Military Review

Artificial Intelligence: Reality or Future?

25
Artificial Intelligence: Reality or Future?For many millennia, man has tried to determine how he thinks, what processes are taking place in his head. So in the field of artificial intelligence (AI), scientists have to solve an even more difficult task. Indeed, in this area, specialists will not only understand the essence of the intellect, but also create intellectual entities.


First of all, it should be noted that artificial intelligence is a rather young science. The first experiments in this field appeared shortly after the end of the Second World War, and the term “artificial intelligence” appeared a little later - in 1956 year. At the same time, if in other areas of science it is rather difficult to make a great discovery, then this area of ​​science opens up great prospects for the manifestation of talent.

At present, the problems of artificial intelligence include a large list of various scientific fields, including such general concepts as perception and training, as well as special tasks, in particular, proof of theorems, playing chess, and diagnosing diseases.

In this area, the analysis and systematization of intellectual tasks is carried out, thus, artificial intelligence concerns all spheres of human intellectual activity, and therefore it can be considered a universal field of science.

From the foregoing, it can be concluded that the field of scientific intelligence is a very interesting field of science. It is noteworthy that a single definition of AI does not exist. In various scientific works devoted to him, there are various interpretations of this phenomenon. They can cover not only thought processes, but also formulations regarding the behavior of an individual.

If you carefully study history development of artificial intelligence, it can be seen that research was conducted in several directions. Hence the conclusion that between those scientists who were engaged in research of human abilities and those who were engaged in problems of rationality, there were certain controversial situations.

The scientific approach, which focuses on the study of man, should be based on the formulation of a large number of hypotheses, as well as their experimental evidence. At the same time, an approach focused on the study of the concept of rationality is a kind of combination of technology and mathematics.

In order to test whether a computer is capable of performing actions like a human being, an approach was created that basically relied on the Turing test. He got his name from the creator, Alan Turing. The test is used as a satisfactory functional definition of intelligence. An English mathematician who laid the foundations of computer technology, in 1950, published a scientific article called Computer Computers and Mind, in which a test was proposed that could be used to determine the intellectual level and nature of computer intelligence.

The author of the test came to the conclusion that there is no point in developing a large list of requirements in order to create artificial intelligence, which to everything else can be very controversial, and therefore proposed a test that was based on the fact that in the end it would be impossible to distinguish the behavior of an object endowed with artificial intelligence, from the behavior of human beings. Thus, the computer will be able to successfully pass the test if the experimenter who asked him questions in writing is unable to determine from whom the answers were actually received - from a person or from a specific device.

At the same time, the author derived a formula that defined the boundary when artificial intelligence could reach the level of natural. According to Turing's findings, if a computer can deceive a person when answering 30 percent of questions, then we can assume that he has artificial intelligence.

However, in order for the computer to be able to answer the questions posed, it must do a lot of work. So, in particular, it should have such capabilities as natural language information processing tools that would allow to communicate quite successfully with the device in one of the languages ​​existing in the world. In addition, it should be equipped with knowledge presentation tools, with which the device will be able to record new information in the memory. There should also be tools for automatically generating conclusions, which would provide an opportunity to use the available information to search for answers to the questions posed and to formulate new conclusions. Machine learning tools are designed to provide the computer with the ability to adapt to new circumstances, and in addition, to detect signs of a standard situation.

The Turing test consciously excludes the possibility of direct physical interaction between the person conducting the experiment and the computer, because the process of creating artificial intelligence does not require a physical imitation of the person. In this case, in the case of using the full version of the test, the experimenter can use the video signal to verify the computer's ability to perceive.

Therefore, when passing the full Turing test to the above means, it is necessary to have machine vision for the perception of the object, as well as the means of robotics for the possibility of manipulating and moving objects.

All this is ultimately the basis of artificial intelligence, and the Turing test has not lost its importance even after half a century. However, it should be noted that scientists who study and create artificial intelligence almost never solve problems aimed at passing this test, believing that it is much more important to study in detail the principles that underlie intelligence than to create a copy of one from carriers of natural intelligence.

However, the Turing test was recognized as a benchmark, but until recently, scientists could not create a program that would successfully overcome the test. Thus, scientists could without any problems determine with whom they are talking, with a computer or with a person.

However, a few months ago, the media reported that scientists for the first time in fifty years managed to get very close to creating artificial intelligence, which was capable of thinking like a human being. As it turned out, the authors of the program were a Russian group of scientists.

At the end of June in the UK was held the world scientific competition of cybernetic intelligence, which was held under the auspices of the University of Reading. The competition was held in the main cipher center of Blatchley Park. Russian scientists presented a program called "Eugene". Besides her, 4 programs also took part in the testing. The Russian development was recognized as the winner, answering the percentage of questions posed to 29,2 in the same way as a person. Thus, the program lacked the entire 0,8 percent in order for the expected event to occur so long ago - the emergence of artificial intelligence.

American scientists also keep up with the Russians. So, they managed to create software bots that were designed specifically for computer games. They had no problems and quite confidently overcome the modified Turing testing. It should be noted that this was done with much greater success than people who were tested with bots. And from this it is possible to draw certain conclusions that artificial intelligence has managed to reach the level where the automatic system is no longer able to determine where a person answers, and where the computer answers.

Of course, it is too early to say that it is too early to overcome such a specific version of the Turing test, which is a game schutter, an indicator of the creation of artificial intelligence by a person. At the same time, it gives the full right to say that artificial intelligence is gradually approaching human, and also that game bots have already reached the level of development at which automatic systems designed to determine human behavior can be quite successfully deceived.

The creators of game bots were scientists from the University of Texas Jacob Schrum, Risto Miikkulainen and Igor Karpov. They managed to create artificial intelligence that can play the game on a human level. A huge virtual platform was created on which many bots and real people fought. Most played anonymously. More than half of the game bots were identified by judges as people. At the same time, they considered some people to be bots. Thus, the conclusion suggests itself that computer characters already behave like people in games.

The experiment was conducted as part of a competition called BotPrize, which started in America back in 2008 year. Scientists and developers whose computer programs will be able to deceive people can become its participants. Posing as quite real players. But the first successes in this field were achieved only in 2010 year.

The winners will receive an award in the amount of 4,5 thousand pounds, and will continue to work on their programs. And there is still something to strive for, because in order to recognize the creation of artificial intelligence, the program must convince everyone that he is a person in the course of a conversation. And this requires deep knowledge about the work of the human brain and the principles of the formation of speech. Currently, no one has yet managed to pass the Turing test in its original form. But it is quite possible to assume that this may happen in the near future ...

Materials used:
http://expert.ru/2012/09/28/iskusstvennyij-intellekt-na-podhode/?n=66992
http://www.dailytechinfo.org/infotech/4043-iskusstvennyy-intellekt-v-lice-igrovogo-bota-obygral-lyudey-i-proshel-igrovoy-test-tyuringa.html
http://cybernetic.me/rossijskie-uchenye-vplotnuyu-priblizilis-k-sozdaniyu-iskusstvennogo-intellekta/
http://ru.wikipedia.org/wiki/%D2%E5%F1%F2_%D2%FC%FE%F0%E8%ED%E3%E0
http://vladimir.socio.msu.ru/1_KM/theme_062.htm
Author:
25 comments
Information
Dear reader, to leave comments on the publication, you must to register.

I have an account? Sign in

  1. aksakal
    aksakal 15 October 2012 09: 15
    +2
    I am glad that the Russians in this direction have not lagged behind. And one more thing: Russians have a backlog in microelectronics. Is it really necessary to overcome it? Chips already produced by the Russian industry may well be carriers of artificial intelligence, and artificial intelligence algorithms can easily level and nullify all the advantages that amers now have in the field of processing information based on old architecture (the so-called von Neumann architecture) .
    My assumptions do not apply to those microelectronic devices, the functions of which are to receive information and its primary processing, in other words, all these different kinds of sensors of a different nature. Here the Russians will not be able to "get around at the turn", they will have to develop and catch up with all this, and on their own - such things are not sold, more precisely, they are sold ready-made without the right to open. And maybe even with bookmarks. IMHO
    1. Green
      Green 15 October 2012 12: 17
      +3
      I know what I'm talking about. Russia is in the ass, with regard to real nanoelectronics. (Please do not confuse with the crap that Medved implies) The chip factories that are being built in Russia now are either electronic consumer goods or dumping technologies 20 years ago. And the limit of 1000 nanometers can not be overcome with one throw caps. (Critical for microelements) This business cannot be overpowered by leaps and bounds, it’s just that you can’t save technology. Overcoming this limit, it is necessary to compensate for the natural trembling of the earth and much more. And in this we are behind. Behind hopelessly and FOREVER. They lagged long before Chubais and Medved. Left behind under Leonid Ilyich. I am not Russophobe, and I love my country, but reality is cruel. Hurray patriots with 2 classes of education can be free. The nano-president with a nano-brain will tell them about nano-achievements.
      1. aksakal
        aksakal 15 October 2012 13: 03
        0
        Quote: Verde
        I know what I'm talking about. Russia is in the ass, with regard to real nanoelectronics. (Please do not confuse with the crap that Medved implies) The chip factories that are being built in Russia now are either electronic consumer goods or dumping technologies 20 years ago.

        - and I’m just about - and whether it is necessary to catch up? For example, if you are behind in carriage production, the western ones are prettier and more comfortable, and the stroke is softer and different, but the first and most primitive cars with a carburetor in the form of a living granny have already appeared -)))) (just read who you used as a carburetor great Daimler-)))), isn’t it more rational to start intensively developing a car than to catch up with the west in carriage production? Let the west continue to perfect its carriages, and we will use their carriages (even for hard currency), in the meantime we will go ahead in the car -)))). To do this, you need to at least discern the potential of the car and take the right decision. Perhaps the analogy is unsuccessful, but still ... Stop clinging to the carriages and all the more so to get discouraged due to lag in this area.
        Usually opponents of AI are IT people of all stripes who are firmly convinced that the current architecture in the field of information processing is the most correct, the rest are a priori wrong. How can bread be taken away -))))
        1. Green
          Green 15 October 2012 13: 19
          +2
          The next step is quantum computer technology. and here the Americans are not frail development. You cannot make up for some algorithms and you won’t make an alternative. These are dreams. In the development of quantum computers, Americans are even more attentive. They buy up whole collectives all over the world, bitches of children and do it right. Forget it. Well right. not serious it. With such degenerates in power. Well right. Algorithms, algorithms ... What architecture? What are the algorithms? what will you run them on? These are FUCKED ALTERNATIVE algorithms. On the accounts? Enough already, about this special way, or we will go the other way. We will go our own way. And this way in the ass. Went many times already in this special way. We need massive TITANIAN investments in the fundamental, we need billions of dollars of infusion and the mobilization of the potential that is still warming. But not a special way. Enough, were already. you are apparently just far from the topic.
          1. aksakal
            aksakal 15 October 2012 14: 05
            0
            Quote: Verde
            The next step is quantum computer technology. and here the Americans are not frail development

            - according to quantum calculators, everything and amers on a theoretical level. Russia in theory does not lag behind and is completely up to date. That's when it goes into the technological plane, then it’s dangerous.
            Quote: Verde
            Well right. Algorithms, algorithms ... What architecture? What are the algorithms? what will you run them on? THESE FOOT ALTERNATIVE ALGORITHMS

            - I will repeat the post below: "Why, losing to this Deep Blue millions of times in production, nevertheless managed to do almost the same thing with almost the same result? The difference is in the algorithms. Deep Blue is simply stupid computing power, just stupidly sorting through all possible scenarios on a chessboard, of course, there are algorithms that discard worthless options right away, otherwise it is simply impossible to consider everything. And kaissa, due to the limited power of it, cannot do this closely. The developers, among whom Smyslov took part, immediately went along the path of imitating human thinking - the use of memory when considering positions, the inherent set of standard positions, of which there are a great many types of Sicilian defense in chess, and the use of associative algorithms in order to see the similarities between the existing the situation on the chessboard"- don't you catch it? That is, due to algorithms, it is possible to reduce the required power by a factor of millions! Is that not enough for you? Here is Deep Blue, not having the memory of the past (well, like past games, I mean a reworked memory that can be used in current actions, well, type of experience), every match is forced to play as a beginner, traveling at a monstrous speed.Opponent made a move - deep blue launches its power and calculates all the options several moves ahead, and there are a great many such options. And you need to calculate it in a limited time - the chess clock is ticking.
            And if you just apply the algorithms "taken" from the brow (more precisely, peeped from nature)? The person looks at the board and immediately deja vu - the current situation on the board reminds him of a game he once played or learned. Well, with differences in details. Can you imagine how many unnecessary moves are already possible, more precisely, unnecessary elaboration of unnecessary moves can be discarded! With the help of some algorithms it is possible to achieve a better effect at capacities that are an order of magnitude lower (several orders of magnitude!) Than in DeepBlue.
            Verde, how to convey to you even more clearly?

            Quote: Verde
            We need massive TITANIC investments in the fundamental
            - the realization and cognition of how people think is fundamentalism. This is one of the subsections of nonlinear dynamics. Well, at least a little respect on this subject. But the improvement of chips is not at all fundamental, it is technology, at best, applied science.
            1. snek
              snek 15 October 2012 14: 30
              0
              Quote: Verde
              You cannot make up for some algorithms and you won’t make an alternative. These are dreams.

              I absolutely agree - trying to beat hardware with software is futile. Moreover, in the same states they are developing both.
              Here is an example in the field of natural language recognition (human speech versus "machine languages") using the example of how the Watson computer plays English. TV show version Your game:

              here is more about him
              http://en.wikipedia.org/wiki/Watson_computer
              and there you can already follow the links
              1. aksakal
                aksakal 15 October 2012 14: 51
                0
                Quote: snek
                I absolutely agree - trying to beat hardware with software is futile. Moreover, in the same states they develop both. Here is an example in the field of natural language recognition (speech of people against "machine languages") using the example of how the Watson computer plays English. version of the TV show Own game

                - I'll look at your link at home. Now I have to repeat: "In the field of information processing, there are subregions where Russians will not be able to get away from stupid computing power - well, such as a computer simulation of a nuclear explosion or cases related to weather forecasting. There are no solutions, just computing power is needed. Accordingly, all the same, Russians will have to be able to create supercomps, too, and everything related to this is chips and other. Well, there is a sub-area related to decision-making. By the way, recognition (in military terminology, we have in front of us an enemy tank or some kind of tractor ?) - This is a subdomain of decision theory, such as what class the observable belongs to? IMHO, this particular area is of greater importance in military affairs, and therefore, the emphasis here is on the development of a new type of algorithms - they will give a greater effect than the Amer way, a way to bluntly increase computing power - that is, the amers cannot get away from the development of AI algorithms, and the Russians cannot get away from improving the "hardware". But something else pleases - in theoretical studies, in understanding how people think, the Russians are still ahead. Amers are all toys.
                For an illustrative example - entry into memory. Pure computer memory is completely unsuitable for use as an experience. Not only that, everything is turned upside down there - are you proving to me that this is correct? For example, just a white wall and the same wall with an inscription on the wall - in the human dimension, in terms of the information provided, there are two very big differences. Just a white wall for a chela hardly "weighs" at least a byte. In the computerized dimension, both the white wall and the wall with the inscription will weigh almost the same, depending on the number of pixels. You persistently prove to me that it is necessary to stay on this "inverted", purely computer dimension, imposed by amers (not out of malice, they just went along this path and led us all). This just suggests that we will have to improve the hardware, where we are already well behind.
                Another example is how the computer remembers the human face and how people remember. Comp remembers stupidly - the whole face is in pixels, it requires a lot of power, every new face to remember so - torment.
                A person (presumably, no one in the world will say for sure yet) - first forms in his memory the etalon of the face, as it were, such a super-average person where the parts of the face should be located there, approximately the shape of these parts and their relative position. Significant deviations from the reference are perceived as ugliness. And now all the new faces there is no need to remember completely - it is enough to write in the memory only the differences of the face of a new acquaintance from the reference, that's all. Memory capacities for this need much less than for computer storage, in addition, the thus-processed info is suitable for associative use, which allows you to use past experience in actions in the present.
                Still persist in iron?
                1. snek
                  snek 15 October 2012 15: 12
                  -1
                  aksakal, in your comments you contrast hardware and software, saying that while the Americans are developing processors, we will make wonderful algorithms. So the Americans are smashing both hardware and software. UAVs, cars that can drive without a driver are all applications of the so-called weak or applied artificial intelligence. They already have all this in practice.
                  Quote: aksakal
                  Another example is how the computer remembers the human face and how people remember. Comp remembers stupidly - the whole face is in pixels, it requires a lot of power, every new face to remember so - torment.

                  This and other similar remarks of yours suggest that you are familiar with the development of those 10 or more years ago.
                  here is an example of facial recognition software
                  http://electronics.howstuffworks.com/gadgets/high-tech-gadgets/facial-recognitio

                  n1.htm
                  here are some relevant wikipedia content
                  http://en.wikipedia.org/wiki/Eigenface
                  http://en.wikipedia.org/wiki/Facial_recognition_system
                  Unfortunately, everything is in English (again, this is not a good symptom).
                  1. aksakal
                    aksakal 15 October 2012 15: 54
                    0
                    Interestingly, a little later I will try to get acquainted with your links.
                    Are they still behind here? - ((((.
                    In terms of driving without a driver, it’s just an example that amers are trying to get out on computing power, they put computers with such speed that they themselves later complain that it’s cheaper to put Schumacher himself in this sense in the literal sense! Well, again, why so much power? - algorithms again. To solve a bunch of diffurs during the movement - if this is intelligence, then it is really a very WEAK intelligence -)))))). People do not solve any diffurs in motion, do not overestimate the human brain, it just goes -))))). And while no one knows how.
                    Presumably, the brow is actively involved in the so-called forgot what logics, the author is an American of Azerbaijani origin Zade (for some reason, Azerbaijanis are terribly proud of him, but he is an Amer, he studied in America and his becoming a scientist was in Amer. It’s the same as Isosif Prigozhin to call a Russian scientist - in fact, this laureate Nobel - a Belgian scientist of Russian origin). Creating a 3-D model in our heads so that we can even imagine ourselves from the side and even from above, and it’s relatively true + these are the logics when estimating the distance with any maneuvering = successful driving without any solutions to the most complicated diffurs. A successful solution to this issue - and you do not need a tricked out powerful iron. That's AI. And the fact that you call AI in the Amer’s sense is so-so AI.
                  2. aksakal
                    aksakal 15 October 2012 17: 36
                    +1
                    Quote: snek
                    aksakal, in your comments you contrast hardware and software
                    - misunderstood me and in general what I want to convey. What we are talking about in this article is a completely different level of information processing, and a fundamentally different one! There is even more difference than between "For example, if you are behind in carriage production, the western ones are prettier and more comfortable, and the stroke is softer and different, but the first and most primitive cars with a carburetor in the form of a living granny have already appeared -)))) (just read who you used as a carburetor Great Daimler-)))
                    "between the first carriages and the latest models of cars. Accordingly, not only software - hardware is needed different, on completely different principles of work, not the one on which you are rested with a mountain - this miserable" on-off ". I myself do not know, on what principle. But I can guess - for example on the principle of holography. How do you?
                    Let's explain for a start. Do you remember what holography is based on? A coherent ray is taken, split into two rays - the reference and the object, the one that is the object, is sent to the object. Then both rays converge on the photographic plate, between them, naturally, there is a phase difference, which creates an interference picture - in fact a 3D model of an object.
                    Here is the most interesting:
                    1. People also create a 3D model of the surrounding space. You don’t need to understand this too straightforwardly, people don’t emit any lasers from the eyes, there may be other mechanisms - for example, shifting the information obtained using photons to other frequencies, it turns out that the object beam has arrived, as if in vain everyone wanders here and there these alpha and other rhythms, folding with the frequencies received from the eyes into an interference pattern.
                    2. If we cut a holofoto with an object, say a rose, in half, we get two roses. If four - four roses. And so many times. In other words, at every point of the holofoto all information about this object is contained. There was such a scientist - I don’t remember the name, he cut many brains of rats, hoping to find a place in the cortex where the memorized is stored. Have not found. I came to the conclusion that what is memorized is remembered by the whole bark like a holographic one.
                    3. Golofoto already initially has the ability to recognize objects, and at an associative level, that is, even if the object does not coincide in details, but coincides in the main, recognition will be carried out correctly.
                    I can also indicate a bunch of suspicious properties of holography, but I think this is enough.
                    And now let's think - is it possible to create "hardware" not on the basis of your dreadfully dreaded binary "on - off", but on the basis of holography? And five does not need to be understood too straightforwardly, I mean based on the interaction of vibrations? In theory, there is such a possibility. Why do you accuse me of opposing
                    Quote: snek
                    aksakal, in your comments you contrast hardware and software
                    - This is generally a different principle of information processing, and the Russians are not far behind here! And you don’t need to develop your on-off chips further, get busy with a fundamentally different chip! And in the field of improving chips "on - off" do not catch up, this time is lost outright.
            2. borisst64
              borisst64 15 October 2012 16: 28
              0
              Quote: aksakal
              calculates all options


              I read somewhere that for 100-cell checkers there is no program because of the colossal number of options. And all that is two boards wider board.
              1. aksakal
                aksakal 15 October 2012 17: 41
                0
                Quote: borisst64
                I read somewhere that for 100-cell checkers there is no program because of the colossal number of options. And all that is two boards wider board.
                - I agree, but also read about the power of Deep Blue (there is some kind of floating point) - there are no question about the hundred-cell checkers
  2. Igor
    Igor 15 October 2012 09: 35
    0
    Quote: aksakal
    And one more thing: Russians have a backlog in microelectronics. Is it really necessary to overcome it? Chips may well be carriers of artificial intelligence.


    Chips are already one of the parts of microelectronics.

    Quote: aksakal
    Chips already produced by Russian industry may well be carriers of artificial intelligence.


    The equipment on which these chips are manufactured, everything is old and foreign, amers have long passed us around the chips.
  3. aksakal
    aksakal 15 October 2012 09: 56
    +2
    Quote: Igorek
    The equipment on which these chips are manufactured, everything is old and foreign, amers have long passed us on chips
    - you do not understand the message. So what if those chips are old and foreign? I meant that the improvement of chips, literally the struggle for every angstrom of miniaturization, in which every "mined" angstrom is given with more and more difficulty, may no longer be so relevant.
    For example, the old Soviet machine "Kaissa", made on slop by the present times, beat the grandmasters in the same way as the super-modern "Deep Blue" did recently. Why, while losing to this Deep Blue in the flight a million times, nevertheless, the kaisa managed to do almost the same thing with almost the same result? The difference is in the algorithms. Deep Blue is just stupid computing power, just dullly sorting out all possible scenarios on the chessboard, of course, there are algorithms that immediately discard unsuitable options, otherwise it is simply impossible to consider everything. And kaissa, due to the limited power of this, cannot come close. The developers, among whom Smyslov took part, immediately followed the path of imitation of human thinking - the use of memory when considering positions, a set of standard positions, of which there are a great many in chess, such as the Sicilian Defense and others, the use of associative algorithms in order to see the similarities between the existing situation on the chessboard and any standard position with further refinement of this standard position to the current situation on the chessboard. All this gave its effect.
    The situation is similar when using AI in military affairs.
    IMHO, the finalization of the chips you love so much. spending on this "catching up" huge funds - into the furnace! The future is not for your favorite chips, but for those who will quickly understand how a living creature thinks and embody it in "hardware". Moreover, the hardware itself is not required for this directly from the latest generations.
    1. Igor
      Igor 15 October 2012 10: 26
      +3
      That is, you propose to create such an AI that will think not only as a computer that goes through all possible solutions, but will be able to think outside the box, like a human being, but such an AI will be able to improve its platform and create other platforms on its base without human help! And what will she do with a person who tries to limit her in this?
  4. aksakal
    aksakal 15 October 2012 11: 04
    0
    Quote: Igorek
    That is, you propose creating such an AI that will think not only as a computer, which will sort through all possible solutions, but will be able to think outside the box, like a person
    - thank! Very flattered. But I regret to say that I am not proposing that work in this direction, albeit sluggishly, is going on without my desire or reluctance, and regardless of my suggestions or suggestions-))))).
    Quote: Igorek
    But such an AI can improve its platform and create other platforms on its base without human help! And what will it do with a person who tries to limit it in this?
    - Output? Do you propose stopping scientific and technological progress? Will it succeed? Like already tried, the result is known.
    I think this is still a long way off, at least not in our life with you, but because this is not our headache, but our sons, and even grandchildren. and further. Because I’m not worried.
    In the field of information processing, there are subregions where Russians will not be able to get away from stupid computing power - well, such as a computer simulation of a nuclear explosion or cases related to weather forecasting. There are no solutions, just computing power is needed. Accordingly, all the same, Russians will have to be able to create supercomps, too, and everything related to this is chips and other. Well, there is a sub-area related to decision-making. By the way, recognition (in military terminology, we have in front of us an enemy tank or some kind of tractor ?) - This is a subdomain of decision theory, such as what class the observable belongs to? IMHO, it is this subdomain that is of greater importance in military affairs and therefore the emphasis here is on the development of a new type of algorithms - they will give a greater effect than the Amer way, a way to bluntly increase computing power.
    1. Green
      Green 15 October 2012 12: 21
      +2
      I say the same. Relax. The reality is that there are brains in Russia. There are original ideas that are in demand and are being implemented successfully. IN THE WEST,
  5. Igor
    Igor 15 October 2012 11: 36
    0
    Quote: aksakal
    - Output? Do you propose stopping scientific and technological progress? Will it succeed? Like already tried, the result is known.


    No, it’s just to limit it in development, because the creation of such an AI is like creating a new life that will surpass a person in its development in all respects, and as we all know, the weak will gradually die out completely. A person’s connection could be a counterbalance to this form of life cars in one piece, but that's completely out of the realm of fantasy.

    Quote: aksakal
    I think this is still a long way off, at least not in our life with you, but because this is not our headache, but our sons, and even grandchildren. and further.


    I think so too.
    1. aksakal
      aksakal 15 October 2012 12: 21
      +1
      Quote: Igorek
      . A counterbalance to this form of life could be a combination of man and machine into one whole, but this is completely from the realm of fantasy.
      - why? Implantation into the cranium, for example, of a microcalculator with a good interface "digital algorithm - brain algorithm", such that they would not even feel its work, but the question "how much is 263 multiplied by 365?" an answer like "95995" immediately appeared in my head - this is a big psychological problem for us, akin to eating an insect alive like a thread. The commandos, by the way, argue about the great nutritional value of this, however, they are silent about the taste. But I'm not sure that this will be as much of a problem for my son as it is for me, and even less sure that it will be a problem for posterity further away. Accordingly, I can assume that the clearly visible line "natural man - artificial robot" will be erased due to the assumption you made, to the extent that it is impossible to unambiguously define this or that individual of the future. I don't see anything fantastic in your suggestion.
      Quote: Igorek
      I think so too
      - on that and decided -)))))
      1. Igor
        Igor 15 October 2012 12: 43
        0
        Quote: aksakal
        why? Implanting into the cranium, for example, a micro calculator with a good interface "digital algorithm - brain algorithm"


        No, I’m not talking about this, you can implant a chip in your head right now. I’m talking about connecting human DNA and the DNA of a computer (I know that a computer does not have DNA, but I will call it a three-letter word laughing ) and then the living organism will consist not only of organics, but also of synthetics and will be able to produce it without any outside interference.
        1. aksakal
          aksakal 15 October 2012 13: 21
          0
          Quote: Igorek
          I'm talking about connecting human DNA and computer DNA (I know that a computer does not have DNA, but I will call it a three-letter word)

          - I don't know, why go so far? Why create DNA for robots too? Not necessary. Start self-replication - when some bots create their own kind, and even an evolving self-replication, with the improvement of bots with each generation - and DNA is not needed at all. I meant something else - after a hundred years, a person who was born naturally and with good finances gradually replaced all the failed living organs with artificial ones - everything artificial - the liver, the stomach ... Even the brain - by the way, is a science dealing with issues transferring the structure of the brain and even the structure of the personality to an artificial medium, called "settleretika", is quite dynamic -))). This is the person - who is he? A robot or is it a man? I wanted to convey this idea -)))))
          1. Igor
            Igor 15 October 2012 13: 48
            +1
            Quote: aksakal
            That person - who is he? Robot or people?


            If this brow remains feelings: anger, love, joy, disappointment, etc. then you can safely call it a man, and if instead of a brain they put a processor to him ... all the joys and disappointments of this life, then despite the fact that all the other organs from his meat he will still remain a soulless machine.

            Quote: aksakal
            - I don’t know why to go so far? Why create DNA for robots as well? Not necessary. Run self-replication - when some bots create their own kind, and even evolving self-replication, with the improvement of bots with each generation - and DNA is not needed at all.


            But then they won’t need a man, like these machines.
            1. aksakal
              aksakal 15 October 2012 14: 10
              0
              Quote: Igorek
              If this brow remains feelings: anger, love, joy, disappointment, etc. then you can safely call it a man, and if instead of a brain they put a processor and he’s up to .... all the joys and disappointments of this life, then despite the fact that all the other organs from his meat he still remains a soulless machine
              - a good criterion -)))). I agree! Emotions are difficult to formalize and, therefore, to translate into algorithms. But "difficult" does not mean "impossible". Let's imagine that an emotional robot has been created. Then I will repeat my question in the above post, but considering that robots are also experiencing
              Quote: Igorek
              feelings: anger, love, joy, disappointment, etc.

              - then what is the quality of the criterion?
              1. Igor
                Igor 15 October 2012 14: 53
                0
                Of course, you can reproduce a certain emotional reaction of the machine to a particular event, but it will be an artificial reaction deep down in her soul to this event and it does not cause any feelings for her (a kind of mask, people of course also often wear such masks (they lie, pretend ), but most of their feelings are sincere, they can react differently to one or another event). Yes, and a person forms his personality throughout his life, no one uploads a file to his head what it should be, of course, on the formation of his personality Parenting, a school, etc. are in effect, but first of all, he decides who to be: a scoundrel, a hero, etc. Yes, and the car cannot be broken morally, but only physically, but a person can be broken both morally and physically.
                1. aksakal
                  aksakal 15 October 2012 15: 31
                  0
                  Quote: Igorek
                  Of course, you can reproduce a certain emotional reaction of the machine to a particular event, but it will be an artificial reaction deep down in her soul to this event and it does not cause any feelings for her (a kind of mask, people of course also often wear such masks (they lie, pretend ), but most of their feelings are sincere, they can react differently to one or another event). Yes, and a person forms his personality throughout his life, no one uploads a file to his head what it should be, of course, on the formation of his personality Parenting, a school, etc. are in effect, but first of all, he decides who to be: a scoundrel, a hero, etc. Yes, and the car cannot be broken morally, but only physically, but a person can be broken both morally and physically.

                  - so far no one can say anything for sure - this is all a big white spot in science, and it pleases that it was the Russians who advanced a little in this spot (and it's a pity that just a little) in this white spot.
                  Speech about AI here is a little different. Here was a good attempt by the Kamovites - to make one combat helicopter with one pilot. It is precisely because of the unavailability of AI technologies that this venture failed. Let me illustrate with an example. To fit into a turn on a car (in addition to driving, you also need to hold a machine gun in your hands - for the best possible analogy with the Ka-50), you need to press the brake, do some manipulations with the gearbox (go to the lower shift), do the appropriate manipulations with the steering wheel, at the end of the turn, manipulation with the gas and shift to top gear. In the case of an automatic transmission, everything is greatly simplified. If you still "twist" the car (introduce a road scan with the creation of a 3d model, equip it with biometric sensors and, in general, brains for making a decision, and a bunch of bells and whistles), then you can achieve even fewer manipulations to make a turn. Or even no manipulation at all - he ordered mentally, the driven machine performed. It is clear that in this situation the machine must extract the information necessary for effective assistance to its "owner" not only from its owner, it must also take into account the contextual information on the situation. Let me explain briefly - after work I get into the car and just say "home" to the driver. According to the situational context - the end of the working day, I look tired and so on - guess three times where the driver will take me? After all, the word "house" in human language has many meanings - it is an abstract building, and a concrete building, and someone's abstract dwelling, and my dwelling is concrete.
                  In order for military equipment to become truly smart, you will have to work a lot on AI. And in no case should anyone be allowed to come forward while Russia is ahead - this is not only about Eugene in this article, there are a bunch of specific articles, according to which I also judge - Russia is still ahead.
    2. Green
      Green 15 October 2012 12: 24
      +3
      I think this is not so far. And people are vile and stupid looking. I think that if AI destroys us all once, then it will be absolutely right. But seriously, but the future is a fusion of the biological brain with the electronic. And this will be an evolutionary process that will put an end to our mind in the usual sense. But it will create a new, more perfect form of the mind.
      1. aksakal
        aksakal 15 October 2012 13: 13
        0
        Quote: Verde
        I think not so far

        - it is difficult to accurately pronounce the dates -)))). This is evolution, and it can be of three types:
        1. Type one. The evolution of active subjects in a passive environment. An example is the evolution of the brow, which is better and better adapting to an environment that does not change much.
        2. Type two. The evolution of passive subjects in an active environment. An example is the evolution of a car. cars themselves do not evolve, but an active environment in the form of human civilization quite successfully forces them to do this.
        3. Type three. Evolution of active subjects in an active environment. If we do not take into account the selection of agricultural animals - there, due to the large difference in the rate of evolution, a more second type of evolution is obtained, then this type has not yet been realized in nature. The current start of work on AI is also going along the second path, "Evolution of passive subjects in an active environment," but it can reach the point where it will be possible to start evolution according to the third type, and the evolution rates will be of the same order, then it is generally difficult to say with what evolution can proceed. Perhaps, "Evolution of active subjects in an active environment" is evolution squared in terms of speed .-)))))
  6. Egoza
    Egoza 15 October 2012 14: 31
    +3
    AI is, of course, good. But most importantly, WHO and HOW will use it. Yes, and human emotions can hardly be embedded in AI. But a person in the process of thinking is often guided by emotions.
  7. KA
    KA 15 October 2012 22: 55
    0
    Being interested in AI problems, I came to the conclusion that at the moment the necessary elemental base for creating a full-fledged AI is not thought out. Although the elements of AI are already appearing and will be even more in demand in the near future.
    Ideas, theories, algorithms ... this is certainly good, but without the technology of manufacturing the element base, they will never be realized! Therefore, all our advanced achievements are implemented abroad, and not with us.
    1. postman
      postman 15 October 2012 23: 59
      +2
      Quote: KA
      AI came to the conclusion that at the moment the necessary elemental base for creating a full-fledged AI is not foreseen

      elemental base and AI are practically not correlated.
      You are on the path to productivity.
      and the problem of a full-fledged AI, not in the number of Terabytes or GHz that the machine can / should overpower, but in the postulate / message itself:
      AS ??
      Here it’s not even the language of software or the environment, but the principle is different.
      in any case, not binary logic (1 and 0 / Yes - no), and those ternary (+ 0 +) / Yes - not sure, no.
      The thing is different.
      1. aksakal
        aksakal 16 October 2012 07: 50
        0
        Quote: Postman
        You are on the path of productivity. And the problem is a full-fledged AI, not in the number of Terabytes or GHz that the machine can / should overpower, but in the postulate / message itself: HOW ?? There is not even the language of software or the environment, but the principle another. in any case, not binary logic (1 and 0 / Yes - no), and those ternary (+ 0 +) / Yes - not sure, no. The thing is different.

        - Plus, at least one like-minded person understands what it is about. Up there, my long posts, in which everything I tried to convey and did not convey.
        But at the same time, you can build AI on simple logic. Binary logic corresponds to the linear processing of information and thus the linear part of the equations. But there is, after all, the nonlinear part of the equations, it has been much more studied, but there is the Tuckens theorem, it says that any arbitrarily complex nonlinear function can be represented in a linear form. In other words, the famous Lorentz butterfly, represented by a system of three diffours, can be represented as a set of terms and variables in the first degree. Only to get a clumsy butterfly, you need ... 120 (!) Of these terms, and that would more or less - and even more. In short, we are trying to describe with simpler tools - as a result, we complicate it.
        You can create AI on the binary logic, but the software there will be such that it can hardly be created and it is unlikely that there will be such a computer even at the final stage of Moore’s law that will pull such software. In short, it is impossible. There is a law of conservation of energy, there is a law of conservation of mass. And if there are two information-related elements, then the law of conservation of complexity applies between them. We complicate one element - thereby simplifying the second. If the eye is simple and can only distinguish pixels, then for him even a child’s drawing of the house will be of such complexity. If the eye is complicated to the point that it will be able to perceive lines, intersections of lines, then the drawing will be just a drawing. Nobody formulated this, in any case I have not met, so for now write this law in my name, the name of Aksakal -))). Or send a link where they got ahead of me -))))
        It is necessary to complicate the basic element from binary logic to a much more complex one, presumably the basic element should be some kind of oscillatory circuit such that the interaction of the waves of two such oscillatory circuits will be a complex dynamics with chaos elements. Well, who said that the brain is a simple thing?
        That something like this.
        1. postman
          postman 16 October 2012 10: 23
          +1
          Quote: aksakal
          But at the same time, on a simple logic, you can build

          possible. but so far only theoretically. and indeed on a binary it will be difficult
          But the main question is nothing to build on (theoretically, it is possible to build on a mechanical computing device).
          the question is: HOW does all this work.
          For mankind, the difficulty to comprehend / understand himself, using for this, as a tool of himself. Briefly, the brain must understand how the brain functions.
          Just everything that you describe (above): the methodology / approach of a mechanogenic civilization.
          Simply put: sorting by signs with the search for the coincidence of most of the signs and the development of a solution.
          If (condition), the result of the comparison, the development of a solution (Yes, no, or).
          Moreover, or (almost everywhere) is implemented artificially (programmatically), and not on hardware (+, 0, -).
          Living (intellect) works differently, not by brute force, especially when solving an abstract problem.
          1. aksakal
            aksakal 16 October 2012 13: 25
            +1
            Quote: Postman
            For mankind, the difficulty to comprehend / understand himself, using for this, as a tool of himself. Briefly, the brain must understand how the brain functions.

            - you are a very interesting conversationalist -)))).
            Quote: Postman
            For humanity, complexity

            - but people can go beyond themselves and watch (read, study yourself from the outside). The body is studying methods of treating many diseases.
            Quote: Postman
            especially when solving an abstract problem

            - it's pretty easy to formalize. Chel constantly operates with abstract things, jumping from one level of abstraction to another as needed. Let me explain: when a child develops abstract thinking, he begins to draw, well, the same person in a very peculiar way - four sticks as a star in a five-pointed star and a round in place of the fifth. This is the first abstraction, "man in general". An abstract scheme of a chela, when compared with an abstract scheme of an animal, for example - the same 4 sticks, but connected to a fifth horizontal stick and a round at the end - allows you to instantly recognize a person and instantly distinguish it from any other living creature. It is clear that the most important thing is left in abstraction, details are discarded. In the image of a chela, for example, the length of the arms is taken according to the prevailing size, and for example, the thickness (and the hand of a bodybuilder may differ in this size from the hand of a supermodel in this part, sometimes so that it even happens to be noticeable with the naked eye) and others are simply discarded. Is it really difficult to algorithmize this rule? There can be several levels of abstraction in total. For a tiger - to distinguish it from an insect - the level of "insects" (round and many crooked sticks around the radius) and animals (already described). To distinguish a tiger from its potential victims - also its own level, which reflects the main distinctive (read informative) features. To distinguish a tiger from a lion - the level of felines in general, with distinctive features between felines. The difference between the tiger Petya and the tigress Masha - a normal person does not descend to this level, it is to the Bagdasarovs and Zapashny. Personally, I prefer to operate at the level of "tiger in general" and closer in any way, I don't know about you.
            And here people, depending on the context, can freely "run" along the levels of abstractions they have formed and thereby successfully solve many problems. This is not difficult to formalize.
            Personally, another thing complicates me. Many components of thinking in general (well, I suspect) are impossible without emotions along the following chain - emotions, turning into impressions and being processed (laid in the background by memory for certain events) into memory, form the personality structure, and many components of thinking are created precisely by the personality structure . In other words, these components are impossible without the AI ​​living for a certain time in a social environment, while endowed with a whole set of emotions, this is really difficult to formalize, and without these components the AI ​​will be clumsy to nothing_)))), if at all possible
            1. postman
              postman 17 October 2012 23: 24
              +1
              Quote: aksakal
              - you are a very interesting conversationalist -)))).

              ATP (thanks i.e. you then)

              Quote: aksakal
              - but people can go beyond themselves and watch (read, study yourself from the outside).

              To study the liver or kidneys by a person of another person using his brain is not difficult. A dozen ruined lives and you know.
              but to study the brain with the brain '(not anatomically, of course), but functionally (by the process of thinking) is unlikely.
              To understand is to be able to recreate.
              this is how we understand and create AI. Although far from that.
              The IR sensor (microscope) will not understand and study the IR sensor (microscope) / if I could think of course /,

              Quote: aksakal
              - formalizing is quite simple.

              It is difficult just.
              How does cancer determine top-bottom?
              The organ of balance of the crayfish is a chamber with grains of sand (the stalk of the "antenna") - he himself puts the sandstone there, after each molt.
              Everything is just clear, reproducible, but here to reproduce in nature the ENTIRE equilibrium sensor, and so accurate, no.
              And the process of formation, transmission of a signal to the brain of cancer (of which the Internet is practically), analysis, development of a control signal ......
              The CHIP, communicator and software repository will probably be the size of the cancer itself.

              Sandstone caught in delicate tissue
  8. tambu
    tambu 16 October 2012 02: 55
    0
    An artificial mind has been created that can evolve independently, said American researchers from the University of Michigan.




    According to the British weekly New Scientist, artificial intelligence takes the form of complex computer programs. These electronic organisms, in turn, have their own "DNA" (original program codes). The discovery is that scientists were able to endow these "DNA" with the ability to independently mutate and acquire new properties and abilities.

    As a result, digital organisms have learned to use this feature, which was previously only characteristic of biological structures, experts said. A process of "multiplication" of these organisms arose, as a result of which each new generation of them becomes more and more "smart and skillful".

    According to experts, the world is on the verge of the emergence of "smart machines" capable of thinking and performing a number of functions inherent in the human brain.

    http://www.medik-portal.ru/augyst2010/iskysstvennuiintellekt
    1. aksakal
      aksakal 16 October 2012 07: 26
      0
      Quote: tambu
      As a result, digital organisms have learned to use this feature, which was previously only characteristic of biological structures, experts said. A process of "multiplication" of these organisms arose, as a result of which each new generation of them becomes more and more "smart and skillful". According to experts, the world is on the verge of the emergence of "smart machines" capable of thinking and performing a number of functions inherent in the human brain.

      - in my opinion, banal so-called genetic algorithms, some thread modification
      1. postman
        postman 17 October 2012 23: 14
        0
        What does digital organisms mean?
        Digitized organisms introduced as monocells (imitators of a living unit) into a programmed computer environment?
        So in the digger, the worms multiplied. And now I can plan such bots in a day, even in C #, they will multiply and recycle

        If there is something similar to "Roy" or at least like S. Lem in "Invincible", then oh, how far away is it. Rather, the Moscow metro will gain reason because of its complexity.
        What was meant. I don’t get it?
  9. Aryan
    Aryan 17 October 2012 14: 32
    0
    why spend a thuja heap of bablos on nanochips and programs,
    When is it easier to teach a frog to solve the Pythagorean theorem? fool
    1. postman
      postman 17 October 2012 23: 08
      0
      Quote: Aryan
      When is it easier to teach a frog to solve the Pythagorean theorem?

      Probably only likely using the frog's brain (separately isolated and properly commutated).
  10. Uruska
    Uruska 18 October 2012 14: 17
    +1
    This "group of Russian scientists" developing AI, at the same time develops "psigenerators", and then puts them on the computer of Russians. Therefore, round-the-clock dialogues are heard in my head.