Many of the questions that pondered the minds of the largest human civilization, over time, become trivial and obvious. They are included in the school curriculum, and their lack of understanding is a sign of mental retardation. This is a normal process of evolution of knowledge. But challenges remain from the category of "eternal" that have not been solved, and it is unknown whether they have a solution at all. These include understanding what is intelligence, definition of artificial intelligence, and all related practical issues. Scientific basis of modern developments in this field was the work of Norbert Wiener in which he showed that a basis of information processes occurring in a living organism are complex positive and negative feedback, which can be modeled in artificial automatic device.
The practical importance of the topic became apparent not so long ago, if we apply the time scale of the history of human thought. Today, however, "intelligent agents" type Copernic'and or recognition systems are installed almost on the each computer. The reason why this question did not find a clear solution so far, is that for its understanding it is necessary to apply different principles of thinking, different from those that are basic for most media, not only "artificial" and "natural" intelligence. It is impossible to understand something from the inside. Necessary to look from the outside. Only this approach can explain the phenomenon in its interaction with others. In other words, in order to understand how we think, we must think otherwise, or not think at all... And understand this is necessary, as our "iron" of the creation is already close to the level when the attitude towards them, just as technical AIDS, it may be invalid and cause strategic errors.
To begin the analysis of existing artificial intelligence systems and make assumptions about their further development of the preferred program of recognition of graphic images and speech. This task is considered isolated, still does not answer the question of what artificial intelligence or ability to think at all, but allows you to come close to him. It's no accident that the classics of Marxism-Leninism was tied up in his philosophical works, the question of the emergence of the "reasonable person" with the development of speech.
If we talk about the voice processing software from Dragon Systems already practically decided this question. The fifth version of the Speech SDK from Microsoft, being, like all works of this company, much more cumbersome (more than 500 MB on the system drive!) and demanding of resources, also allows you to provide a suitable for dictation quality. Sustainable recognition of words and even whole phrases achieved after several hours of training and adaptation of the system to pronunciations and generate a statistical model of the speech of the user. It can reach 95-98 %. If we compare, then, perhaps, people will not be able to parse spoken language more precisely.
The principle of the mentioned software is a mathematical model of transformation of acoustic signals in a numerical sequence, each of which corresponds to a particular word from the pre-loaded dictionary. Dictionaries can be supplemented by the user, and the probability of selecting from a list of similar numeric parameters of words changes depending on the frequency of their use by a specific user. Recognition of graphic images, from standard font to legible handwriting, known to most readers via the software from the company ABBYY (Fine Reader), is based on the same principle.
However, the similarities between artificial and natural or biological systems is of a purely external nature. It is necessary to clarify terminology. In the case of listed software products we are dealing with the detection or recognition of images, but speaking about biological systems in General and man in particular, to speak about their understanding. What's the difference?
Dictating his badly Russified "Dragon" (Dragon Systems), it is easy to find the printed phrase that "the Sun was brightly blue in the sky." And the Dragon can be proud of what has defined this character sequence with 90% probability, as the word "shone" and "blue", from his point of view, almost do not differ. If the program is to consider more data, not only in words but also of phrases, (the last versions of a range of Natural Speaking already do it), and the process of training will be not days, but months, which in human terms is quite fast, of course, next time she'll write down that phrase correctly. But the principle of "recognition" will still remain recognition, not understanding.
From the point of view of "understanding", it is first necessary to exclude the possibility that narrative comes from the person dog, which is out of all colors of the spectrum see only shades of blue, and, consequently, for her the sun can really "turn blue" as anything else to him remains. But seriously, for a proper understanding of the spoken phrase, you should have an idea about the described images of reality. Must have experience and remember the sensations of heat, summer, the sea or another situation in which this description could take place. The words (no matter what language, as for understanding the choice of language is not fundamental) are only a reflection of this reality. From the point of view of our "Dragon" (or another system of pattern recognition) it would be possible if you equip it with temperature and light spectral sensors and send on vacation, say, in Antalya. In other words, to provide him with the entire set of senses similar to a human, and let go through the learning process with the purpose of accumulating the database of life experience at least five children. The conclusion is simple — adequate human understanding of speech or other inputs is only possible when the identity of the life experience and input devices. After all, the recognition program can be quite perfect and adequate, but only to itself. All her "senses" consist of a microphone and keyboard, and life experience — is the hours of training and memorization of sentence patterns at the dictation that she actually "lived" as a unique virtual "creature", in collaboration with their surrounding world, the limited voice of the "master".
But this is still not enough for "understanding". One of the most important features of systems of artificial and "natural" intelligence is the ability to self-education. In systems of pattern recognition, in addition to the primary training initially installed dictionaries and symbol tables, there is the process of correcting recognition errors, which also is stored and becomes part of their "life experience". This is very similar to training. However, the possibility of stimulation here is very limited. Correctly a certain word (or character) is "default", but the error requires manual input of desired values and possible additional training. In this system, in General, still, correctly or not was determined by this sequence. Negative emotions accounted for by user-handler. With Pavlov's dog, the situation is more rigidly: for the correct sequence of actions is sugar, but for a mistake — and you electrocuted yet (or just blow) to. It would be fair if the manual correction of recognition errors (or, more generally, also correct from the point of view of the host reaction to a command or signal) called "negative emotions" from the system. Imagine — you make a correction to dictated text, and your "Dragon" plaintive cheeps of resentment and feelings for their incompetence (or maybe it is?)!
In nature, there are two mechanisms of improvement of the behavioural reactions (in our example this is the output in text form is dictated by the recognized material). For the simplest organisms is, generally, the principle of natural selection, when the most adequately responds to the surrounding environment and individuals are the most prolific, and the responses are fixed in the genetic code. For the most advanced representatives of and, respectively, for more complex behavioral reactions becomes the main mechanism of motivation through emotions. Not just physical pain, hunger, or the desire to reproduce, but also the whole world of suffering and joy — often it is not clear at first glance reason — it becomes the force that motivates a person (or only person?) the path of learning and development.
Remember the Book of job (Chapter 5): "But man is born to trouble as sparks to fly upward". The only question is, positive or negative motivation (stick or carrot) prevails in each specific life. Usually have to deal with both variants, but in someone as lucky...
This implies that, in addition to the inadequacy of perception of the world, the second feature of contemporary artificial intelligence systems is the lack of incentives or motivation for the development of both the individual and species level. But this difference is not insurmountable.
If to talk about development at the species level, it is already known, "mutating" viruses, which, although not an intellectual in the full sense, but already quite capable in the process of natural selection in an artificial environment to pass on to their "descendants" the most pernicious features of preventing their capture.
The question of motivation or incentives to develop on an emotional level is also completely solvable. It's enough to understand that the emotion, positive or negative, is only, respectively, positive or negative feedback, characteristic of any automatic device. Let me explain: a negative emotion or negative feedback is the process in which the inherent desire for its cessation. Accordingly, positive emotion, or positive feedback, is the process of carrying the trend to its continuation or multiple recurrence. Options for their technical realization are numerous, and virtually any device — from classic toilet tank with a float valve to the modern electric — contains a few examples of feedback. Natural and natural system is literally full of their various options. It is important to note that, in contrast to the negative, positive feedback is always prone to saturation, that is, the termination of its "positivity" over time, or after repeating several cycles. This is a key feature. Purely worldly conclusion — the suffering is infinite, and the pleasure is always limited ("eternal high" is unattainable). So in the above quotation from the book of job, unfortunately, more truth than we would like. And who ever heard of a satisfied and happy the subject has created something worthy of attention, except in the culinary field?
The internal contradictions, conflicts and frustration has always been the norm in a creative environment. You can attempt to deduce the General law of the transformation of the amount of dissatisfaction and suffering in the momentum of development, but the categories are too General and vague. One thing is clear — the Lord God through the application of the theory of evolution and the theory of games fully equipped us with complex, conflicting desires and related emotions for our own development for the benefit of common progress. Moreover, the extent of the personal suffering of each of the participants does not matter and merit is not, what is the ultimate goal of the progress — God only knows. As this is an obvious point to combine with the image of the Christian God, loving and forgiving me is not clear. Maybe someone on the forum will explain. Irina Bogushevskaya on this subject there is a wonderful song with the words: "Heaven would be like to be with us"... Poetic and scientific perception of the world often lead to the same conclusions.
But back to our sheep and the Dragons, with some signs of intelligence. What if to stimulate learning, the Lord God, had done so with us, give them unlimited suffering and temporary joy?
It turns out that such work is already underway. As an incentive, the selected basic sense of hunger, or to explain to those who don't know what it is, physiological sense possible cessation of existence in the absence of some external components. This is a unique series of "Gastronomical" extracting electricity from the processing of proteins and carbohydrates in a special bioreactor. The first machine in a series of ever-hungry creations of man is "chew-chew" (Chew Chew), designed by Stuart Wilkinson (Stuart Wilkinson) at the University of Tampa (South Florida). The inventor believes meat is an ideal food for its establishment, however, while the "chew-chew" eats the sugars, as this, to the delight of the technicians, it leaves at least "feces". The main strategy of the robot is exploring the environment is quite real, for the detection and consumption of food. The adequacy of perception of the world is quite conventional and is achieved using ultrasonic sonar. "Chew-chew" until you can sneak up on you quietly, as it consists of three trailers on wheels, about a meter long each and similar to the train of the children's railway. The constant feeling of hunger — its normal condition, as the batteries require constant recharging. A continuation of the project should become an underwater car, eating fish and are intended to patrol the coast and protect swimmers from sharks. As much as the programmers were not mistaken and have not created real steel "jaws", because the swimmer more easy prey than the shark and Autonomous "chew-chew" could easily be retrained in "couscous". And what thoughts are in his ever-hungry electronic brains when he will be watching defenseless "Creator", floating on top of the surface?
There is no doubt that the development of science and technology in the field of artificial intelligence creates a whole set of issues that require resolution. Among them are moral and ethical issues. And hungry, hungry for meat, "chew-chew" — perhaps not the most important of them. Some of them will be discussed in subsequent articles in this series.
Now I would like to formulate a conclusion which may help you to take a different view on the surrounding objects, animals and people.
It is that almost everything near you who have in one way or another the power of reflection of the world, to the suffering or joy and shows some signs of intelligence, not always artificial, and often it is of natural origin. What you don't notice them, does not mean that they are not. They're just not compatible with your unique perception of the world.
Maybe not so wrong Hindu monks and followers of some ancient teachings, odushestvlyayuschaya all surrounding objects, and gives them the ability to "think", to suffer and rejoice?