"Innovators" - a book about people who have committed the digital revolution
Books / / December 19, 2019
Can a machine think?
When Alan Turing was thinking about designing a computer with a stored program, he drew attention to the statement made by Ada Lovelace century earlier in her final "Note" to describe the analytical engine Babbage. She claimed that the machine will not be able to think. Turing wondered if a machine could change its own program based on the information processed by it, whether it is not a form of training? Could this lead to the creation of artificial intelligence?
[…]
Turing has long been interested in how the computer could repeat the work of the human brain, and its curiosity was warmed further work on machines that unscrambled coded posts. At the beginning of 1943, when at Bletchley Park was ready of Colossus, Turing crossed the Atlantic and went to Bell Labs, located in Lower Manhattan, in consultation with the team working on the speech encryption using electronic devices (scrambler) - a technology that could encrypt and decrypt phone calls.
There he met with a colorful genius - Claude Shannon, who, as a graduate of the Massachusetts Institute of Technology, wrote a thesis in 1937, has become a classic. In it, he showed how Boolean algebra, which is a logical equations offers may be displayed using electronic circuitry. Shannon and Turing began to meet for tea and to conduct long conversations. Both were interested in the science of the brain and to understand that their work in 1937 was something common and fundamental: they showed how a machine that operates with a simple binary commands can be placed not only mathematics, but also all kinds of logic tasks. And because the logic was the basis of human thinking, the machine could reproduce human intelligence theory.
"Shannon wants to feed [the car] is not only data, but also the works of culture! - once said Turing colleagues at Bell Labs at dinner. - He wants to play her some music. " At another dinner in the dining room Bell Labs Turing broadcast their high-pitched voice, audible to all those present in the room:
No, I'm not going to design a powerful brain. I'm trying to design just a mediocre brain - such as for example the president of the American Telephone and Telegraph Company.
[…]
"It is believed that computers can only perform tasks for which they are given the command, - he explained in a talk given at the London Mathematical Society meeting in February 1947 of the year. - But you must know, that they have always been used in this way "He then discussed the possibility of new computers with stored programs which can? themselves alter table command, and continued, "They might be like the disciples who have learned from his teacher, but added much more his. I think that, when it happens, will have to admit that the car demonstrates the presence of intelligence. "
When he finished the report, the audience fell silent for a moment, stunned statement Turing. His colleagues at the National Physical Laboratory in general did not understand the obsession Turing creation of thinking machines. Director of the National Physical Laboratory of Sir Charles Darwin (grandson of the biologist, the founder of the theory of evolution) in 1947 wrote to his superiors that Turing "Wants to spread its work on the machine even further in the direction of biology, and to answer the question whether it is possible to make a machine that can learn from their experience. "
Turing bold idea that machines can ever think like humans, while aroused fierce opposition, and still is. It emerged as quite the expected religious objections and non-religious, but very emotional both in content and in tone. Neurosurgeon Sir Geoffrey Jefferson in a speech on the occasion of awarding the prestigious medal Lister in 1949, said: "I agree that the car just reasonable [as a man], we could not before she could write a sonnet or compose a concerto under the influence of their thoughts and emotions, and not because of random selection characters. " Turing answer a reporter from the London Times, it seemed to be somewhat frivolous, but thin: "The comparison may not be entirely fair, as a sonnet written by the machine, measure better than the other machine. "
Thus the foundation was laid for the second pioneering work of Turing, "Computing Machinery and Intelligence," published in the journal Mind in October 1950. In it, he described the test became known later as the Turing test. He began with a clear statement: "I propose to consider the question of whether machines can think." With passion rather peculiar student, he invented the game - and it is still playing and still debating. He offered to invest in the real sense of the question and he gave a simple operational definition of artificial intelligence: if the answer machine the question is no different from the answer that gives a person, then we will not have any reasonable grounds, that the machine is not "I think".
Turing test, which he called "The Imitation Game" is simple: the examiner directs written questions man and machine in another room, and trying to determine which of the answers belong man.
[…]
The most interesting objection, especially for our story - the objection of Ada Lovelace, who wrote in 1843:
Analytical Engine is not intended to create something really new. The machine can perform all that we are able to prescribe it. It may be followed by analysis, but can not predict any analytical dependence or the truth.
In other words, in contrast to the human mind, the mechanical device may not have the freedom or will move its own initiative. It can only do what is programmed. In his article in 1950 Turing section dedicated to this statement and called it "Lady Lovelace's objection."
Brilliant answer to this objection was the argument that in fact the machine can learn, thereby turning into thinking the actuator, which is capable of producing new ideas. "Instead of writing a program to simulate the adult mind, why not try to write a program that simulates a child's thinking? - he asks. - If you run the appropriate training process could eventually get the intelligence of an adult. " He acknowledged that the process of learning the computer is different from the child's learning process: "For example, it is impossible to provide the legs, so that he can not offer to go to collect the coal bin. Probably, it can not be the eyes... It is impossible to send the creature to school - for other children it will be a laughing stock. " Therefore, the baby car should be trained in a different way. Turing proposed a system of punishments and rewards that will encourage the car to repeat certain actions and avoid others. In the end, such a machine could develop their own ideas and explanations of a phenomenon.
But even if the machine can simulate the mind, Turing argued criticism, he would not mind at all. When a person passes the Turing test, he uses words that are associated with the real world, emotions, feelings, sensations and perceptions. A machine does not. Without such links language becomes just a game, divorced from meaning.
This objection led to the refutation of all test outlast Turing, who formulated the philosopher John Searle in his essay in 1980. He proposed a thought experiment, called the "Chinese Room", which in English speaking person who is not knowing the Chinese language, provides a complete set of rules that explain how to make any combination of Chinese hieroglyphs. It is passed a set of characters, and it is a combination of them, taking advantage of the rules, but without understanding the meaning of phrases, he had compiled. If the instructions are good enough person could convince the examiner that he really speaks Chinese. Nevertheless, he would not understand any compiled his own text, it would not contain any sense. In the terminology of Ada Lovelace: he would not have applied for the creation of something new, but simply to perform actions that he was ordered to carry out. Similarly, in imitation of the Turing machine game, no matter how well it can simulate the human mind will not understand or be aware of anything that is said. In fact, to say that the machine thinks no more sense than to say that the person next to numerous instructions, understand the Chinese language.
One response to Searle objection was the claim that, even if the person does not understand the Chinese language, the entire system as a whole, gathered in the "Chinese room" that there is a man (data processing unit), instructions for the handling of characters (the program) and files with the characters (data) may really understands Chinese tongue. There is no definitive answer. Indeed, the Turing test and the objections to it remains to this day the most discussed topic in the cognitive sciences.
For several years after Turing wrote "Computing Machinery and Intelligence," he seemed to enjoy participation in the skirmish, which itself provoked. With caustic humor, he countered the claims of those who chatted about the sonnets and sublime consciousness. In 1951, he made fun of them: "One day the ladies take their computers with them for a walk in the park and will be talking to each other:" My computer told me this morning are funny things! "As noted later his mentor Max Newman," his humorous, but brilliantly accurate analogy, using which he expounded his views made him delightful interlocutor ".
And now the fun part: together with the publishing Corpus we will present two copies of the "Innovators" to our readers. To participate in the raffle share this article in Twitter, Facebook or "VKontakte" and give a link in the comments. Do not forget to add the e-mail address! The draw will take place on September 24th: we will define the winners at random and read them in this article. Good luck!
"Innovators," Walter Isaacson
buy litres.ruBuy at amazon
Draw Results: