Physics of the Impossible: Robots

It is quite possible that I am already fed up with your articles on physical topics. However, let me once again disturb you with an offtopic and tell you about Robots. I hope that you learn a lot of new and interesting. And if you like this article, then, quite likely, you will also like these: Teleportation , Invisibility . I prepared the article during the week, so there should be no errors. But if you find something wrong in the text, please report it. Well, let's continue to develop physically.

(Almost all the information in this article is taken from the book of Michio Kaku "The Physics of the Impossible . " Thanks to him, not to me.)

As always, let's start with a quote:

One day, 30 years will not pass, we will quietly cease to be the smartest on Earth © James McAleer

The history of the development of artificial intelligence

The robot Most likely in our civilized world there is not a single person left who would never hear this word. However, few people know that actually the word “robot” was coined in 1920 by the Czech playwright Karel Čapek . In Czech, it means “hard tedious work”, and in Slovak it is simply “work”.
The idea of ​​a mechanical being has long captured the imagination of many. The iron lumberjack from the Magic Country, the robots children from Spielberg’s “Artificial Intelligence”, the killer robots from the Terminator are machines that can think and act like humans.
In the first century, Heron of Alexandria (he was credited with inventing the steam engine) made automatic machines, and one of them was able to talk, according to legend. Nine hundred years ago, Al-Jazari invented and designed automatic devices such as water clocks, various kitchen tools and musical instruments, driven by the power of water. In 1495, the great Italian artist and Renaissance scientist Leonardo da Vinci drew a diagram of a mechanical knight who could sit, move his arms, his head and open, close his jaw. Historians consider this scheme the first realistic project of a humanoid machine. And in 1927, robots became the heroes of one of the first and most expensive silent films of all time - the movie "Metropolis", filmed in Germany by director Fritz Lang.
But who is by now the most influential person in the field of AI? This man is Alan Turing . You probably do not hear this name for the first time, since all digital computers obey its laws. A little deviate from the topic:

In 1931 The Viennese mathematician Kurt Godel made a real sensation in mathematics when he proved that there are true statements in arithmetic that cannot be proved by means of arithmetic alone. For example: the Goldbach hypothesis that any even integer number more than two can be written as the sum of two primes. This hypothesis has not been proven so far. Gödel's proof in the dust broke the dream of all mathematicians from the time of the ancient Greeks: to prove all true statements in mathematics. It turned out that mathematics was completely unfinished and imperfect in construction, and that it would hardly ever be possible to complete construction. Turing also participated in this revolution, showing that if a machine needs infinite time for any calculations, it is impossible to calculate.

Tests for robots

The question of whether it is possible to create an AI smarter than a human is still open. Here are some well-known tests for robots during the passage of which they will be able to compare with us in intelligence.

• Turing test . Tired of the fruitless and endless philosophical debates about whether the machine can “think” and whether it has a “soul”, he tried to bring clarity and accuracy into the discussion about AI and came up with a concrete test. He offered to place the car and the person in separate isolated and sealed rooms, and then ask both of them questions. If you find yourself unable to distinguish a machine from a person by the answers, we can assume that the machine passed the Turing test.
• Test Chinese room . John Searle, a philosopher from the University of California at Berkeley, suggested the following test to prove the impossibility of creating a machine that could think like a human being: robots must understand not only the syntax of a language (for example, Chinese), but also its semantics. That is, they must not only blindly manipulate the symbols of the language, but also understand the meaning embedded in them.

Approach "Top down"

Attempts by scientists around the world to create robots met with at least two serious problems that do not allow significant progress in this direction: it is pattern recognition and common sense. Robots see much better than us, but do not understand what they see. Robots hear much better than us, but do not understand what they hear. To solve this problem, the researchers tried to use the top-down approach. The goal of the scientists was to program all the rules and laws of pattern recognition and common sense and record these programs on a single CD. They believe that any computer in which you insert this disc, instantly aware of themselves and become reasonable, not worse than a person.
In the 1950s and 1960s, tremendous progress was made in this direction. There were robots capable of playing checkers, solving algebraic problems, raising bricks from the floor, etc. In 1969 robot necks made a real sensation. This wheeled cart with a camera at the top and a PDP type computer could move in space by analyzing objects around.
However, shortcomings of this approach soon became apparent. Bulky clumsy robots appeared that were able to navigate in a room in which there are objects with only right angles. It's funny, but the fruit fly, the brain of which contains only 250,000 neurons and which, according to computational power, is not suitable for any robot, is oriented without any difficulty and moves in three dimensions.
Thanks to this approach, scientists began to realize that the game of chess or the multiplication of huge numbers involves only a tiny fraction of the human mind. Victory in 1997 IBM's Deep Blue computer on the world chess champion Garry Kasparov was proof that you can not even think about winning chess. This does not mean that Kasparov does not know how to think deeply; it only means that when playing chess you can do without deep thoughts.
In general, attempts to program all the laws of common sense and drive them into one computer failed because common sense has too many laws. A person masters them effortlessly - after all, from birth, he constantly faces reality, gradually absorbing the laws of physics and biology. With robots everything is different.

Bottom up approach

The essence of this approach is to imitate evolution, to force the robot to learn from its own experience, as a baby learns. After all, insects, for example, are guided by the movement not by scanning the picture of the world, breaking it into trillions of pixels and processing the resulting image. No, the insect brain consists of "neural networks" - self-learning machines that slowly, bumping into obstacles, master the art of properly moving in a hostile world. It is known that at MIT (Massachusetts Institute of Technology) it was very difficult to create a robot moving around the room using the “top down” approach, but using the “bottom up” approach, robots can rush around the room in just a few minutes.
One of the new projects of Rodney Brooks (director of the MIT AI laboratory) was COG - an attempt to create a mechanical robot with a six-month-old baby. Externally, the robot is a jumble of wires, electrical circuits and drives, but it is equipped with a head, eyes and hands (images nearby). There is no program in it that determines any laws of the mind. Instead, the robot was taught to focus the eyes and follow a human trainer who is trying to teach the robot simple skills.
')
A detailed example of this approach.

Conclusion

Some scientists believe that someday these two approaches will merge together, and such a merger may be the key to the creation of real artificial intelligence and humanoid robots. In the end, when a child learns, he uses both methods.
Will computers outnumber us? Of course, in principle this is not prohibited by any laws of nature. If robots are self-learning neural networks and if they have reached a level of development that allows them to learn faster and more efficiently than we do, then it is logical to assume that over time they will surpass us in reasoning. In the distant future, robots or human-like cyborg may even give us immortality. Marvin Minsky adds: “What if the sun goes out or we destroy the planet ourselves? Why not do better than ourselves, physicists, engineers or mathematicians? Perhaps we need to be the architects of our future. If not, our culture may disappear. ” Although not all the fundamental laws of AI are open, progress in this area is taking leaps and bounds, so by the end of this century we can expect something similar to the terminator :)

Question. Well, in order for us to have something to discuss, let me ask you this question: If you had a robot that is able to do some work quite adequately, then what would you ask to do or ask him to do in the first place? For example, I hate cooking breakfast. Lunch, dinner easy, but not breakfast! Therefore, I would ask him to make me some morning meal every day.