Satellites, tanks and AI. Augmented Intelligence

43
Satellites, tanks and AI. Augmented Intelligence


AI for military purposes or, as some call it, augmented intelligence


There is so much confusion in the first parts. So much at first glance different and absurd at the same time. Incomprehensible satellites, strange Tanks with electronic filling, in the title there is a hint of a voice assistant from Yandex or Sberbank. Now I will connect all the parts in order. I will disappoint you right away: there will be no talk of any “Terminators” or Skynet. The consideration will be what is available and what you sometimes have to deal with, well, and a little imagination based on really available developments.



What do we know about neural networks and AI based on them, where do we meet with them?

Usually these are annoying voice assistants in the support services of banks, insurance companies and mobile operators. Alice, Marusya, Alexa from the same opera, only better. What else can they do and where can you encounter them? Those who drive a car and accelerate above the allowed speed receive chain letters with recognition of the make and number of the car. The system was finished to the level that it distinguishes well between splattered car numbers in a dense high-speed stream.

Soon there will be a system that distinguishes between the busiest business people in the world who just urgently need to talk on the phone while driving, and a system that captures the throwing of garbage from cars. I look forward to a system that distinguishes between humanoid individuals with erroneous DNA that throw cigarette butts out the window or balcony. The use of AI to combat this dead-end branch of human development is more than justified.

Comic and not so videos with the replacement of the main characters with other characters are no longer as surprising as before. For such tasks, solutions based on neural networks are most often used. Deepfakes and the generation of pictures of naked girls based on a text description are from the same series. The most realistic three-dimensional models based on a text description with subsequent printing on a 3D printer are on the way. After that, you can rightfully consider the future to have come.

What else can be presented as examples, besides neurons, for the complete exposure of girls in photographs in swimsuits? A bunch of financial tools for predicting market behavior, trading Robots for high-frequency trading on stock exchanges. Actually, where did all these things go to the masses. In some banks, the decision to grant a loan is no longer made by a girl with a calculator and 1C courses, but by an AI-based system.

Based on the analysis of your electronic payments, they are able to predict the level of credit confidence in you with a certain accuracy. Branch prediction and data prefetch systems in modern processors. All kinds of speech recognition systems, images, patterns of behavior. Many complex problems are modeled with the help of genetic algorithms. And often the output is very strange, but working solutions.

The relevance of the use of AI based on neural networks increases many times when it becomes necessary to solve poorly formalized tasks. Together with the ability to process and analyze huge amounts of data, this allows you to calculate options for a large number of variables. At the same time, there is no influence of the human factor, that is, it is independence from the mood, personal problems and emotions that are inherent in people. The main tasks that can be given to the mercy of artificial intelligence are in the plane of accelerating decision-making and automating routine processes.

For what purposes is the use of neural networks and AI based on them justified?


1. Pattern recognition


In social networks, many have seen or have long been using the ability to recognize faces in photographs. You upload a photo from a party, the social network selects happy faces in rectangles and offers to indicate participants from among friends on them. Then she automatically begins to recognize them in the following photographs, then completely guts the entire photo gallery and finds drinking buddies there in other photographs, which very much resembles a wife or mother. And if earlier there were certain problems with recognition, now the AI ​​​​of the social network determines exactly where the cat's face is and where the owner's face is. Even when the cat's is much better and more pleasant than the owner's.

With image recognition, their classification in photographs and videos, social networks can easily cope with a huge influx of users. It’s hard to say how many thousands of photos a particular social network shovels every second, because people upload photos and videos quite often. Some continuously.

Now back to optical reconnaissance satellites.

Here they are uploading photos and videos from the battlefield in real time. They need to be processed somehow, looked through, find the necessary objects, classify them, if necessary, take the materials received yesterday or the day before yesterday, and compare in order to understand what the enemy has done there.

This is where the fun begins. Who will take care of this routine?

There is a lot of data, and they are heterogeneous, and the human visual channel is rather weak for these purposes. You have to carefully look through the square after the square, remember and compare with what was on the previous squares, remember what happened yesterday.

This is not at all like watching a demobilization album. In this case, you must act quickly and very carefully. Only the eyes are not made of iron, they are characterized by fatigue, the brain can also get tired of such work, although the possibilities of the human visual cortex are fantastic. And this is one of the troubles. The human brain easily draws what is not there. Computer vision developers will confirm this for you.

On the other hand, train AI for certain images, teach it to process data from different ranges of the electromagnetic spectrum and let it thresh itself, recognize fortifications, equipment, fighters. Pull up if needed historical data for yesterday, the day before yesterday, from last week and compares with those already available. On finding objects and suspicious things informs the operators, and they make decisions.

For these purposes, huge data centers are not needed, because several dozen “Instasams” upload much more data in high-definition videos and photos processed with ennobling filters into their public pages than our satellite constellation.

Accordingly, the question is brewing, where to get such data centers? Preferably mobile. Well, so that they can get away from the enemy dronescreated back in the Soviet Union.

Most likely, they will have to be taken in the same place as domestic video cards, processors and high-definition monitors or solid-state drives. There is no listed element base in washing machines and microwave ovens. Even buying something old is a problem. The same neuromorphic IBM chip developed in 2014, go and try to find it.

It is necessary to fence a chain of mediator countries and try to please them so that they do not accidentally remember that they can suffer for mediation. You will have to develop a lot from scratch. And this is a rather long and costly process, taking into account the fact that we do not have not only such a number of specialists, there are not even theoretical groundwork in science. For more than 30 years, we were very willing to do nothing but buy high-tech developments, and with an open soul, at best, we sold for pennies, or even gave away just like that as unnecessary.

I will not sing odes to gallium/indium arsenide again. Judging by the latest observations, technologies based on it are being developed and developed in the countries of Southeast Asia. It is possible that those cunning to the perspective and breakthrough guys will reach the research work of Vsevolod Sergeevich Burtsev on discrete optical computers.

Perhaps we will soon see computing systems based on other physical principles. For 30 years, photonics, microelectronics and technology have advanced quite far, and now much is real that could not advance further than theoretical developments before. Or lack of funding and any interest, as in our case. And we, as always, will try to copy in haste and try to catch up with the rushing train.

Nevertheless, the task of pattern recognition is feasible with our current capabilities, despite all external limitations. For the operation of neural networks, cascades of video cards are usually used, comparable to mining farms. After the turmoil in the cryptocurrency market, there is an excess of video cards, in the secondary market there is a blockage in some places. Of course, we don’t see computers from nVidia directly specialized for such purposes, but we are quite capable of assembling a cluster of commercially available video cards for AI that recognizes images.

2. Forecasting ambush sites


Murphy's Laws of Combat say that professional soldiers are predictable, and the biggest danger is unpredictable amateurs. Training of professional military takes place according to certain standards using specific methods. Techniques systematically drive in a certain model of behavior in a given situation. And so for a long time. Only amateurs improvise.

In addition, any initiative in the army is severely punished, especially in ours, and indeed any domestic civilian will say that, as a result, the initiative strongly loves the initiator. As a result of these factors, a certain pattern begins to be traced in the actions of the military. In machine learning theory, this is called behavior patterns. If in a simple way, then templates. Often noticeable even in everyday life.

This is especially true for our military. Since the fundamental basis of the combat training of the domestic military is marching in step and singing songs in chorus on the move or on the spot. It is enough for the enemy to turn on “Travel March No. 1” at full volume, as, according to the conditioned reflex developed by many years of training, our fighters will begin to stagnate and pull their legs with a go-ahead. Reflexes will work, and they will not be able to do anything with themselves.

The apologists of this pedagogical method are blindly sure that in this way an invisible telepathic connection will be established between colleagues at the quantum level, which cannot be drowned out by any electronic warfare on radio tubes with a radioactive cathode. Otherwise, one cannot explain the antediluvian state of the means of communication in the troops, which are irrevocably lagging behind civilian models. Someone has to object?

I will answer using the example of a twenty-year-old Sony Ericsson T100 phone. Already in those years, the radio module of this simple and compact phone automatically tracked and selected the appropriate cell tower according to the signal level, jumped over frequencies and channels when switching between base stations without manual intervention. Audio codecs provided sufficient sound quality, while compressing the signal, providing noise immunity and speech masking.

It had a graphic display and a very user-friendly menu. Over the years of work, he has experienced more than one fall, flights of the body from a bicycle and with a bicycle, many days of forty-degree heat and twenty-degree frosts. The only consumables were batteries and SIM cards. It is still in working order. Now give an example of a similar one from the modern in the domestic military mobile communications. Such is the ambush with the connection we have.

Let's get back to armed ambushes.

Any ambush made according to the canons (read - patterns) of military art is usually done according to certain rules and has specific restrictions. Restrictions limit the possible options for the applied weapons, the size of the ammunition load and the number of fighters. Well, a professional in his right mind and intelligible mind will not ambush in an open field from a howitzer with five shells and a platoon of unarmed soldiers. Although the unpredictable domestic history keeps a lot of such examples, and modern too, and cases with an unarmed column in an open field without cover and at least some kind of support in general have become textbooks.

With modern means of observation, this absurdity will be detected quite quickly from the air or space, and then they will be quickly butchered from the ground. A professional sniper will not hide on a children's swing or carousel, unlike a crazy amateur. Due to these factors and based on experience, AI can be trained to predict possible ambush sites.

Every ambush is associated with certain preparations. At this stage in the development of science, armed fighters at a certain point in space cannot yet materialize out of nowhere, unlike distant and very distant relatives when dividing the inheritance of the deceased or funny rear characters who check their combat readiness.

Heavy weapons and ammunition are brought in by vehicles, while preparing an ambush, certain earthworks can be performed, and these actions leave traces behind them. The technique leaves a track and a thermal footprint on the surface of the earth, earthworks are visible from the air. And if we combine the data of ambush location patterns and indirect signs in the form of traces, then it is possible to predict possible ambushes with some certainty. To train AI in finding such places, you can take the available data on ambushes that the enemy used. And on their basis to train AI.

When we know the weapons and their capabilities used to organize ambushes, the patterns in which ambushes are organized, we can expect them to appear in a specific place with some certainty and act in advance. Process a possible place in advance so as not to tempt fate, and keep an eye on it just in case.

To some, this whole idea with patterns of behavior and forecasting may seem like naive fantasy and an impossible idea. Every second around the world, websites and search bots collect statistics on user behavior on websites and in mobile applications. It would seem, what is the use of what is there, how many times the user clicked and where did the link go?

True, on the basis of this data, analytical companies build a digital user profile for subsequent sale to advertising or marketing companies. And those based on these models build their shares and implement strategies. IT giants are practically government organizations, they can share a digital profile with a user's social graph with their beneficiary.

Just before the start of the Sochi Olympics in decent England there was a small scandal. One chain supermarket, following the shopping of one young girl, sent home a catalog of products for pregnant women and offers with discounts on products for expectant mothers. The parents of this high school student were seriously excited and angry, they went to sort out the desecrated honor of their modest girl. A public scandal arose, and the showdown began.

It turned out that in the system of analysis of purchasing activity, a pattern was triggered that is typical for women who find themselves in a position. We all know perfectly well that at such moments the taste, nutritional and behavioral patterns of women change. Here, a system was trained on them, and, as it turned out, the system is much more accurate than parental control. Who cares - the schoolgirl graduated from high school in the status of a young mother.

Now it's time to think about tanks.

Do it somewhere. Probably, there are no exact figures to understand how many tanks were destroyed in tank battles, and how many were destroyed by ATGM crews. Perhaps the percentage of those shot down from anti-tank systems is not small, but not only tanks get from these calculations. This is a very formidable weapon, despite its small size.

As I wrote above, the sites of many ambushes can be predicted with varying degrees of certainty and checked for accuracy using artillery or the tank guns themselves. Maybe aviation help. Or maybe just the crew will not climb there, knowing the potential danger.

3. Patterns of operation of communication devices


Satellites have long been able to listen to conversations on the ground.

Prior to Gorbachev's rise to the presidency of the USSR, imported spy satellites had the ability to listen in on land-based telephone conversations. Well, those with a disk dialer, for those who remember. Now few people use wired phones with push-button dialing, most have switched to cell phones that work via radio. Our army, as far as possible, tries to keep up with the times, and gradually, thanks to cheap radio stations from Ali, it is switching from wired communications to radio. The enemy is using with might and main modern imported means of communication with fantastic capabilities by our standards.

We cannot influence the supply of modern means of communication. For the most part, you don't need much. One can benefit from such a deep penetration of foreign communications. Well, firstly, to distinguish - where is your own, where is someone else's. Where it is lilac and miserable, then ours is toiling, and where it stably jumps over frequencies and maintains the availability of the communication channel with all its might, therefore, not ours.

According to the patterns of operation of communication facilities, it is possible to determine where are ours, where are strangers, and thereby supplement the existing picture from the same observation satellites.

In addition to the patterns of communication means, there are patterns of using these tools. During the offensive, one can identify some patterns of using means of communication, others during defense, and still others for ambushes. Preparatory situations for this all also fit into certain patterns. It is not necessary to decipher the contents of the conversation. With even simple encryption, this is a lengthy operation. It is enough to observe the means of communication and the patterns of their use.

The notorious US Internet traffic tracking system "Prisma" did not decrypt the data. It simply correlated data about who, to whom, when sent a message, and the size of this message. When there are several dozen such messages, they are of no use as from a goat's milk. But when there are thousands of such messages, it is already possible to build fairly clear assumptions. Here, as in quantum mechanics, we have a lot of random events and, when a sufficient number of them are collected, we can make certain statements. That is why in government communication channels, when a communication session is terminated, pink noise or recorded meaningless snippets of phrases are fed into the channel.

Many things and phenomena can be digitized, patterns of behavior can be identified and, based on them, work ahead of the curve. The same applies to the enemy and his actions. There are also people from blood, flesh, personal problems, certain cultural customs and peculiarities of thinking, as a result - peculiarities of behavior. These features can and should be used.

4. Guidance systems


Earlier I wrote about the use of kinetic weapons launched from spacecraft to combat large aircraft and swimming vehicles. Many should have had a natural desire to put a curb handle or a crooked key to their temples and twist it well. It is quite natural if the control of such devices is carried out in the old fashioned way with high delays, costs and mandatory errors in the execution of the human factor. If an order can be ambiguous, it is sure to be misunderstood. Murphy's laws work great in combat conditions.

Any of the most modern and sophisticated transport aircraft cannot make a quantum leap and leave the space-time continuum in an unknown direction. Such feints are usually subject to budgets and wives who have seized on their husband's salary card. So, all possible states of an airliner are limited by certain boundaries in the phase space of states. He is able to circle as he likes and do aerobatics available to him, only they fit into the models of possible behavior, which are limited by the phase portrait of this aircraft, and therefore are predictable. Unless a higher power from outer space helps him.

With a fairly high probability, you can predict where the board will be in five, six, ten, fifteen minutes, half an hour. The transport worker has much fewer initial conditions describing its possible behavior compared to weather conditions. Forecasters seemed to have mastered making more or less accurate weather forecasts for the day ahead and almost approached the shamans in this field.

Returning to the transport workers in the sky, based on the initial data, we can calculate the optimal moment for launching a pebble with greetings and best wishes on the first space mission. Or on the second.

Here a large transport worker took off, gained altitude, took up its echelon. Where can he go, what will the crew perform so suddenly? Give up on the Seychelles, as ours once amused? It is unlikely. Either fall through the ground or turn back. Calculating the probabilistic trajectory of a transporter without any AI is not such a difficult task as to cause a combinatorial explosion. The possible states of the liner on the basis of its phase portrait in space are known with a certain accuracy, and on the basis of these data it is possible to choose a suitable spacecraft from which respect and respect will be sent down at one or another space speed.

In the event that a space hotel is discovered, and sooner or later everything will come to this, a transporter or a large long-range bomber has nowhere to go. Jumping from one echelon to another is pointless. A blank at a speed of 10 km / s will not play a special role, that the board will sharply drop the height from the conditional 5 m to 000 m.

Well, two seconds longer will fly. Trying to sharply maneuver and twist the loops is also not an option. We must think that it will be more dangerous for him: when a blank flies into him from above or the crew tries to create Pugachev's cobra. In addition, such aircraft fly in an area limited by their air defense system and sometimes follow each other, and for these reasons they do not have such room to maneuver. If he escapes outside his own air defense system, he can easily and warmly be met by a completely different air defense system.

Of course, this is possible with a sufficiently large number of spacecraft in orbit. If their number is comparable to the number of UAVs in our army, it is not necessary to rely on operational detection, and even more so on intercepting a target on Earth. In addition, it is necessary to take into account possible emergencies that simply love to happen to our spacecraft, and keep the satellite constellation in excess. And it is very desirable that the filling of the spacecraft be on a modern elemental base, and not from what was lying around in the warehouse.

5. Serious scientific work


I am well aware that my writing is fantastic, at times counterintuitive and mostly naive. Only more educated and literate people did not sit idly by and defended doctoral dissertations on the topic. If you still have strength left after reading, you can search for a work available in the public domain called "An Unsupervised Machine Learning Tactical Inference Generator". There are more serious things described than here. In general, D. Ezra Sidran is quite a prolific author on this topic.

Afterword


It's probably a tradition here. In which case, make historical references to the times of the world wars and give examples of the heroism, courage and unpretentiousness of the soldiers of those times. For everyone who wants to do this again in the comments, I ask you to recall the school curriculum of the Russian language and literature with the wonderful poem “Borodino” by M. Yu. Lermontov. If in those days he wrote for his contemporaries “heroes are not you”, what can we say about the current generation? As soon as you feel like writing about how it was before, read the beginning of the poem "Borodino" and think about these lines.

For example, I can tell a story about grandfather Vanya (a veteran of the Great Patriotic War, Ivan Dmitrievich). On a dare, he took the M12 lerk in his hand and cut a thread on a steel bar without a wrench and all sorts of adaptations. Such a man had power in his hands. The current generation of young people with great difficulty imagines what a lerka is and why a tap is needed.

In the commissariat, now the military commissar, in the presence of his mother, gives the recruit two sim cards and a bank card so that parental chats in instant messengers and groups of unfortunate people in social networks sleep peacefully. Therefore, I express my thoughts based on modern realities. And they are such that some heroes of cyberspace can hardly build communication in real life. A kind of pronounced introverts with the habits of a sociopath in one bottle. Are there commanders among the readers who can compare the conscription of the early 2000s and the modern conscription? Tell me, which are more frostbitten?
Our news channels

Subscribe and stay up to date with the latest news and the most important events of the day.

43 comments
Information
Dear reader, to leave comments on the publication, you must sign in.
  1. +9
    3 March 2023 05: 30
    Inet is still a spy ... he exposed his personal data there, then bankers and advertisers will not leave him alone for the rest of his life, even worse when this data gets to Ukrainian call centers, they milk our gullible citizens like on a conveyor belt.
    It is interesting that our generals stuck in the 20th century are interested in the novelties described in the article, are they able to digest in their heads all the novelties of enemy military science for the benefit of our cause?
    In China, some students managed to deceive the AI ​​in the video surveillance system by pulling photographs of strangers over their faces ... a waste of fiction is cunning.
    1. +12
      3 March 2023 06: 14
      AI would not interfere with us now, in some Moscow institutions. Maybe things would have gone better, otherwise there was absolutely no hope left for the existing pension antillects, and the road is excessive and the harm is immeasurable. The most pressing problem for the country right now.
      1. +11
        3 March 2023 08: 05
        From the last. A car brand "Ford" in case of delay in payment on account of a loan will remind the owner that he did not make another installment through the bank - for him, the car. If the owner does not respond, the car will not open the door, and at the next delay in payment, it will generally leave itself for a fine parking lot.
        Not sure if it's a joke or not. Who will buy such "independent" cars? )))

        And the article is good, written with humor, I liked it. Author - thanks! hi )))
        1. +13
          3 March 2023 08: 55
          And the article is good, written with humor, I liked it. Author - thanks!
          I join unconditionally!
          Not sure if it's a joke or not.
          In every joke there is only a fraction of a joke)) And this artificial intelligence must be taught not to be afraid to take more responsibility - otherwise if you forgot to turn off the light in the cabin, then after closing it starts to blink. That is, you see, the car sees that the light not disabled. And what does she do? Turns it off? Fig there! She blinks at them! Yes, turn it off if you're so smart! You see that the owner forgot))
      2. +1
        8 March 2023 06: 28
        pension antillects are available because there was a generational gap in the scientific community, including among the teaching staff of higher educational institutions, because my age and a little older than 40-45 in the 90s and early 2000s were massively dumped on the blessed lands of light elves, and specifically engineers/scientists. Our bohemia was not needed there, and the humanities there washed toilets or cleaners.
    2. fiv
      +14
      3 March 2023 09: 40
      They were recently given several tablets (domestic, of course) with increased mechanical strength and tightness. You can put tea, coffee, and according to some reports, even cognac on them.
  2. +13
    3 March 2023 08: 14
    The article is great. It's a pity that few people are interested. If you don't want to comment, please upvote the author. And it will turn out, as with drones, they also ignored, ridiculed. now suddenly all the experts.
    1. +15
      3 March 2023 09: 11
      Who made fun of? And, I remember, I remember - "witnesses of the almighty rab sect." Yes, yes, during the Armenian-Azerbaijani war, many asked "how are we doing with this matter," for which they were minused with condescending smirks - "what the hell are these drones for us, we won't even let them take off as our almighty slave." And in the morning they woke up. In general, I agree - stop preparing for yesterday's wars, we must prepare for tomorrow's, if we don't want to throw off the whole world again on modern devices. But this is not for citizens, but for officials, this is already a question, they manage the budget .. We have already seen how effective it is. It is necessary to restore order in this area and introduce such a measure into the law as the confiscation of "honestly acquired property through overwork", otherwise they do not understand.
      1. +13
        3 March 2023 09: 23
        And I found more)))
        In the West, many books appeared on sale, written by AI and only slightly edited by the authors who gave the machine a theme, plot and character names.
        Jokers say: "And who will read them? Toasters?" wassat )))
        1. +14
          3 March 2023 10: 15
          Have you heard that rules have been developed for artificial intelligence so that it does not harm people? It would be nice to have such rules for officials. And then, having learned that a billion was lost due to artificial intelligence in Sberbank, they ordered the introduction of artificial intelligence in all divisions of Roscosmos. In general, in vain we are afraid of losing our jobs because of artificial intelligence.
          No intelligence will work for 15 sput))
          1. +12
            3 March 2023 10: 41
            "Three Laws of Robotics" by Isaac Asimov.
            This is a collection of short stories, surprisingly easy to read literature. Lots of humor. And yet, everything is clearly outlined. My old book of the Soviet edition fell into fragments, and I, not knowing how to glue sheet by sheet, keep it in a plastic bag as a precious artifact)))

            There was also the first case of earning a journalist with the help of artificial intelligence. The journalist undertook to write a custom article, and in a few seconds the article was ready. The employer was pleased, and the journalist received a check for $600. This case was described in the news feeds two or three weeks ago.
    2. +9
      3 March 2023 14: 29
      The article is great. It's a pity that few people are interested.


      Hello Gardamir!

      She was interested, maybe many. Just to comment on such articles, you need to be at least a little in the subject. But even in this case it will be difficult to object to the author or to supplement the article.

      In general, the author is a great fellow. The deepest respect to him for the sensible analysis of very tricky questions.
      I somewhat disagree only with the fact that the development of neural networks in Russia is in full swing. There are quite working networks. In addition to banking or advertising applications, for example, early medical diagnostic systems are being exploited - this was done by Rosatom, Safe City complexes, etc. Yandex announced preparations for the imminent launch of a chat bot like the sensational ChatGPT, etc. Yes, and cruise missiles have long been using a similar method for correction. True, the elemental base is there ... peculiar.

      There are also developments of neurochips.
      Motiv, with the participation of Kaspersky Lab, is testing a trial sample of a new neuromorphic processor.
      The Altai processor consists of 256 neuronuclei, forming 131 neurons with 072 inputs each, which together gives more than 512 million synapses. Process technology 67 nm. Performance - about 28 billion actions per second. A single neurochip in terms of energy efficiency bypasses modern graphics accelerators by orders of magnitude - power consumption is only 67 W.
      "Altai" can process up to 2200 video frames per second. The chip from Intel with similar power consumption parameters (0,65 W) delivers performance at the level of 11 fps, while the competitor from Nvidia consumes 779 times more power (30 W) at a performance of 15 fps. For the Russian processor, proprietary software has already been developed (including DDK and SDK), which will greatly simplify the integration of the neurochip into modern systems.
      The updated version implements the ability to scale using the technology “Neurochip → neuromorphic accelerator (16 chips on the board, the ability to connect directly to a computer) → backplane (up to 16 accelerators, connection via USB 3.0 interface)”.
      Under the auspices of the same Rosatom, the Modul company is developing three neural processors based on a 5-nm process technology and its own NeuroMatrix architecture for embedded systems, server solutions and electronic gadgets.
      By the end of 2024, the company intends to start mass production of the Neuro-B SoC with a power consumption of 12-20 watts. It includes a control CPU with eight RISC cores operating at a frequency of 1,5 GHz, as well as a NeuroMatrix neuroaccelerator with 8 NMC5 cores at a frequency of 1 GHz. Claimed areas of use for this processor: medical equipment, unmanned vehicles, smart cameras, drones.
      In 2025, it is planned to create a Neuro-C chip with a power consumption of up to 200 W, consisting of a control processor with 12 RISC cores at a frequency of 2 GHz and a NeuroMatrix neuroaccelerator with 64 NMC5 cores at a frequency of 1 GHz. The chip is focused on cloud computing and ANN training.
      In 2026, it is planned to introduce the Neuro-M SoC with a power consumption of 1–2,5 W (a control processor with four RISC cores operating at a frequency of 1,5 GHz, a NeuroMatrix neuroaccelerator with one NMC5 core at a frequency of 1 GHz) for low-power electronic devices.
      Regrettably, the pilot batch of Altaev was riveted by Taiwan, which refused further cooperation due to sanctions. And if the 28-nm process is still possible to launch in Russia (in the presence of direct budget funding and the absence of bureaucratic obstacles, extortion and sabotage) (although production will be guaranteed to be unprofitable due to the lack of a domestic market with a capacity of billions of chips), then how will the “Module ” with their 5nm projects is hard to even imagine.

      Best regards,
      hi
      1. +5
        3 March 2023 22: 36
        Just yesterday I finished reading a good book of 2002 from the MPEI publishing house about neural networks and neuroprocessors. And there, in the list of produced neuroprocessors, there were domestic ones. But where were they all the time and why didn't they develop further? This is reminiscent of the situation with microwave transistors. According to the directories, they are and they seem to be quite good in terms of parameters, but it is unrealistic to find them. It is easier and sometimes cheaper to buy abroad.
        I am pleased to know that our people have not abandoned this topic and are somehow developing it. The example of the chip you gave is interesting, only how close it is to the famous IBM neuromorphic processor, which they released back in 2014. Now I don’t remember exactly, it seems a year earlier they released a classifier made on optical elements with amazing possibilities of parallelism in processing.
  3. +8
    3 March 2023 13: 02
    Die eigentliche, höchst gefährliche Idea, ist die vollkommene
    Vernachlässigung NATÜRLICHER INTELLIGENZ zu Gunsten
    einer höchst fragwürdigen und geistigen Versklavung der Massen
    fördernde Gleich-Schaltung des Denkens, mit fremdgesteuertem
    Reset-Knopf auf Basis einer als "künstliche" Intelligenz getarnten
    Maschinen-Welt in den skrupellosen Handen der ewig gleichen
    Machtgierigen Eliten, die DANN endgültig die totale orwellsche
    (Georg Orwell 1984) Kontrolle über ALLES und JEDEN erlangen
    konnen...!!!!!

    Diese todliche Gefahr ist real und es ist die Pflicht jedes einzelnen
    INDIVIDUUMS sich nicht in diesen Moloch aus a-sozialen Netzwerken
    und imaginären, KI-gesteuerten Wunsch-Welten versenken und
    versklaven zu lassen...!!!
    Es lebe die Individualität, die Kreativität des Einzelnen und die Fähigkeit,
    WIDERSTAND zu leisten um diese menschenverachtenden Pläne zu
    vereiteln und deren Akteure mit aller Macht zu zerstören...!!!!!
    1. +9
      3 March 2023 13: 47
      I agree, colleague, I agree)))
      But what to do with this? Last week, the news feed published the "faces" of ancient and ancient Russian cities, created by artificial intelligence. That is, the city in the form of a man. And I will tell you this: wonderful, truly fabulous images! Very very!
      AI did not make a mistake anywhere, treated our history very carefully, with a share of romance.
      By the way, the artists began to fuss and massively began to remove their paintings from the Internet. After all, it is on them that Artificial Intelligence learns. The illustrators were especially excited. But I agree. So far, AI has not been able to replace humans in creativity. After all, he began to study recently, and his abilities are at the level of a small child, he is still a child.
      But evil programmers have already threatened, they say, after a year and a half of learning on the Internet, AI will surpass any person in mental abilities. And even all of humanity, they threatened...
      And this is where I burst out laughing!
      Aggregate humanity is not difficult to surpass, because it itself is at the level of a small child, and even capricious, sometimes stupid.
      Only a few people are smart.
  4. fiv
    +9
    3 March 2023 13: 36
    Cybernetics
    (In Concise Philosophical Dictionary 1954)
    CYBERNETICS (from another Greek word meaning helmsman, manager) is a reactionary pseudoscience that arose in the USA after the Second World War and became widespread in other capitalist countries; form of modern mechanism.
    ...
    Thus, cybernetics identifies mechanical, biological and social relationships and patterns. Like any mechanistic theory, cybernetics denies the qualitative originality of the regularities of various forms of existence and development of matter, reducing them to mechanical regularities.
    ...
    Essentially, cybernetics is directed against materialistic dialectics, modern scientific physiology, justified by IP Pavlov (see), and the Marxist, scientific understanding of the laws of social life.
    This mechanistic metaphysical pseudoscience gets along well with idealism in philosophy, psychology, and sociology.

    Cybernetics vividly expresses one of the main features of the bourgeois worldview - its inhumanity, the desire to turn the working people into an appendage of the machine, into an instrument of production and an instrument of war.

    At the same time, cybernetics is characterized by an imperialist utopia - to replace a living, thinking person fighting for his interests with a machine both in production and in war.

    The instigators of a new world war use cybernetics in their dirty practical deeds.
    1. +1
      3 March 2023 21: 21
      As a result, the machine successfully replaces people even in those professions that until recently seemed difficult to automate. As long as people are cheaper than cars. I think it's time to admit that in the next hundred or two years, Homo Sapiens in its current form will go to the dustbin of history. If not before.
  5. +7
    3 March 2023 14: 17
    AI based on artificial neural networks (ANN) operates with templates in the same way. Only the set of templates for INS is incomparably larger than a person can keep in mind. Well, a computer with ANN-based software is tireless, which is why it outperforms a person when processing large amounts of data.
    And the ANN algorithms are not that complicated, the mathematical apparatus has already been worked out. A team of students should be assembled, set a competent task, and provided with the necessary data in the form of satellite photographs. And there will be a hitch. Are there enough of them, with the required frequency of shooting, with the required resolution?
    About patterns. That would be someone who came up with the pattern of the aggressor country. It should be taken into account in different countries the number of military bases in the world, their dispersal, military budget, involvement in military conflicts in the world, and many other direct and indirect parameters. What country in the world do you think would most fit this pattern?
    1. +2
      3 March 2023 23: 00
      A team of students should be assembled, set a competent task, and provided with the necessary data in the form of satellite photographs. And there will be a hitch. Are there enough of them, with the required frequency of shooting

      There is no need to collect anyone. Many domestic financial institutions have long appreciated their capabilities and have been using them for several years. Not without failures and absurdities of course. Well, I traded in the minus for 2 yards of AI dollars of one green bank. It happens to everyone? He is still far from the famous 50 yards of the euro from Societe Generale performed by a man.
      As for the availability of pictures of the area in sufficient quantities, I can’t say anything. Our satellite constellation is provided in a vanishing amount, probably coinciding with the number of UAVs in the army. Some devices appear to be in a deplorable state. In the first part of his opus, he just walked over this topic and the fact that the militarization of space is a logical next step in the development of military affairs.
      By the way, it was very good to collect data from drones and upload them for processing and image recognition. Thereby supplementing the existing picture from satellite images.
      At the same time, I perfectly understand that for the next few five years, this will be done by AI based on fighters. So to speak, they will work off the maternity capital.
  6. +2
    3 March 2023 17: 08
    Something the author has gathered together
    In some banks, the decision to grant a loan is no longer made by a girl with a calculator and 1C courses, but by an AI-based system.


    there is no AI here, the usual scoring system based on statistical methods is often used on "microcredits" or as a rough input filter, an employee, or even more than one, is behind the decision to issue more or less serious amounts, in the case of mortgage loans, there are several employees make the decision, plus individual supervisors/moderators conduct checks. And after the "triumph" of chartGPT, there was some kind of unhealthy hype on the Internet about AI (although this is not AI, there is no understanding of the "context").
    1. +2
      3 March 2023 21: 57
      In one bank, I won’t say what color, but the advertisement of this bank, beloved by grandmothers, is often spinning, so they wound up AI until all sorts of chartGPT appeared. I can't say how it works in others.
      There is data about antifraud. There AI plows for a long time and thoroughly. Although some things could be solved faster with digital filters of the third kind.
      1. +1
        4 March 2023 11: 29
        I will just repeat the scoring systems are not AI, well, or you should decide what you mean by AI, the same with antifraud.
  7. +1
    3 March 2023 20: 02
    A funny feuilleton turned out. True, in some places it looks like a pamphlet. But the mood is uplifting. As for AI decision making, I don't think it will ever happen. Man and his brain are very imperfect, or perhaps incomplete.
    But, nevertheless, the world around us is infinite in every sense, and our knowledge about it is finite, which is 0 in comparison with infinity. This also applies to the environment, and to the person himself and his brain. And the possibilities of the brain cannot be fully explored. You can "disassemble" a living organism for study, but you can't "assemble" it back - something very important is lost.
    And AI will always be a product of the brain, and more of an enhancement and empowerment than a replacement.
    1. +3
      3 March 2023 21: 23
      Yeah, and also organic substances cannot be synthesized artificially, as vitalism teaches us, right?
      1. +1
        4 March 2023 10: 28
        And what about this? Someone created at least one living creature in the laboratory? Not grown, not cloned, namely, created, from scratch? At least one living cell? So far, laboratories have created "conditionally living" cells using different biomaterials.
        1. +1
          4 March 2023 12: 59
          And here it is? Yes, and here it is.
          Quote from shikin
          And the possibilities of the brain cannot be fully explored. You can "disassemble" a living organism for study, but you can't "assemble" it back - something very important is lost.

          Once upon a time, the synthesis of organics was described in exactly the same way. I couldn't help but draw a parallel. Science will study everything, give it only the will, time and means.
          Is it worth the task to make a living cell? Well, just what will be gained from this? Experience? OK then. But how to use it, if it will still be easier to clone and edit ready-made cells? If nature does not give something, then we do it ourselves. But if she gives out something, and even in excess, then why strain?
          1. 0
            4 March 2023 13: 05
            Do you seriously think that it is possible to know the infinite to the end? And measure infinity? Do you think that we know (and will know) all the laws of the universe?
            1. 0
              5 March 2023 02: 42
              Do you think that we know (and will know) all the laws of the universe?


              It's just a matter of time, if our civilization exists for another million years, then our galaxy will definitely be explored up and down at least.
              1. 0
                5 March 2023 12: 06
                I have a feeling that you do not understand what infinity is and believe that there is a "last horizon". At one time, one smart person said that the microcosm is as inexhaustible as the macrocosm.
                1. 0
                  5 March 2023 22: 05
                  I repeat in mathematics, we can work with infinite sets, but if we return to the laws of the universe, it is not yet clear what prevents us from knowing them all if we are talking about hundreds of thousands or millions of years?
                  You very unfoundedly called the human brain imperfect, although at one time the same Maxwell, by the power of such an imperfect brain, on paper, by pure mathematics and imagination, “deduced” electromagnetic waves, the existence of which Hertz later experimentally proved, this is only one example of knowing the laws of the universe. Moreover, it is likely that the laws of the universe work “approximately” lonely everywhere and the hydrogen atom in us and in another galaxy hundreds / millions of light years away has the same properties, that is, you understand what I mean, we will not need to explore every “corner” of the universe.
                  1. 0
                    6 March 2023 10: 35
                    You understand infinity, apparently, exclusively in the spatial sense. And infinity is in everything - in the laws of the universe, in the fields, in the microcosm, in the awareness of the surrounding world. Man and his brain are evolving, this evolution is not completed and cannot be completed as long as humanity exists. Man's knowledge of the world is limited by his sense organs (especially at the beginning of knowledge) and his own ideas. Truly new discoveries that expand the horizons of knowledge and change the previous picture of the world are extremely rare, the rest is a compilation of what is already known. And the process of cognition cannot be finite.
                    Look back at 100 years, 200, 500, compare knowledge. But biologically man has not changed over the past thousand years.
                    And about the "corners of the universe" - are you sure that the levels of the universe are not an infinite number? What do we know today? Multiple levels?
                    1. 0
                      6 March 2023 10: 59
                      You understand infinity, apparently, exclusively in the spatial sense. And infinity is in everything - in the laws of the universe, in the fields, in the microcosm, in the awareness of the surrounding world.


                      You are apparently not a mathematician, so infinity scares you.

                      Truly new discoveries that expand the horizons of knowledge and change the previous picture of the world are extremely rare, the rest is a compilation of what is already known. And the process of cognition cannot be finite.


                      Over the past 200 years, there have been enough such discoveries, electricity, radio waves, nuclear fusion, the photoelectric effect, Einstein's work, jet propulsion, etc., and this is just for 200 years, let me remind you that if our civilization lives the same way, developing at least another hundred thousand years, we (mankind) will get to almost all the "laws of the universe."


                      And about the "corners of the universe" - are you sure that the levels of the universe are not an infinite number? What do we know today? Multiple levels?


                      This is not a constructive dialogue, so you can come to Russell's teapot, if you make a statement that there are an infinite number of levels of universes, then you need to back it up with something.
                      1. 0
                        6 March 2023 11: 53
                        I think these questions are rather philosophical - ideological.
                        Question: humanity has known the universe, all the laws of the universe - what's next? End of evolution?
                        At least, this question in relation to space is more obvious: we have approached the boundaries of the known - what's next?
                      2. 0
                        6 March 2023 12: 19
                        I have no idea why ask such questions now, when we come to such a period, then it will be clear, maybe evolution will stop, maybe we will create new laws ourselves, we still have to go (my completely rough estimate without justification) for a hundred thousand years.
                        If you dream up, then most likely for a “jerk” humanity will first of all have to solve the problem of aging, since after all 30-40 years of active working life is very little (if we are talking about galaxies, universes), such an option could be the creation of AI in iron and, accordingly, the transition from protein life to cybernetic.
                    2. +1
                      9 March 2023 01: 34
                      There are many different infinities in mathematics, each with its own power. There is such a term for infinities in mathematics. The cardinality for the set of real numbers differs from the cardinality of the set of complex numbers. And there are also hypercomplex and p-adic numbers. In Lie groups and fractals, in general, horror is going on. So let's leave mathematical infinities alone. And physical too.

                      And the process of cognition - can not be finite

                      In fact, there is no mutually unambiguous and precise definition of what knowledge is. As there is no complete understanding of how neural networks work. To say when something is over that we don’t know is not quite the right approach. Khrushchev promised to build communism in 30 years.

                      But biologically man has not changed over the past thousand years.

                      Unfortunately, it became a highly social animal, an upright primate, sometimes with a pronounced herd behavior.
    2. 0
      5 March 2023 02: 51
      Man and his brain are very imperfect, or perhaps incomplete.


      What do you think is imperfect? Any current living species is perfect, it has won the toughest and most uncompromising competition of “evolution”, millions of species have become extinct, all who are now winners, there cannot be imperfect or incomplete.

      But, nevertheless, the world around us is infinite in every sense, and our knowledge about it is finite, which is 0 in comparison with infinity.


      The series of natural numbers is infinite, but this does not prevent us from studying these series, their properties, and finding patterns.

      And AI will always be a product of the brain, and more of an enhancement and empowerment than a replacement.


      Real AI (with understanding of the context, possibly emotions and desires) implemented in hardware will simply be the next evolutionary step, we, as protein organisms, will simply lose the evolutionary struggle to cybernetic creatures.
  8. +1
    3 March 2023 23: 10
    Quote: depressant
    Aggregate humanity is not difficult to surpass, because it itself is at the level of a small child, and even capricious, sometimes stupid.
    Only a few people are smart.

    Here you are right...damn right...and these whims and stupid actions bother me a lot. hi
  9. +1
    4 March 2023 07: 14
    Author! Thanks for the great posting!
    While reading, I tried to remember what I disagreed with. And a lot of things caught on and urged to argue ... I realized that it was necessary from the very beginning to take notes or just not bother ...
    The strongest stimulus is the word "pattern".
    1. +1
      9 March 2023 01: 37
      The strongest stimulus is the word "pattern"

      I tried to keep the article in terms that AI and machine learning specialists work with.
  10. +1
    4 March 2023 12: 11
    Thinking about the potential military applications of these technologies, you are right, but choose the wrong viewing angle. Our country may be in a good place in the world for the development of such systems, but here is the large-scale production of "iron" and "stones" for them - we can say that we are not developed. The things you describe imply that you will need horse-sized volumes of calculations and horse-sized throughput lines for them, huge databases - and for all this you will need DCs, expensive and fancy. And yes, fortified - because all this is military or double infrastructure.
    All this is $$$, and in addition to money, it is also a hell of a lot of work to get the opportunity to do all this "at home". Now this is a matter of viewing angle.
    Because while we don’t have this, the potential enemy is much closer to this. And he can do it and has the resources to do it. He is closer, much closer to this than we are.

    So it’s time for us not to think about how we could use such intelstraps, but about what options can be used to protect against them in the foreseeable future.
    After all, an AI that tells where an enemy ambush can be can very well tell where this ambush can be organized (and how), and why are there people there? After all, there are wonderful tools in the form of remotely controlled modules.
    The same SWORDS robot paired with a well-thought-out system for landing it in the potential ambush area.
    1. +1
      9 March 2023 02: 48
      Let's continue to scoop up oil and gas for sale and mine ore for remelting into metals in order to sell it too. If they don't interfere with it. I mean sell. And then we will try to buy something with the proceeds of $$$ through third countries.

      The things you describe imply that you will need horse-sized volumes of calculations and horse-sized throughput lines for them, huge databases - and for all this you will need DCs, expensive and fancy

      Return from the days of dialup and rotary phones to the current realities. About 7 years ago, the notorious automotive company Tesla announced its specialized chip with AI elements for its autopilot. I know that this autopilot is not quite complete yet and sometimes requires human intervention. But we do not expect this in the near future. This chip allows you to recognize images from car cameras, process radar data and control the car in heavy traffic and in real time. The finished module is assembled in the size of an A5 diary. The well-known company Apple rented 5 PB of disk storage from Google. There may be more now. On this storage, the apple company stores media files of its users. For understanding, 1 PB is a storage for two server racks plus a small rack for network equipment. It is smaller in size than the legendary and modern domestic system for counter-battery combat based on the Kamaz vehicle. nVidia offers its computer on a specialized processor for the needs of AI the size of a large system unit. Even three such system units do not take up much space. I just don’t think at all that you will have data from a fifth or tenth of what Apple has. Not a single satellite in our country will give out video with a resolution of 4k and a frequency of 120 frames / s. The situation is similar with drones.
      I remember with what memory and pomposity we have been reporting through the central channels for a decade about the construction of a new gas or oil pipeline, about powerful gas pumping stations. The money received was again invested in mining and incomprehensible projects such as maternity capital or the Internet in each school. Why the hell did the mother capital give up if there are no kindergartens nearby, and those that are are overcrowded? What's the use of the Internet at school when there are not enough teachers? And what have we had all this time with the development of our own electronics industry and information technology? How many DCs were built for civil needs?
      We officially live in the information age. Information is the new oil. And this is completely independent of the viewing angle and coordinate system. A clear adversary has long understood, taken action and is in full readiness. But we prefer to extract our own oil in the hope that someone will process the information for us.

      So it’s time for us not to think about how we could use such intelstraps, but about what options can be used to protect against them in the foreseeable future.

      We still ride on armor, because there the probability of surviving is higher than under armor. We have many models of equipment before firing, they must stop and aim. Because there is still no barrel stabilization system, nor normal aiming mechanisms. Do we have at least one ATGM attached to ground armored vehicles able to work when the equipment rushes over rough terrain? What do we have to protect against UAVs? It's time for us to think about where we are taking people. One more mobilization and there will be no one to work in the country. And this horde at the front must be fed and supplied. Only no one will do it.
  11. 0
    4 March 2023 19: 09
    I remember some fantastic story (maybe Harrison, I don’t remember), a planet with AI developed to the limit. Everything is calculated, all the nuances are open and recorded, the outcome of any event can be influenced, perhaps, by a butterfly that accidentally happened to be in the right place and at the right time ...
    Yes, there are wars in this world, but not in our usual sense. At some specific moment, one country declares war on another country, AI calculates all the nuances and determines the winner with a 99,99% probability. The loser pays an indemnity, and calms down until better times)
    Campaign, and we are all going to this)
  12. 0
    10 March 2023 00: 01
    The article is really good, humorous and tasteful.

"Right Sector" (banned in Russia), "Ukrainian Insurgent Army" (UPA) (banned in Russia), ISIS (banned in Russia), "Jabhat Fatah al-Sham" formerly "Jabhat al-Nusra" (banned in Russia) , Taliban (banned in Russia), Al-Qaeda (banned in Russia), Anti-Corruption Foundation (banned in Russia), Navalny Headquarters (banned in Russia), Facebook (banned in Russia), Instagram (banned in Russia), Meta (banned in Russia), Misanthropic Division (banned in Russia), Azov (banned in Russia), Muslim Brotherhood (banned in Russia), Aum Shinrikyo (banned in Russia), AUE (banned in Russia), UNA-UNSO (banned in Russia), Mejlis of the Crimean Tatar people (banned in Russia), Legion “Freedom of Russia” (armed formation, recognized as terrorist in the Russian Federation and banned), Kirill Budanov (included to the Rosfinmonitoring list of terrorists and extremists)

“Non-profit organizations, unregistered public associations or individuals performing the functions of a foreign agent,” as well as media outlets performing the functions of a foreign agent: “Medusa”; "Voice of America"; "Realities"; "Present time"; "Radio Freedom"; Ponomarev Lev; Ponomarev Ilya; Savitskaya; Markelov; Kamalyagin; Apakhonchich; Makarevich; Dud; Gordon; Zhdanov; Medvedev; Fedorov; Mikhail Kasyanov; "Owl"; "Alliance of Doctors"; "RKK" "Levada Center"; "Memorial"; "Voice"; "Person and law"; "Rain"; "Mediazone"; "Deutsche Welle"; QMS "Caucasian Knot"; "Insider"; "New Newspaper"