Artificial intelligence in the Russian army

Uranus-9 is a full-fledged combat robot with artificial intelligence

Recently, the leading armies of the world have been paying special attention to the development and development of artificial intelligence technologies. The Russian armed forces are no exception, and our science and industry, on their order, are also conducting relevant developments. Some projects have been successfully completed and their results are finding practical application, while others are still in the early stages, but their prospects are already clear.

The importance of direction

The importance of AI technologies for the development of weapons, equipment and the army as a whole no longer raises questions. Over the past several years, the top leadership of the armed forces and the country has repeatedly raised the topic of such developments. The need to create such projects and their special importance for defense were noted.

These issues were last discussed at a recent meeting with the President, Defense Ministry officials and defense industry officials. In this context, the head of state pointed out the need to create promising robotic systems with AI and an increased degree of autonomy. Such technologies should find application in controls from unmanned aerial vehicles. aviation and autonomous uninhabited underwater complexes.

Combat RTK in the exercises "West-2021"

It should be noted that AI can be useful not only in the field of UAVs and AUVs. With its use, it is possible to create modern highly efficient ground or surface RTK for various purposes and new samples of high-precision weapons... In addition, AI can find application in on-board equipment of military equipment and in command and control facilities - in these cases, it will take over part of the calculations and reduce the workload on a person.

It is important that all such possibilities exist not only at the level of general ideas. The forces of various scientific and design organizations carry out research and develop real products for various purposes with one or another use of AI. Some of these samples have already been brought to use in the military, while others are expected in the foreseeable future.

История вопроса

It is curious that domestic research on the topic of AI and the development of certain technologies of this kind for implementation in real projects began several decades ago. The Soviet anti-ship missile P-700 "Granit" is often cited as an example of the early application of such technologies. It was created in OKB-52 / NPO Mashinostroyenia during the seventies and entered service with the Navy in 1983. It was distinguished from previous domestic anti-ship missiles by a more developed control system with some intellectual functions.

Experienced UAV S-70 "Hunter"

A salvo with several P-700 missiles allows you to successfully attack a group surface target. The high efficiency of such a strike is ensured by special algorithms for the operation of individual missiles and the entire salvo. The missiles are reportedly capable of determining the composition and size of a group target, as well as identifying the type of a ship's detachment and identifying the main objects in it. Also in the automatic mode, the distribution of targets between missiles and the formation of an attack is carried out.

According to some reports, the P-800 Onyx missile system has similar capabilities. At the same time, a newer element base and improved software make it possible to increase the overall combat effectiveness. Apparently, similar developments are used in other projects of guided weapons that have already been put into service and are only being developed.

Recent achievements

To date, a new generation of AI systems has been successfully created. The first models of equipment equipped with such means are already entering service. Other developments are also being tested, which will also soon fall into the army. Such work covers several areas and is carried out in the interests of the ground forces, fleet and air force.

Tracked prototypes of the Marker program

The ground-based Uran-9 RTK developed by the 766 UPTK company has advanced farthest at the moment. After passing all the necessary tests, in 2019 this complex was put into service. Now the production of serial products continues, and the troops are mastering new equipment. During the recent exercises Zapad-2021, the Uran-9 complex was used for the first time in the same combat formations with the "conventional" equipment of the ground forces.

RTK "Uran-9" is a self-propelled tracked vehicle with a combat module equipped with cannon-machine gun and rocket-propelled armament. In autonomous mode, using AI, such a robot is capable of moving along a given route, observing, searching for and hitting targets. In this case, the decision to open fire remains with the operator.

With the use of AI, a control system was built for the promising S-70 "Okhotnik" UAV, developed by the "Sukhoi" company for the Aerospace Forces. Currently, such a drone is undergoing the necessary flight and other tests. According to known data, both the independent performance of tasks and the group use under the control of a manned fighter are being practiced.

It is expected that the AI ​​"Hunter" will fully solve the tasks of piloting with the passage along a given route and solving the assigned tasks. The UAV will be able to conduct reconnaissance and open up enemy defenses, as well as strike at ground targets. Perhaps in the future, AI will also master aerial combat.

"Marker" on a wheeled chassis

Reserve for the future

In parallel with the development of equipment for delivery to the troops, the necessary research is being carried out and a scientific and technical reserve is being created for the following projects. Some programs of this kind have already been brought to the test, and their progress is regularly reported, while others remain secret for the time being.

An important role in these processes is played by the Marker program implemented by the Advanced Research Foundation. Its goal is to create a complex of control with AI, suitable for use on different platforms. The design of such a complex has already been completed, and several experimental RTKs of different shapes have been built. Now they are undergoing various tests, incl. in real-life conditions.

Based on the results of the Marker project, a unified AI control system suitable for use in various ground-based RTKs will appear. Its presence will greatly simplify the development, production and operation of new robots, because their creators do not have to design the most complex components from scratch.

In the spring of last year, the Ministry of Defense launched a closed competition for conducting research "Research on the creation of an experimental sample of a complex for the development, training and implementation of deep neural networks for a new generation of military systems with artificial intelligence." Almost 390 million rubles were allocated for the work. and three years, results are expected no later than November 2022.

Poseidon underwater vehicle. This technique also needs AI.

The details of this project, which also received the code "Kashtan", have not yet been disclosed. Nevertheless, the very name of the research project shows what topics the military department is interested in, and what new developments may turn out to be.

Efforts and results

Thus, in recent years, artificial intelligence technologies have received all the necessary attention in our country, due to which their development continues. Developments in this area are regularly introduced into real engineering projects and reach, at least, landfills. Moreover, some samples have already entered service or are being prepared for this.

Obviously, this development will continue and again give the desired results. Already in the coming years, we should expect new military RTKs with AI to enter service, and in addition, non-military models may now appear. They will be useful to military, rescue, environmental and other organizations.

In general, the situation is conducive to optimism, and allows us to expect that in the field of artificial intelligence and related technologies, our country will maintain a leading position in the world. At the same time, one should not forget that other powers are also actively engaged in the development of AI, and their projects are quite capable of getting ahead.

Comments

[English tarantas]

Oh, the hand of a master is felt, new and useful information - one sentence, a conclusion as in a school essay. Thank you Kirill for your time.

[nikvic46]

Good technology. And great specialists created it. It is one matter of learning. The combat situation is another matter. Having expended energy, where will this technique be refueled? The dugout in the photo is not the work of AI.

[hagen]

Quote: nikvic46

Having expended energy, where will this technique be refueled?

One of the main tasks of commanders of all ranks in battle is to control the combat effectiveness of their own forces. Those. always upon reaching a certain level of consumption of l / s, ammunition or fuel, the commander must report on the command "up". This algorithm is probably embedded in the software of this "device". So there won't be a sudden empty tank or store. There will always be a message to the operator about reaching the "NZ" level for any of the combat or technical components. Considering that there is a diesel engine, the organization of refueling, as well as ammunition, will be decided at the stage of preliminary preparation for the task. Maybe in position, or maybe by withdrawing them from combat positions to the MTO area.

[Knell wardenheart]

If my memory serves me, there have long been devices that allow the technique to dig itself, there are also machines for digging trenches for a long time. Of course, it will not look as aesthetically pleasing as a man-made one, but if necessary, it will completely fulfill its function. At the moment, the autonomy of such machines is, yes, a moot point. As far as I understand, the functions of already existing samples are mainly reduced to four things - routine security tasks, one-time operations of low depth as the vanguard of the main forces (breaking through the defense with the opening of firing points, for example. At least such tests were carried out by the Americans), contributing to the strengthening of stationary defense finally ambush functions. For these tasks, some kind of increased autonomy is not required - the cars will either have the ability to periodically leave for the refueling point, or they will have access to servants at the end of their mission or in the process.
For tasks similar to the task of modern infantry, completely different vehicles are needed - for different parameters of autonomy, dimension, armament, and probably also booking / price determination. Such machines are also being developed - these are various variations of "dogs" in my opinion, they have recently begun to be armed with, for example, sniper weapons, everything goes to the point that such devices in the future can TRY to replace the infantry on the battlefield. With a huge amount of NO and IF, however.

[nikvic46]

Well, if this technique has become commonplace, then please. But I did not see a single UAV on the street. Also Vanya and Petya dig trenches, wells, trenches. Or do we want to watch all the rehearsals on TV?

[ViacheslavS]

Well, this is not the main problem, as a last resort it will simply become a stake. The question is different when they talk about AI, is it possible to use this technique when clearing the city and are not afraid that it will open fire on its own people, will not turn over on the march, without recognizing the same pile of bricks, it will not be misled by an enemy who simply put on the same uniform or what she will do if she sees the same replica, recognizes it as her own, or will destroy it. All systems are their own alien, on the basis of any means of communication can be drowned out by electronic warfare, you need to rely only on vision systems, as people actually do.

[Plate]

By the methods you describe, a person can be deceived.
In general, you can just show the faces of all your soldiers that are here to the car. What is there, a high-quality photo, let it be 5 megabytes. Even if there are 100 soldiers in the conflict zone, only 000 megabytes. I saw someone's car with military equipment, punched through its base. There were no matches, which means the enemy - we wet him. And you can not even give a damn what form he has there.
And in order not to be able to rewrite this database, it is necessary to somehow restrict access to it. The most reliable way is to develop a unique connector for connecting to the car. And only through it it will be possible to change the embedded instructions. And there also all sorts of password checks will have to be passed.
But this is just an example.

[ViacheslavS]

In general, you can just show the faces of all your soldiers that are here to the car. What is there, a high-quality photo, let it be 5 megabytes. Even if there are 100 soldiers in the conflict zone, only 000 megabytes. I saw someone's car with military equipment, punched through its base. There were no matches, which means the enemy - we wet him. And you can not even give a damn what form he has there.

Not to mention possible recognition errors, because even 99% (and this is the result for combat conditions still needs to be obtained) it will not be a satisfactory result when this machine opens fire on its own in one percent of cases and on the turn considers the enemy to be its own (errors of the first and of the second kind), the second moment this robot becomes a carrier of incredibly secure information.

And in order not to be able to rewrite this database, it is necessary to somehow restrict access to it. The most reliable way is to develop a unique connector for connecting to the car. And only through it it will be possible to change the embedded instructions. And there also all sorts of password checks will have to be passed.

This does not protect against reverse in any way, any connectors will do if there is no third party in the cryptographic scheme, and even if the data is encrypted on a typewriter, but the key is also somewhere on the typewriter, they will get to it.

[Plate]

Not to mention possible recognition errors, because even 99% (and this result still needs to be obtained) will not be a satisfactory result when this machine opens fire on its own

Well, yes, but you won’t achieve 100% even with people, no matter how hard you try. I'll tell you another minus: I just thought about it. No one will walk on the battlefield with a face for show. Goggles, a healthy helmet, and even a gas mask. It turns out that the enemy fired gas, you put on a gas mask, and your enraged robot tank started firing at you.

the second moment this robot becomes a carrier of incredibly secret information

And what is secret in the faces of ordinary fighters? They calmly show on TV, the Ministry of Defense in their Internet groups and channels too.

This does not protect against reverse in any way, any connectors will do if there is no third party in the cryptographic schemes, and even if the data is encrypted on a typewriter, but the key is also somewhere on the typewriter, they will get to it.

It's right. They'll get to everything. Therefore, you can never develop protection and think that it is forever. Regular change of codes, passwords, keys is a classic.
Despite all the problems, the use of robots on the battlefield is where you need to go. That's where really - a steel fist.

[ViacheslavS]

Well, yes, but you won’t achieve 100% even with people, no matter how hard you try.

It’s easier with people, there is someone to compare with, and who will be to blame if the recognition system made a mistake of the second kind (there is a person in the database, but we didn’t recognize him false reject rate) and laid down his platoon.

I'll tell you another minus: I just thought about it. No one will walk on the battlefield with a face for show. Goggles, a healthy helmet, and even a gas mask. It turns out that the enemy fired gas, you put on a gas mask, and your enraged robot tank started firing at you.

So I wrote above how far all the developments of artificial intelligence are from the average static person, even in your example, the fighters recognize their own by silhouettes, by contours, by voice, by their original location.
I repeat, otherwise we would have been massively using unmanned vehicles for a long time.

And what is secret in the faces of ordinary fighters? They calmly show on TV, the Ministry of Defense in their Internet groups and channels too.

Show the faces of all the fighters in the conflict zone?

. Therefore, you can never develop protection and think that it is forever. Regular change of codes, passwords, keys is a classic.

This will not help with the reverse if your database and the key to it are somehow stored on the machine, and the key must be on the machine otherwise its software will not be able to read the data, but there are possible schemes that the key is encrypted with a password, and before switching to combat mode, enter this password and the key itself lies only in RAM, but this only adds complexity, but does not make it impossible to reverse.

[Plate]

It's easier with people, there is someone to compare

I'm sorry, what? I don’t want to dig into the spelling of the words, but without further context, I wouldn’t understand at all. And then, I'm not sure.
Do you need to ask someone? I propose to somehow experimentally determine who is less likely to put his platoon by mistake. If the car, then boldly implement. If a person has it, we refine it and measure it again. It is even more difficult with a person that you have to deal with him: is it a mistake or a crime. At the car, you can simply get into the logs or whatever it has.

I repeat, otherwise we would have been massively using unmanned vehicles for a long time.

Developments, developments, developments... Yes, release only machines on the battlefield and let everyone who is not connected to their control network be brought down. Like, the tank did not see itself from the side of the one it looked at - it failed. Leave a scorched desert.

Show the faces of all the fighters in the conflict zone?

When in Syria someone is interviewed from ours, it’s quite, if my memory serves me right. We don't participate anywhere else now. Purely logical. What is the value of knowing what a sergeant's or private's face looks like? It seems that he is not fighting crime in his city so that something could threaten the family because of this: the family is thousands of kilometers away.

but this only adds complexity, but does not make it impossible to reverse.

In principle, any security step only adds complexity to the cracker, but does not make his job impossible. BUT! Hacking a person's memory I would not call impossible either. So in terms of preserving information, I would no longer hope for people. There are experimental neural interfaces capable of displaying a picture from the brain on a computer. Of course, today it is filled with noise and the user has to work hard and train to do it. But in 30 years, you may have to, on the contrary, strain yourself so that some advertising bot does not count thoughts out of your head, after which you will be bombarded with advertisements for survivalist equipment, finding out that you are going to go camping one day. And what will happen during the interrogation ...

[ViacheslavS]

I propose to somehow experimentally determine who is less likely to put his platoon by mistake.

Now, even in ideal conditions without glasses and masks, it’s good if you manage to achieve a recognition accuracy of 99% in combat conditions, that is, the machine will be wrong 1 time in a hundred.

When in Syria someone is interviewed from ours, it’s quite, if my memory serves me right. We don't participate anywhere else now. Purely logical. What is the value of knowing what a sergeant's or private's face looks like? It seems that he is not fighting crime in his city so that something could threaten the family because of this: the family is thousands of kilometers away.

Obtaining a database of all personnel with photographs and apparently positions, ranks (whatever one may say, but meta-information is required) can be used by the enemy for “social attacks” by relatives of the military, somewhere it can help uncover possible operations, etc., and in many countries the military even banned from using social networks. networks by which they could be identified.

[Plate]

Well, okay. It is necessary to release only cars on the battlefield and say: "Bring down everyone." Then no more mistakes!

[Klingon]

Quote: nikvic46

Good technology. And great specialists created it. It is one matter of learning. The combat situation is another matter. Having expended energy, where will this technique be refueled? The dugout in the photo is not the work of AI.

That's why the AI ​​is there in order to determine its own parameters in PM, to find in time where to get fuel (return to the base or find a refueling robot) while the fuel has not yet been completely consumed

[dauria]

That's why the AI ​​is there to determine its own parameters in PM, to find in time where to get fuel

Where to run away from the owner's sneaker, who turned on the light in the kitchen. With whom to copulate, where to lay eggs. Which apartment to move to, if dichlorvos is sprinkled here. And the main thing is not to ask people for anything.
At least make a cockroach with his reflexes ... AI ...
In humans, intelligence begins with the word "mother". When he extracts from a heap of signals from the picture in the eyes to sounds and smells and assigns to the object a short symbol of four bytes - mother. That's when the robot will be able to write a presentation on the painting "Deuce Again", then we will talk about intelligence.

[The leader of the Redskins]

I agree with you.
The phrase "Artificial Intelligence" implies making any independent decisions. The author has listed the developed / developed complexes, but somehow poorly illuminated their capabilities in this aspect.
There, two cars on tracks, on the road. Will they even reach the specified point independently on command?

[NDR-791]

Quote: Leader of the Redskins

The phrase "Artificial Intelligence" implies making any independent decisions. The author has listed the developed / developed complexes, but somehow poorly illuminated their capabilities in this aspect.

That's exactly what. But this is not AI, it is just an algorithm. Yes, the algorithm is complex, but not creative. That is, it does not imply "soldier's ingenuity." And again, articles in the style of "having no analogue in the world" do not lead to any good because they strongly distort reality in favor of ordinary dreams. And this is fraught.
And in fact, there is not even a single definition of AI.

[Klingon]

No need to clown about, by the way, the "dumb" vacuum cleaner from Samsung is still able to find the charging "base" and recharge itself. The same was done by the dog Aibo from Sonya

[Cat]

a vacuum cleaner from Samsung is able to find the charging "base" and recharge itself.

An amazing achievement! And my electric kettle turns off by itself when the water boils in it. I generally keep quiet about the coffee machine.

[vadimtt]

And he crawls to the tap with water, your kettle? And if famous personalities all drank from the tap, what will they do? And so on. In fact, the concepts are confused here. AI is too general, at least there is "strong" and just AI. All the Wishlist is realized by the "strong" one, but at the moment there is only the usual "Deep learning networks" are the same ordinary AI (probably the final stage in the development of Rosenblatt's ideas), although far removed from the classic "if-then" constructions. Further, only recurrent spike networks, but how to build them correctly is not yet clear, but poking around in living brains has already gone far and general structural ideas are visible. If suddenly where a strong AI is born, then everything is possible further, up to the end of human civilization We need it?

[SavranP]

Yes, you are right, the "AI" described in the article has no desires, no feelings, no instincts. No incentives, no will to do anything, "Oblomov" in kind? But when the "AI" will have incentives, if there are, then the person may have to run from the "sneaker"! Hollywood, in fact, has already shown everything.

[stankow]

Just from the numerous tasks for AI, he has pattern recognition, everything is fine. And mom, and dad, and 3D environment. But to the presentation - far away.

[Plate]

That's when the robot can write a summary of the painting "Again deuce", then we'll talk about intelligence.

A robot can write an essay. A good one can write an interesting essay today. But on the subject given by the person, yes.
A robot today can sing, draw a picture, competently organize patrols of the area ... but it can do all this only at the commands of a person. Without a team, even the smartest robot will stand still.

[Ross xnumx]

To the author - honored (+).

Artificial intelligence in the Russian army

The variety of ideas and developments is amazing. Doubts creep in that all this will find (someday) use ...
A drone looks nice, capable of monitoring and (even !!!), at the command of the built-in AI, to make a decision to destroy an object. And what does this change in patrolling (for what purpose) by the Russian military on the Syrian territory?
Where is this army of "insensitive" and "indefatigable" guards, observers, fighters, terminators? Why are living terrorists still gathering (with impunity) in groups on the territory open for observation and making attempts to harm both the Aerospace Forces forces and the civilian population of the state where the peacekeeping mission is being conducted?
==========
Every time I read such materials, I begin to think about the place of my own residence and the place of habitation of those who put some plans on some kind of AI in the Russian army. Brainy guys concluded that AI will not be subject to passions and desires: only cold calculation and rationality of each subsequent action ...
WHERE IS IT SHE SHOWING HIMSELF IN PRACTICE ???

[SavranP]

Perhaps you do not quite understand the level of modern "AI"? The AI ​​described in the article has no feelings at all, no! There is only a program algorithm set by a developer (systems engineer), and a program created by a programmer, and then written into the AI's memory. The AI ​​mechanism has an extensive and very flexible program that adapts to the circumstances, but it is a program created by a person, which does not leave the "AI" freedom of action.

[Ross xnumx]

Quote: SavranP

The AI ​​described in the article has no feelings at all, none! There is only a program algorithm set by the developer

And who is against it?

Quote: ROSS 42

AI will not be subject to passions and desires: only cold calculation and rationality of each subsequent action ...

[SavranP]

Usually "cold calculation", "rationality" mean feelings? Like a human. No? Then excuse me, I just misunderstood.

[SavranP]

If you doubt the capabilities of the Russians in providing the Russian army, then it is completely in vain. Do you seem to be based on your past experiences? Purely to understand the huge difference with the technology created (20-30) years ago, we can say that the emphasis is on "systems on a chip", that there are several design centers in Russia that are quite authoritative abroad, and so on.

[Vadim237]

"The drone looks nice, capable of monitoring and (even !!!), at the command of the built-in AI, to make a decision to destroy this or that object." You seem to have slept through that moment this summer when a Turkish kamikaze drone attacked the Kurds in autonomous mode.

[Ross xnumx]

Quote: Vadim237

when Turkish kamikaze drone is offline

Thank you for reminding me. Key words here:
Turkish kamikaze drone

[Vadim237]

With an automatic decision-making system to attack or not.

[gregor6549]

I wrote a lengthy comment about AI, but suddenly the network went out and everything written prematurely died. There is no desire to start all over again, so I will just say that AI in military systems is still the same myth, like many other unparalleled ones.
Yes, work on the creation of AI elements is carried out in many countries, but this takes into account the unwritten law that decision-making in military systems should remain with the person. Otherwise, the world is threatened by a real Skynet with no less real Terminators.
In civilian systems, for example, in a rover, the presence of AI elements and the transfer of decisions to control the rover to them is not only permissible, but also necessary, since operational control of the rover from the Earth in real time is impossible due to an unacceptable delay in the transmission of control signals. And apparently there are such AI elements there and they work quite successfully.

[A. Privalov]

You should not be afraid of AI, but natural stupidity and stupidity ..

[dauria]

You should not be afraid of AI, but natural stupidity and stupidity ..

Philosophically It's just not fair.
"He had a happy talent
Without coercion in conversation
Touch everything lightly, ... "
Stupidity and stupidity are inevitable companions of complex algorithms. In fact, these are unforeseen situations that were not taken into account in the process of natural or artificial selection (creation of an algorithm) And it does not matter whether the devices were created by nature or by man.
Once I saw a swift crash on a wide window glass simply because the garden was reflected there. A stone falling from a mountain will never be so wrong.

[A. Privalov]

Quote: dauria

You should not be afraid of AI, but natural stupidity and stupidity ..

Philosophically It's just not fair.
"He had a happy talent
Without coercion in conversation
Touch everything lightly, ... "
Stupidity and stupidity are inevitable companions of complex algorithms. In fact, these are unforeseen situations that were not taken into account in the process of natural or artificial selection (creation of an algorithm) And it does not matter whether the devices were created by nature or by man.
Once I saw a swift crash on a wide window glass simply because the garden was reflected there. A stone falling from a mountain will never be so wrong.

Come on, Alexey. There is no need to dig that deep. You should not create unnecessarily new entities (s), if what is happening can be explained by ordinary human stupidity. This genius has limits, but stupidity and stupidity do not.

[dauria]

if what is happening can be explained by ordinary human stupidity

Exactly - the usual! Have you ever done anything stupid? Similar
nonsense will be done by an artificial brain. It seems that it was about this. And we must also be afraid of his stupidity, since we will allow him to self-study, develop conditioned reflexes and even "think." Unlike those laid down in him by a person, this will already be his personal stupidity and stupidity.

[A. Privalov]

Quote: dauria

if what is happening can be explained by ordinary human stupidity

Exactly - the usual! Have you ever done anything stupid? Similar
nonsense will be done by an artificial brain. It seems that it was about this. And we must also be afraid of his stupidity, since we will allow him to self-study, develop conditioned reflexes and even "think." Unlike those laid down in him by a person, this will already be his personal stupidity and stupidity.

Of course, you and I have done stupid things and we will do more. AI will do too. Only he is a piece of iron, and we are human beings. Our demand is different. Wait, the day will come when it will not be ethical to call AI a piece of iron, and photographs in negligee against their background will be taken for an insult with all that it implies.

[dauria]

Wait, the day will come when it will not be ethical to call AI a piece of iron.

Worse ... the day will come when very rich people, very willing to live forever, will begin to "move" before the end to an artificial carrier of reason. Backing up once a week ...

[A. Privalov]

Quote: dauria

the day will come when very rich people, very willing to live forever, will begin to "move" before the end to an artificial carrier of reason. Backing up once a week ...

I don't see anything bad. Let them create. At first, it will be expensive and unaffordable, and gradually, many will be able to afford it. Today, after all, not everyone has cars, but you must admit that they are more affordable than they were 50 years ago.

[dauria]

I don't see anything bad. Let them create.

Yes, I don't seem to see it either yet ... Moreover, it cannot be stopped. Ocean, space - amazing horizons will open. Or maybe just an iron drinking companion. "For life" to drink, talk. Would he only want to, bastard? Say- "Well, you ram, my friend. You don't know the Laplace transform from the Heaviside function."
Isn't it scary to become unnecessary?

[A. Privalov]

Quote: dauria

Isn't it scary to become unnecessary?

Enough for my lifetime.

[dauria]

Enough for my lifetime.

Yes, there was some serious sensible American. He predicted by 2030 a human brain costing one and a half thousand bucks and eating only 100 watts. It’s a pity, I don’t remember where I read it.

[A. Privalov]

Quote: dauria

Enough for my lifetime.

Yes, there was some serious sensible American. He predicted by 2030 a human brain costing one and a half thousand bucks and eating only 100 watts. It’s a pity, I don’t remember where I read it.

Why do we need to fantasize for many years to come. Here are some people who confidently claim that AI has already been used in a serious mess:

 Show / Hide text

https://censor.net/ru/resonance/3289134/samoe_tehnologichnoe_ubiyistvo_xxi_veka

[Motorist]

Quote: A. Privalov

https://censor.net/ru/resonance/3289134/samoe_tehnologichnoe_ubiyistvo_xxi_veka

Isn't this the same Butusov who was wounded on the Kolchak fronts? Some comments, of course, fire ... But these are marked as spam:

This is a common terrorist act, the Tsim by driving Israel into a terrorist state ...

[Igor Tikhomirov]

It would be nice to add natural intelligence to our realities for a start. Then you can talk about artificial.

[Bez 310]

It is high time to introduce AI everywhere, since, judging by information from open sources, the current military personnel are not all right with natural intelligence.

[Victor Tsenin]

> will maintain its leading position in the world.

It is gratifying that things are going on, but there can be no talk of maintaining positions until we start real import substitution, because the recent news about the stuffing of our UAVs is simply horrible!

[APASUS]

In general, the appearance of such a technique leads to a complete change in the tactics and technology of combat. Interestingly, what has changed in our documentation, regulations and other documents for the same cadets?

[voyaka uh]

Strictly speaking, AI and autonomous systems are not the same thing.
AI is a self-learning autonomous system. She writes (adds) the program code herself in the process of performing tasks.
There are no such combat systems in the troops yet.

And ordinary autonomous systems - with complex software, but written by people.
Such combat systems are already in operation.

[gridasov]

It's a sin to call programming Artificial Intelligence. you can set a systemic benchmark as a program in the analysis of one or another array of information. But analysis is not programming - it is deep forecasting on algorithms built in analysis as a method for processing big data now. In addition, analysis is always an opportunity to enter new data and recommendation functions regarding the assigned tasks. Therefore, there is now a lot of information and the technologies for its collection are relatively effective, but data analysis is still a problem because analysis cannot be carried out by computational methods of mathematics. Since the variable processes change and grow in dynamics.

[ALARI]

The title of the article is like the beginning of the anecdote))

[Catfish]

Like, "Jew is a janitor"?

[Vadim237]

The designers of our robotic military equipment have a question - why the hell are they making such narrow caterpillars to make it easier for the equipment to sit on its belly or the cross-country ability to pass sharply decreases in spring and autumn.

[smaug78]

Uranus-9 is a full-fledged combat robot with artificial intelligence

The RF Ministry of Defense was very surprised ...

[Max PV]

Now, somehow, in what is called AI "horses, people are mixed in a heap." What they do not push under this topic: from analog circuits more or less, to image recognition algorithms, and from primitive programs for a dozen lines of code to expert systems and neural networks. But there is still no full-fledged AI. And it is not needed in the overwhelming majority of cases, requiring huge resources to implement the simplest tasks. What kind of AI can we talk about if you have a finite number of scenarios and methods of action? An expert system is required at most. In fact, the only military applications of AI can be considered the analysis of arrays of complex signals, such as video, sound, and the like, and it is necessary to solve the problem from the fairy tale “go there, I don’t know where, bring that, I don’t know what”, that is, the parameters of the desired are not determined in the assignment. For example: the task of finding a camouflaged person on the ground, when neither the camouflage method, nor what can be seen specifically, nor whether he is there at all, is unknown.

[Daemon]

This year, at Army-2021, there was a round table dedicated to the mathematical support of AI. Gathered very eminent professors, doctors of science, representatives of industry. The first 15 minutes of the round table was devoted to how ordinary mortals do not understand the essence and call everything AI in a row and got crazy over it. And these are people in whom AI was everywhere in positions, on the business card and in the topics of the report. This is a cool marketing ploy that adds the word "AI" to any bullshit and raises both the price tag and the importance of research. They wrote everything above correctly - there is no AI and there won't be any in the near future. There are highly specialized algorithms that automate some functions and processes, but they were written by the person himself.

[Shahno]

// There are highly specialized algorithms that automate some functions and processes, but they were written by the person himself //
In AI, a person does not write algorithms (functions and processes), even in the primitive case of DL, let alone RL.
A person uses code to optimize the environment - for example, a universal approximator ...
How it's done, well ... complete an MS in Data Science, then you'll be in the know.

[Daemon]

I’m sorry, of course, but you had bad courses in AI, ML, DL and Data Science if you were not explained how all this works at the level of software implementation. Starting from the program interface, data loading algorithms (picture, data array), data storage, learning mode, processing and recognition - all this is being implemented (writing the algorithm and code) by a human with pens in programming languages. And it is he who introduces certain restrictions on the operation of the program and determines the final area of ​​application of a particular program. Read at least about OpenCV, Google TensorFlow, Keras, etc. to understand what I mean and how it is used.

[Shahno]

And what does a Keras programmer program? Structure, parameters, no? And the courses? But yes, I only teach sometimes, when I am invited, if the payment suits me, and I have time ..
By the way, google cloud dataflow, not tensorflow ... OpenCV didn't understand what AI has to do with it.
OpenCV for ComputerVision, feel the difference? Vision is still an interface for an object with AI.

[vadimtt]

On OpenCV forks, all AI software in MO and the surrounding area
But Keras is questionable, because the tensor-flow code is partially closed. And in general, a python in a military service looks somehow strange Is it possible to put it beautifully on the architecture of Elbrus in general with gil?

[Shahno]

Yes, I didn’t know about MO. In general, it’s a matter of taste. But OpenCV for AI ... a bunch of special stages in C. TPUs have already been added by the way.
For my pipelines I use opencv mainly as localization ...
As it is clear, from the point of view of openness and finance, the platform is tolerable.
Python, why not, depending on how much they pay, I don't really know Elbrus, but if you tinker ..

[Shahno]

As for tf, and Keras add-ons ... it's still a pretty tough tool for implementing linear algebra paradigms, written in C.
But this is the basis, no more, on which optimization (aka cost function) is implemented, using stochastic mechanisms, for example, or competitive learning, for another example.
Due to the probabilistic scheme, the AI ​​algorithm is, as our partners from Boston say, half art, half not art.
Ps. I haven't taught in Russian for a long time ... Thank you for the opportunity.

[vadimtt]

Unfortunately, the breakthrough in neural networks is over, deeplearn has certainly gone far from perceptrons, but not enough for strong ai. Further breakthroughs are likely to be associated with recurrent spike net or direct modeling of columnar-type cortical structures. But there are many problems with sustainability and there is no understanding of how to solve them. Although nature could

[Shahno]

Yes, something similar was used in one project of the pulse coupled nn type for the image fusion task.

[p0pulivox]

Quote: NDR-791

Quote: Leader of the Redskins

The phrase "Artificial Intelligence" implies making any independent decisions. The author has listed the developed / developed complexes, but somehow poorly illuminated their capabilities in this aspect.

That's exactly what. But this is not AI, it is just an algorithm. Yes, the algorithm is complex, but not creative. That is, it does not imply "soldier's ingenuity." And again, articles in the style of "having no analogue in the world" do not lead to any good because they strongly distort reality in favor of ordinary dreams. And this is fraught.
And in fact, there is not even a single definition of AI.

Outstripped! Now the concept of AI is often used for speculative purposes or illiterate, which we observe ...

[tank64rus]

So far, this is an unmanned combat robot with AI elements and nothing more. It all depends on the operator. During the exercises, they sat in a building. Questions for training operators to control robots during a combat situation. Now there is still no experience and Syria immediately showed all the flaws. We need an advanced control center, we need to determine its place in the combat system, the options for the operators of the interaction of robots with each other and the means of support, etc. etc. At least this, but AI will come in the future in the course of the work.

[Mister who]

AI is easy to create in 10-15 years, fast-track and 5-7 years of funding.

[vadimtt]

This song about easy to create in 10-15 years has been sung since the 50s of the last century.
As well as about limitless thermonuclear energy

[Vadim237]

Now electronics allows you to do this.

[vadimtt]

If it allows, why didn't they create it?
And the main problem is far from electronics.

[Oleg133]

Everything you need to know about AI.

In America, they brought to justice for the accident of a dude lying in the back seat of a car with autopilot

[ivan2022]

We should not forget that artificial intelligence is created by natural brains. If a society is constantly experiencing problems even with the assimilation of ideas: Christian, communist, liberal .... it will definitely have huge problems in creating computers and other new and complex technology. We have many talents of scientists, but the word "authority" among the people means not a scientist, but a bandit ... .. Not the smartest of us "make the weather"