in terms of accumulated investments in the belarusian economy, russia firmly holds first place, over $5 billion. again, about 2,500 companies with russian capital operate successfully throughout belarus. great practical benefits for deepening russian-belarusian interregional cooperation are brought by integration processes developing both within the framework of the union state and along the line. of the eurasian economic union, precisely thanks to the ever closer integration of the companies of our countries, no matter what region they are located in, get the opportunity to work in a single economic field, without barriers and obstacles. this directly concerns the areas of science, technology and innovation, precisely those areas of interaction that were discussed these days at the forum. i would like

to remind you that on january 29 in st. petersburg , alexander grigoryevich lukashenko and i approved a new strategy for the scientific and technological development of the union state. this document provides that in the period until 2035, russia and belarus will implement a number of joint high-tech projects for import substitution, launching new advanced production facilities and introducing a competitive market. capable products with high added value. let me emphasize that a lot is already being done in this direction by the russian and belarusian regions. for example, the ural civil aviation plant and the aircraft repair plant in belarusian baranovichi are working together to create a light multi-purpose aircraft. enterprises of moscow and st. petersburg, moscow and leningrad regions. as well as minsk and

the minsk region has established cooperation in the production of optical devices and photonics devices for communication systems, data transmission and computer science. regional research centers and design bureaus in russia and belarus are working together to design the orbital and ground-based components of earth observation satellite constellations for navigation. and, of course, i can’t help but say. power plants. nuclear contracting organizations from many russian and belarusian regions. to date, bes has already generated more than 30 billion of inexpensive and clean electricity for the needs of the belarusian economy. i will also mention that regional

scientific centers and universities are actively involved in cooperation, which is implemented through the academy of sciences of the two countries, as well as educational exchanges. in particular , about 12,000 belarusian citizens are currently studying in russian educational institutions; in the next academic year, another 1,300 belarusian students will be able to enroll in our universities on a budgetary basis. dear friends, in in conclusion, i would like to take this opportunity to congratulate all our belarusian friends on the independence day of the republic and on the eightieth anniversary of the liberation of the hero city of minsk from the nazi invaders, which are celebrated on july 3. both russia and belarus will always remember that during the great patriotic war, our peoples together defended their homeland and liberated europe at the cost of enormous sacrifices. made a decisive contribution to

saving the world from nazism. in january , alexander grigorievich and i opened a new memorial complex to peaceful citizens of the soviet union who fell victims of the nazi genocide. the creation of this monument is just one of many joint projects dedicated to the great victory. it is important that our regions regularly host events commemorating... the feat of our fathers, grandfathers and great-grandfathers. a good example of such joint work with the direct participation of the regions is the memory train youth initiative, which is being implemented for the third time these days. schoolchildren from russia and belarus, as well as from azerbaijan, armenia, kazakhstan. kyrgyzstan, tajikistan and

uzbekistan, will travel through russian and belarusian cities, and see with their own eyes memorable places, fields of decisive battles of the great fatherland. concluding my speech, i would like to once again wish the organizers and all participants success. the results of your joint activities will undoubtedly contribute to the strengthening of russian -belarusian ties in the interests of our two fraternal peoples. thank you for attention. this was footage of the president’s address to participants in the forum of regions of russia and belarus. rapid development. artificial intelligence is becoming not only a source of pride for developers, but also a cause for concern: whether these researchers will turn out to be so smarter than people that humans will lose all control over them. already now , university networks are discovering abilities that no one put into them. we will discuss all this in the program, the issue of science. watch it

now. too powerful. so that it can be controlled. strong artificial intelligence from the pages of science fiction books is again breaking through editorial pages of major news publications. another open letter from experts in the field of artificial intelligence has been published. a group of experts, mostly former current open ai employees, accuse the industry of being reckless. it is necessary to ensure a certain level of hype, no matter what the instrument. projects in the field of artificial intelligence are expensive and require the attraction of enormous funds, so companies, as it were, need either funds from the state or private funds investors, because of this, any hype, any statements about revolutionary nature will attract much more capital than general arguments about how great it is to live with artificial intelligence, we need to create a common field, that we are

better, that is, we have developed such a powerful artificial intelligence, yes, even we ourselves are scared by this, industry leaders are constantly accused of playing with fire, and partly these fears can be understood, artificial intelligence has achieved incredible things in recent years... with the world, the threats associated with this are growing powerful and developing their communication capabilities , current models become increasingly incongruent. what is strong artificial intelligence and should we be afraid of it? how close are we to creating a fully functional digital mind? this is a question of science, i am alexander gasnikov, rector

of enopolis university, and today our guest is alexander krainov, director for the development of artificial intelligence technologies at yandex. alexander georgievich, hello! hello! alexander, well, the first question is, what is skillful anyway? intelligence, oh, na there is an endless amount of debate on this topic, it seems to me that we are only now agreeing on this definition, classical artificial intelligence is the ability of a machine to demonstrate skills comparable to a person, that is , to demonstrate something that was previously inherent only to humans, but i i would formulate it this way: artificial intelligence is programs, services, algorithms, whatever, created using some kind of automatic algorithm for selecting parameters and values, here is ordinary programming, ordinary compilation programs when the developer himself decides at each moment in time that if this is how we will do it like this, and if this is how we will do it like this, and if knowing what we should receive in

the end, the program selects these parameters within itself independently, this is artificial intelligence, let’s say, a program for playing chess from the sixties. ago, this is all artificial from the nineties, i don’t know, ten-year-old intelligence or not quite so, well, to some extent, to some extent, it’s just that we then it, then it looked so weak that they didn’t call it at all, of course, artificial intelligence, when suddenly they start to beat a person, this just seems like artificial intelligence to us, but here it’s straightforward, no one has yet learned to speak, to directly and unambiguously set a barrier, this is still not artificial intelligence, this is it already. then a logical question arises: what do we get, we have long had a situation where artificial intelligence is superior to humans, or or not quite so, we have been in this situation for a long time, we have been in this situation for a long time, but here we need to understand, what does superior mean, because how long

have we been in a situation where one person is superior to another, yes, but not in everything, that is, there are some very, very narrow areas where artificial intelligence is superior to humans, what is happening recently , those that we previously considered as our absolute advantage, here again we remembered about chess, chess, a very difficult game, these are the prerogatives of a person, or some other games, such as the game go, which are even more difficult and chess, it is implied that there is a person shows intuition, or poetry, prose, that is, writing some works of art, or music, or, say, creating paintings, or a skill such as programming,

convolutional neural networks came up, which were then very popular for tasks, say, image processing, they were very good for playing go, well , great progress was really achieved in this area and this was thanks to the fact that this area was very good, and the company deep did a very good pr job. all this history, and apparently, this is how i imagine it now, that when they were thinking about how to present it all outwardly, well, machine learning will win against a person in go, well, somehow it won in a not very interesting way, or even a neural network is already more interesting, but still what - the peppercorn is missing, here this could not be more suitable: the term artificial intelligence, until that time it lived in computer games, when a person played with a bot, he said, there is a strong artificial intelligence or not a very strong artificial intelligence. and by defining this, the computer plays strongly or not, here artificial intelligence sounded in a gaming environment, it fits right here, but a person

wins a game against a computer, that means a person wins an artificial intelligence, and if an artificial intelligence beats a person at a game, that means an artificial intelligence wins against a person, and then they wrote in all the headlines that artificial intelligence had won over humans, from that moment on machine learning, before that all these algorithms were called machine learning, went through this wonderful rebranding and began to be called artificial intelligence, this is not yet... using this term, it seems to me, has now become accustomed, but mostly they use something else, neural network, yes, machine learning, whatever, but art intelligence is somehow not very uses, but in the entire press, in all media, yes, now we have artificial intelligence everywhere, well, the example that you said, as far as i understand, there was still a very important element... well , let’s say, some kind of self-learning , that is, that the program played by itself, generally on

your opinion, this track, self-learning, it plays an important role in artificial intelligence, it plays an important role, here’s what’s interesting, we slightly misunderstand what self-learning is, well, we are people, when we talk about artificial intelligence, we imagine ourselves some kind of program that lives and learns with it somehow it becomes the best. well, actually this is not true, we humans learn and do something at the same time, if we play some kind of game, we get smarter with every move, well, presumably smarter, at least with each move we play somehow differently, we have a process of executing some program and action that we have and learning occurs simultaneously, with computers this is not the case, a machine can do it a thousand times, 10 thousand times , perform the same action as many times as you like and learn nothing.

the opponent's next move is meant or any move at all, that is, we have some kind of game, in this game in this position the following next move was made, the task of the program is to guess which one in the played game the next move was made, and you can, the more good games you learn from, stronger players, the better the next move you assume, you can learn from people’s games, or you can play, the program can play with itself, it is at this moment, while it is playing . she doesn’t get smarter by herself, but at this moment, having played a lot of games, she accumulates, roughly speaking, tasks, at this moment she is busy formulating tasks for learning, that is, there is some position in which the next move has been made, then her training is to guess which one was made the next move, in this regard, yes, it’s self-learning, but i’ll repeat it, i’ll repeat

a completely separate controlled process, let’s say we have some kind of rides... and autonomous delivery robots, yandex ones, that breed food, here’s a robot for now drives down the street, he doesn’t get smarter at all and won’t get smarter as long as he just drives like that, but he records a taxi, here’s a yandex taxi, let’s or an unmanned taxi, yes, but the car records some events happening around, not just any , everything, then on this experience, on these events is launched separately somewhere on the servers there is a training process, after which it thoroughly tests, if it turns out that yes, it really is better than the previous one, then... an update process occurs, that is, now new versions are being uploaded to all these robots, so they have now become a little smarter, but on his own, but if he is an autonomous robot or drives a car, then he does not get smarter in any way, well, we have now raised a very important question, this is really the so-called training sample, that if it is not very large, then there may be problems with the quality of work, here

now many experts say that modern language models already use almost the entire internet, or at least. you know, with this data you can be much, much smarter, we still don’t know how, we people haven’t yet learned how to present the data correctly, understand which piece of data is in the beginning, which is later, how much data there should be, because there exactly there shouldn’t be perfectly good data, that is, the model should, in theory, see everything, well, in the end , in order to distinguish good from bad,

but we don’t know how to teach, in what proportion, in what quantity, and how to take this experience. not from where, we have some kind of experience, understandable how people learn, but people still have some completely different hardware device, and it’s not entirely suitable here, very expensive, expensive experiments, that is, training one large non-growth model - this is, well, probably now it’s tens of millions of dollars, and if you want, you think like this, let me try, that at the beginning give data of this size, then this, and what if at first try it out like you learned from school textbooks, then add something else.

outgoing on video, you can probably learn this too, there is also, of course, audio information, but i’ll combine it here with video, because video can be full-fledged, audio and video, it’s not like...

the physical world for learning models, again, what does it mean to collect, it’s not like collecting something that we people haven’t collected, no, something that we people have collected beautifully, but we can’t share it, that’s even simple, we have some something ordinary is our natural movement, like we take a bottle of water and pour it into a cup, at this moment a huge number of our receptors are involved, we balance this bottle, we understand how the weight changes when we tilt it, how the weight changes when the quantity changes. water in a bottle, and we do this somehow imperceptibly in the background, we have this information, we take the next bottle, we are already trained, this information, it is not

saved anywhere in any form, is not digitized, it is too simple and natural for us to digitize, this is a huge part of understanding the physical world; it belongs to neural networks and there is no place to get it from yet. thank you, in your opinion, are modern, well, let’s say gpt type models a step forward? well, we don’t fully understand strong artificial intelligence yet, it seems to me that we haven’t agreed on what is considered strong artificial intelligence, that is, what is happening, here they say that intelligence will be strong when it beats a person, okay, so it beats him in some tasks, we’re like, well, no, he’s in general not very strong, because he can do something in this task, but i don’t know, there is a program that beats a person in a goal, and a program that writes poetry or answers questions intelligently may not be able to add 2 + 2. and what kind of intelligence is this, and then the idea appears that well, strong artificial intelligence is a kind of general algorithm that is capable of solving various problems and being

comparable, beating or not beating, but being comparable to a person in different tasks and acting in conditions of uncertainty , they didn’t teach him some specific problem, and he solves it, demonstrates this solution, to some extent this is what text-generative networks show us, because... how they learn, they learn to logically continue the text, they each since they insert the next word so that it would be naturally appropriate here, i still can’t get used to it, i still can’t get my head around the fact that sometimes they even solve mathematical problems in this way, why do i sometimes underline the word, because in in most cases they betray complete heresy, but the fact that sometimes these cubes are put together like this, this random next word is inserted, that this turns out to be the correct solution to the problem. this is of course amazing, unexpected, to some extent yes, this is a step in that direction, but still it seems to me that we are still very, very, very far from strong artificial intelligence,

that is, i do not at all share the opinion of people that we we've already gotten very close to it, no, it seems to me that we 're deluding ourselves too much here, into artificial intelligence, we still don't have a strong artificial intelligence, well, at least a decade, but it seems to me that we just don’t even see it yet, it’s very difficult. will predict how long it will be before him, because he, he is somewhere beyond the horizon, that’s what we see now, what we can predict quite confidently, he’s not there, when he appears there beyond the horizon, i don’t i know what needs to happen in order for us to decide that strong artificial intelligence has appeared, we must have such a program, such an algorithm, such artificial intelligence that can effectively solve a problem that he was not taught to solve. what is missing for this strong artificial intelligence? and there is not enough computing power, and there is not enough data on which to learn,

data from different modalities, not just text, well, in the sense that it’s not that there is not enough data itself, there is not enough skills to learn from this data. we still have the so-called curse of dimensionality, every time we add another dimension, an additional dimension, we add more data, we have very the requirements for computing power are greatly increasing, and we are now running into them, on the one hand, computing power seems to be growing, but all this becomes, the further, the more expensive, you need to understand that this appearance of each language model is super expensive, very few companies can afford it, and right now it’s not paying off right now, such investments in the future in development, in everything else, but... it’s very expensive, right now, and some significant strengthening will lead to an even greater rise in price, that is, i think that we will also get stuck here and slow down in this place, well, i also have

this... some kind of consideration that i am trying to analyze how artificial intelligence differs from us, from people, of course, it really differs everyone and no one is trying to reproduce a person, but of course we periodically spy on a person, because there is nothing better, so this is what happens to us and people, we develop in society, and in general a person’s intelligence is determined by society, there is even such a theory, what alone a person's consciousness will not travel.

this is the general development of intelligence that people have, and these are some mechanisms for the transfer of knowledge that are completely incomprehensible to me, but they are super perfect, because a lot is transmitted by speech, but there are people who are hearing impaired, who are unable to hear speech and yet everything is fine with them, and there is a lot that is transmitted through vision, but again there are people who are blind from birth ; everything is fine with their intellect, of course, it’s probably somehow harder, but in general they are not at all. are in no way inferior to others people, that is, our abilities to somehow interact with each other and transmit information to each other are multifaceted and some alone, even if some channel is defective or cut off, the remaining channels compensate for this, and the neural network is all that is being created now , she appears like a child on a desert island, she grows in absolute, develops in absolute isolation, receiving very little feedback, now some mechanisms of

this reinforcement lowing with human are appearing. reinforcement learning with feedback from a person, when, well, very in quotes, very strained, conditionally, we begin to somehow turn on this network of our society, somehow interact with some algorithms, but this interaction, in other words, we simply tell the neural network how successful it is answered and thus she tries like yes, and in this way we transfer at least some of our data, but this mechanism is so meager, of course, that all the neural networks are still in isolation and to me.

a person with a neural network, i don’t even theoretically i can imagine where this could come from, i understand why we are afraid of this, because such an ideal antagonist for us is this soulless creature that lives somewhere out there and yes, some kind of skynet, from the point of view of drama , this is a better antagonist than even little green men, because they are still at least somehow similar to us, this is a completely different thing, let’s mean in works of art to create an image of an enemy or, well, someone of what? alive and so give rise to fears, but in general to be afraid it seems to me that there is no more fear that artificial intelligence will take over us than we can be afraid that we will be taken over by computers or washing machines. let's summarize what we have come to? strong artificial intelligence is the inevitable future, the question is, when will we get there, apparently it will be...