The Turing test is often discussed without reference to the fact that it is not a test but a definition of artificial intelligence.
Before explaining this statement, allow me to sketch the background of the subject.
Thirty years ago, computers grew so fast and became so powerful that "artificial intelligence" chairs were created in the best universities and it was feared that computers would take over. Nowadays, computers are much more powerful and much more portable, but humans still seem to control them.
The idea of taking over computers was always absurd. A computer takes into account the data provided by human beings, executes a program of instructions written by humans and transmits the output data to its human operator, who can activate and deactivate them at any time. The output data can be used for a variety of purposes, including robot control as in the automotive industry. But we are far from a robot that plays tennis and can beat Djokowic, Nadal, Andy Murray or Federer. The only type of robot that would probably approach him would be one with a pseudo-biological construction that imitates muscle and bone. Such a robot would be an intelligent device, but far from being equivalent to a human clone playing tennis. Such a clone is a distant possibility but it would not be a human design computer.
But even thirty years ago, the subject was not new. The pioneer of computer science, Alan Turing, had analyzed the question "Can computers think" in the 1940s and suggested a test to answer it, what is now called the Turing test? ? Essentially, a human interrogator was sitting alone in a room with a keyboard on which he could ask questions. The written answers would be provided by an entity in another room and published in the interrogator. After about ten minutes of interrogation, the interrogator would declare the human or artificial entity. If the entity had been declared human but in fact was artificial, it would have passed the test.
We could design a refinement of the test by replacing the questions with movements in a chess game. Nowadays, the artificial player almost always defeats any human opponent in chess, but that does not mean that the computer thinks about its actions as a great master does. It is simply a matter of following the instructions of a very long program designed by man. The grand master has the understanding of the whole game; The computer calculates the best opportunities for your next move. It is a simple combat partner for the human, not a substitute for live game that is more popular than ever. The fact that a computer can beat a human does not destroy the appeal of chess, rather than the fact that a cheetah can defeat a man destroys the attractiveness of athletics. Neither computers nor cheetahs take over.
The philosopher John Searle gave an example in the 1980s in which the test consisted of translating Chinese messages into English. It was conceivable that an artificial device could be built to do it perfectly according to a routine programmed step by step. But as Searle pointed out, that would not prove that the computer could think.
When a computer "passes the test", we should ask ourselves: "What did he really do?" And the answer is that the computer simply performed a prescribed task also or better than the human being. It is a mistake to conclude that the Turing test showed that the computer can think. The test is simply a way of deciding whether or not to think about the task at hand. If there seems to be a reflection, we can say, by definition, that the computer shows artificial intelligence.
In summary, the Turing test should be seen as a way to determine if the computer shows "artificial intelligence," not to decide if it really thinks.
And the display of "artificial intelligence" by a computer does not prove that it has fundamental beliefs or self-awareness. In fact, as Bertrand Russell argued, we can not prove that our partners have a spirit like us. We are forced to believe it without proof. But there is no reason to believe the same about computers. (See next note in 'Creeds').
I hope that one day public figures (scientists, broadcasters, presenters, religious, commentators) will see it as a duty to tell us clearly and briefly "where they come from", that is, to publish their beliefs or "My Creed in a few words" (acronym) mycian ").
No comments:
Post a Comment