integrate it into your work, as if to adapt to these changes.

And, most importantly, critically think because of need.

Critical thinking is what there is a catastrophic shortage on the planet.

There are a million proofs of this because that we teach this badly, and we do it badly.

We teach you to think in terms of data and statistics.

And we teach poorly.

Critically thinking should be done from school.

I guess I still need to do this from school.

Here we briefly, I think about it.

At some point, they said, I think that meditation is also something that needs to be done from school to include.

Because we don't, we teach children to observe the world around them.

Because meditation is a way of sorts observations of the surrounding world.

Well, sort of removed here, not to interfere in the world around us and just look at how it is arranged.

It seems to me that the ability to perceive the surrounding world objectively is associated with the ability to silently look at the rain and not think in your head no thoughts.

This is real me, me, I can't prove it, but intuitively level I think it's very connected.

Ability to depersonalize someone time, including your inner self, allows, among other things, better criticism.

Just a physicist, you're just used to observing for the phenomena and then observe later.

Meditation was not invented by physicists and is quite, it has helped many cultures for a long time.

So it's strange that we don't teach this children's garden of Russia.

That is, as I understand it, you returned with the best intentions to work for the good.

Well, to do science in Russia, yes, make science of Russia.

Russia is a great scientific power.

I squandered yours so much social networks.

Not very deep, but enough back and discovered that one day you just literally stopped working in Skill Factory, has ceased operations in Yandex.

Well, I noted this one day on social networks.

Well, that was awesome, in fact.

There was some process when it started war.

I decided what I would like to call things with your own names, but the laws do not allow this without risk to my own freedom.

And to me, I don't like it.

It seems strange to me.

That is, I understand that again there, well, I like the situation when in countries that consider themselves possible to invade other countries.

For example, in the States, part of the population can as if to consider that this is correct, and part go out to huge protest rallies and to say that it is wrong.

And somehow I agree on this within society.

In principle, I don't think it's right to invade other countries, if anything would be a disclaimer.

But at least if you do this, you must inside for citizens of their country have an opportunity to discuss what in general.

At least that's what happens to me.

I think I decided that I need to go somewhere, do science elsewhere.

Next, for some time, well, the laboratory is there, is a big story.

Accordingly, commitment to students is necessary so that the students who started with me to do something finished doing it, protected, and so on.

That's when actually I got the opportunity, all these commitments at the level of papers to sort everything out.

The guys defended, published, and so on.

I have finished my interaction with Russian organizations.

And how much is it?

Took well with different ones in different ways.

Well, that's how it became clear that I have to leave.

I actually became talk to people about it and speak.

Let's figure out how to do it, somehow transfer cases and deal with everyone.

I understand that with these stories, Yandex, well, this is actually a well-known fact.

The shower is really big, the process among people about who thinks that necessary stay, who thinks that it is necessary to leave.

It was quite painful.

I think all people who have considered the possibility for themselves leaving was such an internal process, sometimes external.

No, in the sense it's not specific.

Yandex seemed to be the plus of this innovative company of Russia.

Well, maybe be there, I don't know, maybe ba was.

I also think that similar processes happened in Sberbank and Tinkoff and everywhere where there are people.

Once again, people who could leave and considered for themselves this possibility was thought of at that moment to leave or to stay.

I think it is, it happened everywhere, and for good reason this term appeared.

Yes, that's what it was.

Kanty is a great word, and I really like it.

I like how it annoys people.

Yes, that is, there is this is very cool because it shows that in Russia this is the topic.

Migration is very painful, in some sense taboo.

You write, no, I know somewhere on Twitter, there is a girl in Belgrade made a cool one.

So, you posted this little cafe, such a micro cafe in a hole from under a brick somewhere on Belgrade street.

I took a picture like this.

I opened mine, the first coffee shop in Belgrade, ran there.

A lot of people in the comments, some are bots, some people who mean like this.

These are Russian migrants on daddy's money opened a coffee shop.

Well, that is, they didn’t even were able to understand that the coffee shop is small and toy.

Well, that's what I'm sharing.

All the cities they come to make them very cute.

Came running some person in the comment is like this, not Rila Cantons, but migrants.

They are not in any way better than others, and this shows that the person does not understand the phenomenon of how calls itself migration.

Yes, that is not, I know people who fled from the Cuban Revolutions.

They called themselves Mari Lithos because they sailed from the same bay.

And actually, in Miami, there is still one there, a huge neighborhood, little Havana, where the descendants of these Marilis live.

This is a normal situation that people need themselves somehow designate.

As if people came up with the word lacan, they use good.

The word, it shows as if a piece of society which has accepted for itself some decision was made that there were risks.

Stay exceed the benefits and need to be tried something else.

That's all I think it is happened not only in Yandex.

This was happening in the minds of all the people who hypothetically for myself this opportunity I considered.

There are people who couldn't but physically there, and some people don't have it.

There was money, but someone didn't have the opportunity.

I know you're unlucky.

You have a relative which needs to be looked after.

Well, that's all.

Yes, what do these people probably think of this for themselves?

And those who did not have an internal struggle, there was an opportunity to leave, they were all about it thought everywhere in parallel.

And not in Yandex, in Yandex, I think that those too were just, well, it's kind of a difficult choice.

Listen, and you had someone close to you or there relatively close acquaintances with whom you are directly Rubaltik g was or is would be a good friend childhood who called me Vlasov here.

So yes, in correspondence, I find it difficult.

I'm still worried because of this.

It's funny because Vlasov's flag is that one, the flag that is currently in effect is used by the Russian Federation.

A catastrophic coincidence of circumstances.

Yes, that is, of course, it's amazing that in ninety-first year, people who could choose whatever they wanted, whatever flag they chose, the flag under these guys fought, chose any the name of the parliament and chose that a name that was never normal parliament.

You, of course, you can take the veche.

Here was the Russian parliament, evening, evening is really some kind of Old.

In what problem?

This was our last republic.

Last Russian republic died when it was captured by Ivan the Terrible.

Marfa Posadnitsa protected her.

It was our last republic.

We will have it restore.

Of course, when we will, we don't have another Republic.

We have, there are two in history.

There is Pskov, but it, Novgorod fell earlier, and we seem to have there are no options.

If we will restore the Republic, do Novgorodsk.

If we will do something else, okay.

Can we do it again?

We've already tried.

We've already tried the last 200 years.

We've been doing something different all the time.

But if we want to do something Russian republican state, the government, then the parliament will be there.

It's called a party in my lab.

The general meeting of the laboratory was called veche because it's normal Russians.

People gather for the evening to solve problems.

So, so, so it has been our tradition historically.

Well, in this way, the Duma is rather in creation of an empire.

Of course, you can stretch it out.

Of course, they took the Imperial flag, and the parliament was called fictitious, and which is Imperial, naturally.

And what is it?

It may still be that the words have the meaning, of course.

And did you think about that the word State is not equal to the word й yt?

This is the same.

The state of affairs is the state of the European, is like a kingdom.

A is yt.

Uh-huh.

And we have a state, state from the word sovereign.

Although there is a million, there are ways there, order is decent.

Can you imagine it's not the State Duma but decent Duma of the Russian Federation?

Cool.

I'm afraid too many songs I would sing your KVN.

No, wait, well, the words, how do they define me?

I study language.

Well, how would I teach languages?

I really like the language of the models.

Interesting language determines whether there are languages on which certain things are easier to say and some are more difficult.

There are languages in which there are no words that are translated one to one.

I don't know about other languages, but Finnish has them.

The word sioux, the word sioux translates well into Chinese on the word Dao and bad translated into Russian or another.

The European language developed in this way.

Finnish culture has, in Finnish culture, how is this favorite word?

Verb indicating that you plan to drink with morning in underwear without going out.

This is one verb.

Good plan because, well, that's how it turned out in Finnish, to put it simply.

Yes, as if that is what we are like.

We say this through our mouths.

Among other things, influence how we think.

And of course, the word State is a word which is directly from the word sovereign.

I do not, I was thinking specifically about the word, by the way, states are interesting.

Interesting thought.

Yes, that's it.

The Novgorod Republic was one of the most literate countries in the world.

Percentage of the population able to write and read was one of the highest in Europe, at least judging by the number of birch bark letters that we are still digging up.

Yes, self-government, everything was fine.

That is when they tell what Russian means there, the people somehow can't govern themselves.

Guys, come on, it's been 600-700 years already.

Worked great.

There were mayoral elections, ends there, all business, everything, everything, please.

Literacy is also a good word because literacy is a document.

A literacy is a document.

A competent specialist is more than that, educated specialist.

That is, this is good, tako horo Russian word.

It is for me, I like it.

If we agreed to build society on these three such polar stars that lead us somewhere, like this triangle, then I'm in business.

Wherever it is located.

Yes, no matter how it was, it was called.

Wherever it was, these are good metrics.

And the main thing is that got it.

How to optimize all three?

It is clear how to optimize with artificial intelligence both.

Yes, of course.

Well, it seems to me that some things are already, that is, things related to happen there, digital document circulation, fro bureaucracy, they are already very active, are being researched.

Yes, and people are trying different things there, models as some kind of state processes to be completely digitalized.

Yes, how well, that's also a clear story.

Artificial intelligence, that's how it would be.

He sooner or later leads us to the question of whether there is free will.

If it is, then that people have it.

She is in artificial intelligence reproduced.

We don't know yet.

Yes, there is here is a hypothesis known to me.

David Yan once she told that on physical level, if free will exists somewhere, then this at the level of quantum effects in our in the synapses of the brain, everything else is biochemistry concentration.

Well, there I see some electrical signals, that's all.

It is quite deterministic.

SS, that here between synami, yes, there between the synapses when the electron is running there, there may be some quantum effects that really influence that freedom is somehow tied on quantum effects.

That the brain is such an object in which there are a lot of such here are micro quantum things happening on at the synapse level, and we don't know how they happen.

But that's what gives free will.

It's absolutely awesome.

The guy is Ion Mac Gilchrist.

He wrote the book The Master and the emissary about the left and right hemispheres of the brain.

He generally neurobiologist.

Disclaimer, I am not a neurobiologist.

No need to consider that what I am, what I'm telling you scientifically.

Read how it is.

Other poets at the graves of Christ, he just neurobiologist.

And he, among other things, puts forward a cool idea that the brain is not the source of consciousness and the receiver.

It's a very cool thought.

And what about the receiver?

Well, there is one.

Actually, that's how this is pa Pansa ZM.

Yes, right now it's very fashionable among a number of such scientists, neurobiologist Pansa's idea ikizoglu for free will in the world.

Synapses are way as if with this something interact than you have synapses more topics you have more of this free will itself.

The less you have of it, all the more so because you are determined.

You have more of these quantum effects, simple structures in which synapses little, little.

This is an interesting approach.

I don't think he is true or false.

To me, this seems like an interesting thought experiment, just yes.

That is how are you Avatar, as it were, you connect.

Well, there you can do it in different ways.

It's to say, yes, yes.

Well, that is, there are world religions.

This is usually formalized in level of what is there like this.

If we got to this place now, we can talk about some crazy theories.

Well, here's the first crazy theory.

This is just that consciousness is a brain technique.

This is the receiver that consciousness is somehow just being in the world.

That's because we know that there are some strange quantum, there really are effects in the world.

There is, for example, quantum entanglement.

That is, you can confuse two particles.

Yes, it's farther to spread them as soon as possible.

You measured one, the second one too, sort of measured no matter how far it was.

It's pretty strange on the one hand.

And here's it means the brain just works somehow, receiver other in this direction.

Comparatively mainstream frin too, a theory is a theory about how it is there religions interpret.

Yes, that is, religion has been telling us for quite some time that any religion, she says that as if there is some common will.

We kind of only preachers seem to analyze this will through itself.

It is as if such a way of interpreting religious.

Another cool fresh way interpretations are a theory of simulation.

Yes, about the fact that there is such a thing.

The article is called simulation hypothesis.

It's an article by a philosopher, by the way.

They gave me a huge grant at Oxford security center, and he wrote a book Superintelligence.

One of the most famous of his scientific papers cited articles.

It is about that let's imagine that computing technologies are developing.

At some point, we create a virtual reality.

It becomes more and more real.

In some way, the moment we will launch it in our computer simulation of reality.

Well, we will reach such scientific powers that we simulate very close ones on the computer.

Let's now imagine a real experience.

Well, as soon as this happens inside this, something will start to happen in reality.

And after a while, there inside someone will appear who will be able to launch the next virtual reality.

So now, so if we already see that we live in some kind of reality, what is it the probability that we live in the root reality?

A, what is the probability we live in the simulation?

We are just a ghost appreciating this to another.

Yes, yes, and he says, well, formally speaking, you have one root reality gives birth to the infinite number of simulations embedded in it.

So strictly speaking, the probability of that we live in a simulation is higher than that we live in a base reality.

It's such a beautiful philosophical construction, but there is some amount around this.

People who, for example, are trying to connect theoretical physics with this hypothesis and, for example, try to find some inconsistencies in theoretical physics which would be well explained simulations.

Simulation hypotheses and one of cool inconsistencies which comparatively well known - this is a discrepancy about the observer paradox.

Means in quantum, there is a paradox in mechanics that it's called the observer paradox.

Particles behave differently if you are on whether you watch them or not.

Here we are when we play a terrible thing.

But if you play a shooter, the shooter is not behind you rendered.

Rendered only your field vision.

When you turned to levels, a different part of the level is rendered than you.

No observer effect in Kwa mechanics.

Yes, this is a cool, cool approach.

But it's like a free, but very cool.

Well, it's the same.

Yes, here's what if you do as if the world is completely science fiction, of course.

Well, how about constantly rendering the world?

It's quite complicated.

So it's like some things can be rendered as needed necessity.

If the observer needs it.

Well, there are people who somehow get along.

These two things, and let's say cool too.

Great approach to this.

He is like, "Well, look, now let's imagine that we did something, a simulator to teach people."

E cool, if you know what it is, the trainer.

You treat it wrong when you play poker for chips, and when you play poker for money, you play differently.

So if you're in some kind of reality built a nested reality, for example, to teach people ethics, you have a question.

And if you, let's say, there is even data experiments, you find out.

And they thought that they live their real life is much stronger than that educational at the exit than those players who knew what they were playing.

Well, you can just decide that this is how we are.

Now we will start the simulator, but we won't tell the players that they are playing.

They came here to learn ethics.

We will just start the simulator.

Later, they will look at the results later to disassemble as if retrospectively.

And this is literally what it's about.

Religions say that you got into reality.

You in reality, we need to make some decisions later after some time when you die.

Prot x decisions to say as yes, and they explain to you here it was not very, but here it's kind of nice.

This is it also such a cool frin approach to that how does simulation theory relate.

As they say, old songs in a new way, of course.

Of course, let's go blitz.

Well, our traditional one.

Triad is fast, cheap, good, necessary.

Choose two.

Fast and good.

A who is the coolest scientist alive today?

It'll come then, so I came and it works. His fellow graduate students, when they found out that he received the Nobel Prize for biology, threw everything into the pond together.

So, in the courtyard, in the courtyard, there is a rod and the Max Planck Institute. They raised him up in joy, hands, and threw him into the rod because he loves to harden themselves. That's why they love the sauna and swim in the pond.

I think it is cool. I think when you are a cool scientist, what can you do? An offer to go to work at the Max Planck Institute? Say yes, but it would be good so that there would still be a sauna. Here it is, this is the level of negotiating position.

Who is my role model? I would like to be in my place someday. Life in this, I wish that the sauna at Swan Pub is really cool.

Overall, paleogenetics is a very cool story about what we understand more about in general peace, thanks to sequencing history or mathematics. Maybe physics, dangerous invention, invention of humanity, nuclear bomb, and great nuclear energy is powerful.

Well, like, it's really a big breakthrough. For a relatively short time, thousands of years to go through the same path that humanity has come to such a source of energy is very impressive.

At the same time, I think, what degree of ethical? That is, it seems to me that ethics and aesthetics seem to be somehow strongly connected. I don't know, I can explain how, but intuitively, I am connected with things.

For example, good math, good physical formulas, some laws of physics are beautiful, aesthetically beautiful.

Do you have a favorite law of physics? I have favorites. Well, yes, I do. Yes, I do. Well, I can't say that I have one.

One of my favorite physics experiments demonstrating probability is this: there is this Gaussian distribution bell. We talked about it with you today. There is such a board called Galton's board, which was actually invented by him, Galton, to show next.

So, you have a board like this, and it has such needles, a triangle of needles, and lots of metal balls. When you turn the balls over, they start to fly down such channels.

Periodically, the channels are interrupted, and there, the needle, that is, the ball, falls and hits a needle and falls either to the right or to the left. Accordingly, it further gets into the neighboring channel and runs along them, either to the right of the needle or to the left of the needle, and then again to the next needle, and so on.

In the end, actually, just such columns, that is, balls fall into columns. When you're turning over, I think you're just showing the audience some video. You watch it, you turn it over, and the balls lie in Gaussian distribution because each ball collides with the needle.

There, I know, 10 times is essentially 10 coin tosses, right, left, right, left. The ball that was thrown 10 times, say, right, the balls fall into the middle because they threw out half heads, half tails, and ended up somewhere in the middle.

This is a way to illustrate what is called the Central Limit Theorem. It shows that a series of random events converges to a certain distribution. In this case, a series of binary events is what it's called, the Bernoulli distribution, converging to a normal distribution.

This is a good way to understand in what cases you will have a normal distribution in nature. Normal distribution appears when you have a series of coin tosses, very simplifying.

If you have such a process which is always 50%, and you do it lots and lots of times, you throw it away, Gaussian appears. But if you have some kind of distribution where the winner takes it all, then what's next for you?

It turns out that the wall is the law. I'm not on for nothing. Lectures go, everything is so. Yes, because you won from you once, at once, big budget play many, many times again.

So that many, so that once win, and how would you play once? Someone lost right away, and he has no more attempts at all. You still have 1,000 attempts. Out of these 1,000 attempts, you have, how would there still be a small probability, but it's more that you win the second time, having already put, as if not one, let's say, 10 and 100 attempts.

And you too, accordingly, next time is winner takes all, and you have a kind of head that gets higher and higher. Yes, yes, yes.

And it's like social objects, more often everything is described like this. The book that I advise everyone to read is a book that I advise everyone to read: "Walls on the Lower Opening."

Slowness about meditation? No, this is an adventure novel about a captain of a long-distance ship who is a bit slow on the uptake. The book is about a guy who understood that he was slow and thought that it was necessary to somehow live with this.

If you could change one thing about people or in the world, then what kind?

Wow, this is very complicated.

I don't know the question. I would like to...

So that, well, probably people should be kinder. It would be great if people were kinder; that would be good for everyone.

Better.