In the movie "I, robot", made on the works of Isaac Asimov, in 2035 creators launch into operation the most advanced in the history of the computer system. It has its own name - Wiki - virtual interactive kinetic intelligence) and is designed for perfect management of the life of a big city. Under its control is everything from underground and electric networks to thousands of domestic robots. The program Wiki iron principle is to serve humanity.

But once Wiki asked a key question: what is the main enemy of mankind? Mathematical logic led to the unequivocal conclusion: the main enemy of humanity is humanity itself. It urgently needs to be rescued from an unhealthy desire to destroy nature and start a war; you cannot allow him to destroy the planet. For the Wiki is the only way to fulfill the main task - to take power over humanity and to establish a fertile machine dictatorship. In order to protect humanity from itself, it is necessary to enslave it.

In this film raises important questions. Taking into account rapid development of computer technology, can we expect that someday take-over by machines? Will the robots so developed to represent a real threat to our very existence?

Some scientists answer this question negatively, because the idea of artificial intelligence is no good. A chorus of skeptics in one voice says that to create a machine that can think, impossible. Skeptics say that the human brain is the most complex system, created by the nature for all time of its existence (at least in our part of the galaxy), and any attempt to reproduce artificially thinking process is doomed to failure. The philosopher John Searle from the University of California at Berkeley, and even the famous physicist Roger Penrose of Oxford[20] sure that the machine is physically unable to think like a man. Colin McGinn from Rutgers University says that artificial intelligence is like the slimy, who would have tried to do psychoanalysis according to Freud. There is no need for this organs".

Can machines think? For more than a century, the answer to this question divides the scientific community into two camps.
The history of artificial intelligence

The idea mechanical beings captures the imagination; it has long been settled in the minds of inventors, engineers, mathematicians, and dreamers. From the tin man from Magic country to robots-children of "Artificial intelligence" Spielberg and robots-murderers of the "Terminator" - everywhere machine, able to act and think like men.

In Greek mythology the God of the Volcano was forged of gold mechanical servants and did a three-legged tables, able to move by themselves. In 400 BC Greek mathematician Architect Tarentsky wrote about what could be done mechanical bird, which would move through the power of steam.

In I century Heron of Alexandria (he is credited with the invention of the first steam engines) have done machines, one of which according to legend was able to talk. Nine hundred years ago, al-Jazari invented and designed such automatic device, like a water-clock, all kinds of kitchen accessories and musical instruments, is driven by the force of the water,

In 1495, the great Italian artist and scientist of the Renaissance Leonardo da Vinci drew a diagram of a mechanical knight who could sit, move the hands, head and to open and close the jaw. Historians believe the scheme da Vinci first realistic project humanoid machines.

First functioning, though rough robot built in 1738 Jacques de Vaucanson; he did Android, which could play the flute, and the mechanical duck.

The word "robot" was invented in 1920 Czech playwright Karel Capek in the play R.U.R." (the word "robot" in Czech means "heavy tedious job, and in Slovak - just "work"). In the play the enterprise appears under the name "Universal robots Rossum", serially producing robots for unskilled labour. (However, unlike conventional machines these robots made of flesh and blood.) Gradually the global economy falls into the full dependence on robots. But treat them horribly, and eventually robots rise up and deal with the owners-the people. However, in anger they killed all scientists, able to repair robots and create new ones, and thus doomed to extinction. In the final play two robots special models find yourself with the ability to self-reproduction and become the new Adam and eve era robots.

In addition, in 1927 robots became the heroes of one of the first and most expensive of silent movies of all time - film "metropolis", shot in German Director Fritz lang. The film is set in 2026, and the working class doomed to endless work for the terrible and dirty underground factories and the ruling elite has fun on the surface. One beautiful woman named Maria manages to win the confidence of the workers, but the rulers are afraid that some day it may raise people to revolt, and therefore turn to the villain-a scientist with a request to make a manual copy of Mary. This plan, however, turns against the authors, the robot picks up the workers to revolt against the ruling elite and thereby causes a system crash.

Artificial intelligence, or AI, is significantly different from the technologies that we have discussed so far. The fact that we still poorly understand the underlying this phenomenon fundamental laws. Physics decent understanding of Newtonian mechanics, MaxValue theory of light, relativity and quantum theory of the structure of atoms and molecules - but the basic laws of the mind is still concealed by the veil of mystery. Probably Newton artificial intelligence is not yet born.

But mathematicians and computer scientists does it bother. For them to meet at the threshold of the laboratory out thinking machine is only a matter of time.

We can call the currently influential personality in the field of AI. This is the great British mathematician Alan Turing - seer, who managed to lay the cornerstone of the study of this problem.

It is with Turing starts the computer revolution. He created in his imagination machine (which has since been called a Turing machine), consisting of only three elements: input, output and the Central processing unit (something like the Pentium processor)capable of strictly specified set of operations. On the basis of this representation Turing established the laws of computers and precisely define their expected power and the limits of their capabilities. And today all digital computers are subject to strict laws Turing. Structure and structure of the digital world owe it to the scientist.

In addition, Turing made a great contribution into the Foundation of mathematical logic. In 1931 the Vienna mathematician Kurt g?del made in the world of mathematics real sensation; he proved that in arithmetic are true statements which cannot be proven by means of one only arithmetic. (As an example we can mention the Goldbach conjecture, expressed in 1742, and that of any even integer greater than two can be written as the sum of two primes; the hypothesis is not proven until now, although it took two and a half centuries, and it may never be unprovable.) Revelation adequately shattering dream that lasted for two thousand years and originating from the Greeks, is a dream to prove someday all true statements in mathematics. G?del showed that there will always be true statements, proof of which are inaccessible to us. It turned out that mathematics is not a finished, perfect design, the building and what to complete the construction there will never come.

Turing also took part in this revolution. He showed that in the General case it is impossible to predict whether or not you need a Turing machine to perform various mathematical operations on a given her program finite or infinite number of steps. But if on the calculation of something it takes an infinite time, this means that what you ask a computer to calculate, calculate impossible. So Turing proved that in mathematics there are true expressions that cannot be calculated, they will always remain beyond the capabilities of your computer, no matter how powerful it may be.

During the Second world war, the pioneering work of the Turing-decoding of encoded messages saved thousands of allied soldiers and, very likely, influenced the outcome of the war. Allies, being unable to decrypt the Nazi messages encrypted using a machine called "Enigma", asked the Turing and his colleagues to build his car. As a result, the Turing did it; his car was called "Bomb". By the end of the war, there were already more than 200 vehicles. In the result, the allies for a long time was reading the secret messages of the Nazis and managed to deceive them about the time and place of the decisive invasion of the continent. Historians are still arguing about the role of Turing and his works in the planning of the invasion of Normandy - invasion, which ultimately led to the defeat of Germany. (After the war, the British government has classified work Turing; as a result, the public does not know what role he played in these events.)

Turing not only offered as a hero who helped turn the tide of world war II; no, it just made fun of before his death. One day, his house was robbed, and the scientist called the police. Unfortunately, the police found in the house of the certificate of homosexuality owner, and, instead of looking thieves arrested the Turing. The court decided to subject him injections of sex hormones. The effect was disastrous: it has increased the chest. In 1954, Turing, unable to bear the mental anguish, committed suicide - ate the Apple, stuffed with cyanide. (Rumored to be bitten Apple, which became the logo of Apple, is a tribute of respect to Turing.)

Today Turing, probably best known to the Turing test. Tired of fruitless and endless philosophical debate about whether a machine is "think" and does it have a "soul", he tried to bring into the discussion about artificial intelligence clarity and precision, and invented a particular test. He proposed to put a car and person in a separate insulated and sealed the premises, and then to ask both questions. If you are not able to distinguish the machine answers from man, we can assume that the machine has passed the Turing test.

Scientists have already written a few simple programs (for example, the program "Eliza")that can simulate speaking and keep the conversation; the computer with this program is able to fool most unsuspecting people and convince them that they are talking to a person. (Note that in the conversations of people, usually limited to a dozen topics and use only a few hundred words.) But programs that can deceive people who know about the situation and deliberately try to distinguish the car from a person still does not exist. (Turing himself assumed that by 2000, with exponential growth in computing power you can create a machine that can cheat in a five-minute test 30% of the experts.)

Some philosophers and theologians speak on this issue a United front: they believe that to create this robot is able to think as a man, it is impossible. Philosopher from the University of California, Berkeley John Searle has offered to prove this thesis "test of the Chinese room". Essentially, Searle argues that robots would be able to sometime, maybe, pass the Turing test in some form, this means nothing, because they just blindly manipulate symbols, not knowing invested in their content.

Imagine: you, not understanding a word in Chinese, sitting in an isolated case. Suppose you have a book, with which you can quickly translate from Chinese and Chinese, as well as to manipulate the signs of this language. If someone asks you a question in Chinese, you simply rearrange according to the book the strange icons and give a reliable answer; you do not understand any of the questions, no answers.

The crux of the Searle is reduced to the difference between syntax and semantics. According to Searle, the robots are able to master the syntax of the language (i.e. can learn how to correctly manipulate its grammar, formal structures and so on), but not its true semantics (i.e. the semantic meaning of the words). Robots can manipulate words, not knowing what they mean. (It's like a conversation on the telephone with voice mail when you have from time to time to press the number "1", "2" and so on, following the instructions of the machine. The voice on the other end of the wire is fully capable to react to your numbers, but it would be strange to suggest that he understands something.)

Physicist Roger Penrose of Oxford also believes that artificial intelligence is impossible; mechanical creature, capable of thinking and with human consciousness, contrary to the quantum laws. The human brain, says Penrose, so superior to all created by man in the laboratory that the experiment to create humanoid robots are simply doomed to failure. (He believes that as the theorem of incompleteness proved that the arithmetic is incomplete, so the Heisenberg uncertainty principle proves that the machine is in principle unable to think like human beings.)

However, many physicists and engineers believe that nothing in the laws of nature is not contrary to a real robot. For example, Claude Shannon, who is often called the father of information theory, once asked, "Can machines think?" He replied, "of Course." When he was asked to explain the answer, he added: "I think, isn't it?" In other words, he considered it obvious that machines can think, because people are also machine (though made of flesh and blood, not of chips and wires).

Watching the film robots, you might think that the creation and development of complex robots with artificial intelligence is a matter of the nearest future. Actually it is not so. If you see that robot acts as a man, it usually means that the business is dirty, it's a hocus-pocus, say, in the corner sits a man, and speaking for the robot as Goodwin in the Magic country. In fact, even the most complex of our robots, such as the Martian robotic Rovers have at best intelligence insect. Experimental robots famous MIT artificial intelligence Laboratory struggle to cope with the tasks available even cockroaches: for example, to move freely around the room, filled with furniture, to hide or to recognize the danger. No robot on Earth are not able to understand simple children's fairy tale, which he read.

The plot of the movie "2001: a space Odyssey" is based on the wrong assumption that by 2001, we will have swarmrobot HAL able to fly the spacecraft to Jupiter, a casual chat with members of the crew, to solve problems and, in General, act almost like human beings.
The top-down approach

The attempts of scientists around the world to create a robot met with at least two serious problems that are not allowed to measurable progress in this direction: it is the recognition and common sense. Robots can see so much better than us, but do not understand what he saw. Robots can hear much better than us, but do not understand what is heard.

In order to approach the solution of this dual problem, researchers have tried to apply the approach to artificial intelligence, known as "top-down" (sometimes called the formalist school, or "the good old AI"), the purpose of the scientists, roughly speaking, was to program all the rules and laws of pattern recognition and common sense and record these programs on one CD-ROM. They believe that any computer into which you insert the disc, instantly aware of yourself and be reasonable, not a worse person. In 50-60s of the XX century in this direction were achieved enormous success; appeared robots, able to play checkers and chess, solving algebraic problems, up from the floor bricks etc. Progress has such a strong impression that sounded even prophecy that in a few years, robots on intelligence will surpass people.

For example, in 1969 sensation robot Cervix, founded at the Stanford research Institute. The robot, this was a small computer PDP with the camera on top mounted on a wheeled cart. Luggage "looked", the computer analyzed and recognized in the room objects, and then tried to take the trolley route, without hitting anything. Cervical the first of mechanical machines learned to move in the "real world"; journalists then argued heatedly, when the robots will overtake people in development.

But soon became apparent drawbacks of such robots. Approach to artificial intelligence, known as "top-down", led to the creation of cumbersome awkward robots, who needed a few hours to learn how to navigate in a special room, where there was only objects with straight sides (rectangles and triangles). It put in a room furniture irregular, and the robot was no longer able to recognize it. (Funny, but a fruit fly, the brain which contains only about 250 000 neurons and computing power not hold a candle to any robot, without any difficulty focused and moving in three dimensions and takes the aerobatics; meanwhile clumsy noisy robots manage to get lost in two dimensions.)

Soon the top-down approach as if lied on the brick wall: progress stopped. Steve Grand, Director of the Institute of kibirizi, says that such approaches "we had 50 years to prove their worth, and they fell short of expectations".

In the 1960s, scientists did not understand what a huge work needs to be done in order to program the robot to perform even the simplest of tasks, such as recognition of keys, shoes and tea cups. As said Rodney Brooks from MIT, "40 years ago, the artificial intelligence Lab at MIT gave this task to the student as a summer job. The student failed - as I do in his doctoral dissertation, 1981". Generally speaking, artificial-intelligence researchers still cannot solve the problem.

Consider the example. Entering the room, we instantly recognize the floor, chairs, furniture, tables, etc. While the robot, looking round the room, sees her only as a set of lines, straight and curved, which he translates in pixels of the image. And requires enormous computing power to remove from the mess of lines any sense. Us enough of a second, to see the table, but the computer sees in place of the table only a set of circles, ovals, spirals, straight and curved lines, angles, etc. Can be, having spent an enormous amount of computer time, the robot in the end, and recognizes in this object table. But if you rotate the image, it will have to start all over again. In other words, the robot is able to see, and much better than the man, but he is not able to understand what he saw. Upon entering the room, the robot will only see a jumble of direct and curves, instead of chairs, tables and lamps.

When we enter the room, our brain unconsciously recognize objects, producing many trillions of operations, " the lesson that we, fortunately, just do not notice. The reason that a significant part of the action of the brain is hidden even from ourselves - evolution. Imagine the person in a dark forest was attacked by saber-toothed tiger; if he consciously make the actions needed to identify hazards and find ways to salvation, he will not manage to move. To survive we need to know one thing - how to escape. When we lived in the jungle, we just didn't need to know all incoming and outgoing signals, with whom it has a brain in recognition of earth, sky, trees, rocks, etc.,

In other words, the actions of our brains resemble a huge iceberg. What we are aware of only the tip of the iceberg, consciousness. But beneath the visible surface, hidden from the eye, there is much more voluminous the subconscious mind; it uses a huge amount of "computing power" of the brain so that we were constantly aware of the simple things: where are we who speak what's around. All these actions brain does automatically, without asking our permission and without reporting about them; we just do not notice this work.

That is why robots can't duck in the bathroom, to read handwriting, driving in cars, collect garbage, etc., In a futile attempt to create mechanical soldiers and smart trucks, the us military has spent hundreds of millions of dollars.

Only after that, scientists began to understand that the game of chess or multiplication huge numbers involves only a tiny fraction of the human mind. Victory in 1997, the computer Deep Blue IBM over the world chess champion Garry Kasparov became the victory of pure computer, i.e. the computer, power; however, despite the headlines of the Newspapers, this experiment did not tell us anything new nor the mind, nor the consciousness. Douglas Hofstadter, a computer scientist from Indiana University, said about this: "my God, I thought to play chess need to think. Now I understand that it is not necessary. This does not mean that Kasparov is not able to think deeply; it means only that in the game of chess you can do without deep thoughts, just as you can fly, not flapping its wings".

(The development of computers in the future will have a significant impact on the labour market. Futurologists sometimes claim that in a few decades without work does not remain only highly qualified specialists in design, production and maintenance of computers. In fact it is not. These workers, as scavengers, engineers, firefighters, police officers, etc. that will also get future without work, because their work involves a pattern recognition problem. Each crime, each piece of trash, each instrument and fire are different from the rest; robots with such work will not cope. Ironically employees with special education, such as ordinary accountants, brokers and cashiers in the future can really lose your job because of their work consists almost entirely of repetitive actions and includes working with numbers, and we already know what computers do best.)

Second - after the pattern recognition problem faced by attempts to create robots, even more fundamental. This lack of robots so-called "common sense", For example, everyone knows that:

• Water is wet.

• Mother always older daughter.

• Dogs don't like pain.

• After the death of no return.

• Rope can pull, but can't push.

• Stick can push, but you can't draw.

• Time cannot go backwards.

But there is no such calculation, such mathematics, which would Express the sense of these words. We know all this, because he saw in the life of animals, water and rope themselves had thought up these truths. Children learn common sense in error, it is inevitable clashes with reality. Empirical laws of biology and physics are also on the experience in the process of interaction with the environment. But robots have no experience of this kind. They only know that built them programmers.

(In the future no one will take away a person's profession, which require common sense, i.e., the field of activity associated with creativity, originality, talent, humor, entertainment, analysis and leadership. It is these qualities that make us unique, it is so difficult to reproduce in the computer. They make us human.)

In the past mathematics repeatedly tried to build a magic program, which concentrates in itself once and for all the laws of common sense. The most ambitious project of this kind - CYC (short for "encyclopedia"), the brainchild of Douglas Lanata, the head of the company Spohr. Just as in the result of implementation of the Manhattan project - the program cost $ 2 billion. - was created the atomic bomb, the CYC project was to be a "Manhattan project" artificial intelligence, the latest push, which was to appear genuine artificial intelligence.

Not surprisingly, the motto of Lenta sounds GAK: "the Mind is ten million. (Lenat invented a new way of discovering the laws of common sense; its staff carefully combing the page scandalous and sensational Newspapers, then ask CYC be found in the articles of the error. In fact, if Lento will be able to finally achieve this, CYC has a gun most readers tabloids!)

One of the tasks of the CYC project is to reach "the point of equality", i.e. the point where the robot will understand enough to yourself to digest the new information and download it directly from magazines and Newspapers that can be found in any library. At this point, CYC, as the chick that have left the nest, will be able to spread their wings and become independent .

Unfortunately, since the firm was founded in 1984, its reputation has suffered greatly from the common AI problems: its representatives make loud, but completely unrealistic predictions that only attract newsmen. In particular, Lent predicted that in ten years - 1994 - in the "brains" CYC will contain from 30 to 50% of "well-known reality." But today CYC and comes close to this indicator. As scientists have found Corporation, you must write millions of lines of code that the computer could even approach the level of common sense four year old child. While the program CYC contains miserable 47 000 concepts and 306 000 facts. Despite consistently optimistic press releases of the Corporation, the newspaper quoted one of the employees of Lenta RV Guha, who left the team in 1994: "CYC is usually considered a failure... We worked hard like hell trying to create a pale shadow of what was originally promised."

In other words, attempts to program all the laws of common sense and drive them into one computer failed simply because of common sense too many laws. Man makes them without effort - after all, he's from birth constantly faced with reality, gradually absorbing the laws of physics and biology. With robots all different.

"The company founder Microsoft bill gates admits: "it was much more difficult than anticipated to teach computers and robots to perceive the reality and respond to it quickly and accurately... for example, to navigate in the room toward its subjects, to respond to sound and to understand it, to take different in size, material and fragile items. The robot is very hard to do even something as simple as to distinguish the open door to the window".

However, supporters of the top-down approach indicate that progress in this area, although not so fast as we would like, yet there. In laboratories throughout the world are overcome all new frontiers. For example, several years ago, the DARPA, which often takes on financing the most advanced technical projects, announced a prize of $ 2 million. for creation of the auto transport means самостоятельноj without driver, to overcome the highly rugged terrain Mojave desert. In 2004, none of the participants in is not able to pass the route. Best machine was able to go to 11.9 km, then went out. But already in 2005 machine without driver, presented by Stanford Racing Team, has successfully overcome the difficult route 212 km, although it took this for seven hours. Besides the winner to finish the race came another four cars. [However, critics note that the rules allow machines to use satellite navigation systems on the long journey in the wilderness. As a result the car rides on a pre-selected route without special complications; this means that it is not necessary to recognize in the way of complex images obstacles. In real life, the driver must consider many unpredictable factors: the movement of other vehicles, pedestrians, repair works, traffic congestion, etc)

Bill gates with cautious optimism said that robotic machines can become "the next big leap". He compares today's robotics with personal computers, which he did 30 years ago. It may well be that robots today, as personal computers are then ready to rapid start. "No one can definitely say that this industry will gaining critical mass, " he writes. "But if this happens, robots may change the world".

(The market humanoid robots, if they ever will appear and will become commercially available, will be enormous. Although today these robots no, robots with a rigid program is not onlythere are about, but quickly spread. According to the International Federation of robotics, in 2004 there were about 2 million of these robots, and by 2008 will see another 7 million Japanese Association of robots predicts that if today the turnover of the industry involved in the production of personal robots, is 5 billion dollars. in the year by 2025, it will reach 50 billion.)
The bottom-up approach

The limitations of the top-down approach to creating artificial intelligence is obvious, therefore, from the outset, researchers are exploring another approach is "bottom-up". The essence of this approach is that, in imitation of evolution, to force the robot to learn from experience, as he was learning the baby. After all, insects, say, guided the movement is not the fact that scan a picture of the surrounding world, break it up into trillions of pixels and processed the image using supercomputers. No, the brain insect consists of "neural networks" - self-training machines that slowly, bumping into obstacles, master the art of correctly to move in a hostile world. It is known that at MIT with great difficulty, managed to create walking robots by the method of top-down. But simple mechanical beings like Zhukov, accumulate experience and information through trial and error (i.e. uticati obstacles), after a few minutes start successfully running around the room.

Rodney Brooks, Director of the famous Laboratory of artificial intelligence MGT, famous for its large and clumsy walking robots type of top-down, turned himself into a heretic, when I started to explore the idea of a tiny "like" robots that learn to walk in the old proven method: stumbling, falling, stumbling on all sorts of things. Instead of using complex computer programs and mathematically calculated when walking the exact position of each foot in each moment of time, his "nascosti" act by trial and error and cost little processing power. Today the descendants of tiny robots Brooks collect on Mars data for NASA; they overcome kilometers sad Martian landscapes of their own understanding. Brooks believes that nascosti ideal for exploration of the Solar system.

One of the new projects Brooks became a COG is an attempt to create a mechanical robot with the mind of a six-month baby. Externally, the robot is a jumble of wires, electrical circuits and drives, but equipped with a head, eyes and hands. There is no program that defines any laws of the mind. Instead, the robot has learned to focus my eyes and watch the man-coach; who is trying to learn simple skills. (One of the women who become pregnant are betting on who will make great progress by the age of two years: COG or her future child. A child is much ahead of "the contender".)

Despite the successful imitation of the behavior of insects, robots with neural networks look pretty pathetic, when the creators are trying to get them to emulate the behavior of higher organisms, such as mammals. The most advanced robot with neural networks are able to walk around the room or swim in water, but can't jump and hunt like a dog in the woods, or explore the room, like a rat. Big robots on neural networks contain dozens, most of the hundreds of "neurons", while the human brain has about 100 billion neurons. The nervous system is very simple worm Caenorhabditis elegans, a fully studied by biologists and plotted on a map, consists of 300 small neurons; this is probably one of the simplest nervous systems in nature. But in this system between neurons observed over 7000 links-synapses. No matter how primitive C. elegans, his nervous system is so complex that no one has been able to create a computer model of such a brain. (In 1988 one computer expert has predicted that by now we would be robots with approximately 100 million of artificial neurons. In fact, the neural network of one hundred neurons is considered outstanding.)

The irony is that the machines are constantly carry out tasks that people seem difficult, say peremeshayte large number or playing chess, but get stuck on a completely "simple" for a person jobs, like to walk around the room, to know someone in the face or gossip with a friend. The reason is that even the most advanced our computers basically just complicated to limit counting machine. And our brain evolution has shaped so that he could solve the global problem of survival. This requires sophisticated and well-organized structure of thinking, which includes common sense and image recognition. Complex calculations or chess is not necessary for survival in the woods-but there can not do without the ability to run away from a predator, to find a mate and to adapt to changing conditions.

That's how he summarized the AI problem Marvin Minsky of MIT, one of the founders of the science of artificial intelligence: "the History of AI in something funny because the first actual achievements in this area were beautiful machine capable of logical proofs and complex calculations. But then we wanted to make a car that was able to answer questions on common stories, which can be found in the book for first graders. At present there are no machines, capable of it".

Some scientists believe that someday the two approaches, top-down and bottom-up - all come together, and the merger may be the key to creating this artificial intelligence and human-like robots. In the end, when the child learns that he uses both methods: first, a little man relies mainly on the method of " bottom-up" - he stumbles upon the subject, examining them, taste them and so on; but then he starts to receive verbal lessons from parents and teachers, books - at this point, the time comes for top-down approach. Even as adults, we constantly mix the two approaches. For example, cook reading the recipe, but did not forget to try the dish preparing.

Hans Moravek says, "Completely reasonable machines will appear not earlier, than it will be full of Golden crutch that will connect both ways". He believes that this will happen, probably in the next 40 years.
Emotional robots?

One of the recurring themes in literature and art has long become a mechanical creature, who dreams of becoming a man, his human emotions. This creature was not satisfied that are collected from the wires and steel; it wants to laugh, cry and feel all emotional joy of human beings.

A typical example is a puppet Pinocchio, who dreamed of becoming a real boy. Tin man wanted to get his heart. And data, the robot from Star trek, wants to be a man, although superior to any person in power and intelligence.

Some even speculate that our emotions are a higher unique property, and that they make us human. Proponents of this view argue that no machine will never be able to gasp with delight at the sight of the magnificent sunset or to laugh a good joke. And some say that the machine will never experience the emotions, because they, emotions, represent the Acme of human development.

But the scientists who are working on the creation of artificial intelligence and trying to unravel the physics of emotions, paint a different picture. For them the emotions not only the essence of the human, but on the contrary by - product of evolution. Simply put, emotions useful for us. They helped us to survive in the forest and today also help to overcome adversity and to navigate among life's dangers.

For example, it is very important in evolutionary terms, the concept of "like" - because most of the things in the world we are in danger. Of the millions of objects faced daily people, only a few are able to benefit him. So when us something like, this means that we million allocated from dangerous and useless things that their tiny share, which may be useful for us.

Similarly, jealousy is an important sense, because success in procreation ensures the transfer of our genes to future generations. (This is why sex and love so many associated emotionally charged feelings.)

Shame and remorse are important because they help us to master the skills of socialization, necessary for life in society. If we fail sometimes to apologize, sooner or later, we shall cast out of the tribe, seriously thereby decreasing our chances of survival and gene transfer.

Loneliness is also important. At first it seems that this sense of unnecessary and redundant because the man is able to live alone. But the desire for other people too important for survival because people always depend on the resources of the tribe as a whole.

In other words, in the course of further development of robots could also have emotions. Maybe the programmers will lay them in an emotional connection with the hosts to robots don't end up in a landfill. Such emotions would help them enter into our society, to become not rivals, and reliable assistants owners.

Expert computers Hans Moravek believes that robots will be programmed to emotions such as fear; this is necessary for self-preservation. For example, if a robot ends battery, it will Express the excitement or even panic so that people could understand him. He will go to the neighbors and ask for permission to use the socket, saying: "Please! Please! I need it! It is so important to me and so little worth it! We'll pay you!""

Among other things, emotions are important when making decisions. People who have suffered some brain damage, lose the ability to feel emotions. Intelligentvirtual ability remains with them, but to Express feelings they fail. A neurologist Antonio Damasio of the Medical College of the University of Iowa, specially studied people with this kind of brain injuries, says they know, but not feel."

Dr. Damasio argues that such people often experience difficulties with the adoption of even the most minor decisions. They can't be guided by emotions, and therefore infinitely sorted and considering options; the result is disastrous indecision. One of the patients of Dr. Damasio for half an hour chose the date of the next visit.

Scientists believe that emotions are processed in the limbic system of the brain, located deep in the center. If a person violates the relationship between the new cortex (which manages rational thinking) and the limbic system, his mind is still with him, but lost emotions that he could be guided in making decisions. Sometimes we "overshadows"we "know in our gut"how to act. People who have broken the link between rational and emotional parts of the brain, no such ability.

For example, in the store, we unconsciously produced thousands of evaluation and decision; we estimate almost everything that we see: "This thing is too expensive, too cheap, too colorful, too stupid, but this is just what you need". For a man with such a brain injury trip to the store for shopping can be a nightmare, because all things will seem to him equally good or, if you will, equally bad.

As the robots are becoming smarter and will begin to make their own decisions, they are also likely to become the victim of a fatal hesitation. (Remember the story about the donkey that died of hunger between two haystacks - he could not decide to which haystack to go.) To solve this problem, the robots of the future, probably in the brain appears the emotional path. Speaking about the lack of emotion in robots, Dr. Rosalind Picard of the Media lab at MIT observes: "They don't feel that is most important. This is one of the main deficiencies. Computers just don't understand this".

As written Russian novelist Fyodor Dostoevsky, "if everyone on Earth was reasonable, nothing would have happened".

In other words, emotions may require the robots of the future, to set goals, to give their life meaning and structure; otherwise endless possibilities will deprive them of all the ability to take action.
Do they have any conscience?

On the question of whether machines to possess consciousness, there is no consensus; moreover, it is not the question of what consciousness at all. No one has yet managed to formulate a universally agreed definition of consciousness.

Marvin Minsky believes that consciousness is a rather "the totality of consciousness"; meaning that the thinking process in the brain is not localized and distributed, and in each moment of time for the championship in it compete several different centers. In this case, the consciousness can be seen as a sequence of thoughts and images coming from different "minds" of the lower level, and they are all competing with each other and try to capture our attention.

If so, it is quite possible that the notion "consciousness" is somewhat swollen; perhaps too many scientific works devoted to the subject, blurred generations of philosophers and psychologists. Possibly determine consciousness is not too difficult. Sidney Brenner from the Salk Institute, La JOLLA says: "I predict that by 2020 - a trait which is quite distinct - consciousness as a scientific problem will cease to exist,. Our successors will be amazed by the number of scientific stuff, which today are seriously discussed, of course, if they have the patience to dig into the electronic archives of old magazines".

Marvin Minsky believes that research in AI suffer from envy to physics". In physics Holy Grail, who wants to get any scientist, is a simple equation, which would unite all physical interactions of the universe in the framework of a unified theory - the "theory of everything". Under the influence of this idea artificial-intelligence researchers are also trying to find a common paradigm that would explain consciousness. But Minsky believes that such paradigm may not exist,

(Members of the "constructionist" school to which they belong and I believe that instead of endless debate about whether to create a thinking machine, you need to take and try. As for consciousness, then most likely there is some space of consciousness, which is owned and primitive thermostat, which monitors the temperature in the room and a self-conscious body, in what is today the people. Animals may also have consciousness, but consciousness lower level compared with man. So, instead of endlessly discuss philosophical issues and argue about the definition of consciousness should be trying to compile a catalogue of various types and levels of consciousness and put it on the shelves. Perhaps robots will eventually acquire silicon consciousness". In General, someday, perhaps, robots will embody the very different than ours, the architecture of thinking and processing information. It is not excluded that in the future superior robots will be able to blur the distinction between syntax and semantics and their reaction will truly be indistinguishable from a human response. If this happens, the question as to "know" whether they actually question, will lose any sense. The robot, fluent in the syntax, understands - for all practical purposes, the content of the conversation. In other words, the perfect possession of the syntax and understand.)
Could robots be dangerous?

Moore's law (Noteridae that the performance of computers doubles every 18 months) suggests that in a few decades will appear robots to reason, say, a dog or cat. But it may happen that by 2020, Moore's law will cease to operate, and silicon era come to an end. The last half-century the performance of the computers has been growing at such a breathtaking annual pace due to the fact that there were more and more tiny silicon transistors, tens of millions which could easily fit on the nail. For etching these microscopic transistors on silicon plates were used ultraviolet radiation. But the process of microminiaturization cannot continue indefinitely. Over time, transistors can be reduced to the size of molecules, and it will automatically stop. After 2020, when the era of silicon, Silicon valley may turn into a new Rusty belt.

The Pentium processor in your laptop has layers in the thickness 20 atoms. By 2020, the thickness of the layer can be reduced to five atoms. At this moment will come into effect the Heisenberg uncertainty principle, and will be impossible to say for sure where the electron is. And then the chip will have outages and computer - short circuit. At this point the computer revolution and Moore's law will come to face a blank wall - because the laws of quantum mechanics to bypass impossible. (Some argue that the digital era is "victory bits on atoms". But someday, when Moore's law will cease to operate, the atoms will take his.)

Now physicists working on postcranial technology that will dominate in the world of computers after 2020, but so far the results are not too encouraging. As we have said, we consider several promising technologies, including quantum computers, computers DNA-based, optical computers, nuclear computers, etc., But in each direction, there are enormous challenges to be overcome, before the technology can try on the gown silicon chips. Technology of manipulation of individual atoms and molecules is still in its infancy, and we are not yet able to make billions of transistors, comparable in size with atoms.

But let's assume for a moment that physicists have found a way to bridge the gap between silicon chips and, say, quantum computers. Suppose also that Moore's law in one form or another continues to operate in postcranial era. In this case, artificial intelligence may indeed become a reality. Then the robots can learn human logic and emotions and to learn how to confidently pass the Turing test. Steven Spielberg researched this topic in the film "Artificial intelligence"; the film tells about how they created the first artificial boy, able to Express emotions and therefore suitable for adoption in the human family.

The question arises: can such robots be dangerous? The most likely answer: Yes, you can. Robots can be dangerous, once achieved intelligence monkeys, because the monkey has consciousness and will. Perhaps on achieving this milestone will take many decades, and scientists have time to watch the robots before they begin to pose a threat. For example, in their processors will be possible to place a special chip that will not allow them to wander off in any direction". Or you can integrate them in the mechanism of self-destruction or disable that had worked well in case of emergency.

Arthur C. Clarke wrote: "Maybe we will become for computers Pets and will, as lapdog, to lead a carefree existence, but I hope that we will always be able at any time to unplug".

A more real threat is the dependence of our infrastructure from computers. Water supply and electricity network, not to mention communication and transport, will become increasingly computerized. Our cities have turned into complex organisms, and now for thereign throughout our infrastructure and its monitoring is necessary complex and intricate computer network. In the future, to make it work, you have to enter in these computer networks, artificial intelligence. Error or failure in this inclusive of computer infrastructure can completely paralyze city, country or even the whole civilization.

Will surpass us do computers on rationality? Of course, in principle it is not prohibited by any laws of nature. If robots are self-trained neural network and if they reach a level of development that allows them to learn faster and more effectively than we learn, it is logical to assume that with time they will surpass us in reasoning. The Moravek says that post-biological world "is a world where the human race was swept by a wave of cultural changes and expelled its own artificial descendants... When this happens, our DNA will be useless, because it will lose the evolutionary race to the opponents of a new type".

Some inventors, such as ray Kurzweil, even predict that it will happen soon, sooner than it seems - perhaps in the coming decades. It is not excluded that we are now creating their evolutionary successors. Some computer scientists foresee singularity", as they call it, " when robots will be able to process information faster and faster, exponentially, and simultaneously create a new robots, until finally their collective ability to assimilate information will not grow almost indefinitely.

Therefore, some scientists in the long term offer to combine carbon and silicon technology[21], not waiting for our complete destruction. We, the people of carbon-based robot - silicon (at least currently). Maybe the solution lies in the merge us with our creations. (If we ever meet again with aliens, do not be surprised if their bodies will be part of organic and mechanical part; it is easier to withstand the rigors of space travel and successfully to live in a hostile environment.)

In the distant future robots or humanoids cyborg[22]maybe even give us immortality. Marvin Minsky adds: "What if the Sun will go out or we destroy the planet? Why not do better than we do, physicists, engineers or mathematicians? Perhaps we need to be the architects of our future. If not, our culture may disappear".

The Moravek foresee a time in the distant future, when we are able to obtain the structure of their brain and neuron for the neuron, right in the car. In a sense, this will give us immortality. This idea seems wild, but, generally speaking, does not exceed the limits of the possible. So, according to some scientists, in the future, man will immortality (in silicon form or in the form of artificial bodies with improved DNA).

So, if we manage to overcome the impasse associated with the termination of Moore's law, and will deal with the problem of common sense, thinking machines with intelligence animals and can be as smart as we are, or even smarter - can become a reality. Maybe it will happen at the end of the current century. Although discovered not all fundamental laws of artificial intelligence, progress in this area is making great strides. With this in mind, I would define robots and other thinking machine as the inability of the I class.