Alan Turing developed one of the most important evaluation systems that exist today, because thanks to his test it can be determined if an artificial intelligence acts like a human. In this sense, this article explains what what is the turing test And how does this experiment work? In the same way, the history of the creation of the test is covered, as well as the questions and games on which Turing relied to measure the responses of computers.
Concept and definition of the Turing test
Today, the concept of artificial intelligence is very popular on social networks and other Internet sites due to the increase in this type of program. In this sense, it seems a novelty that humans can communicate with machines through artificial intelligence and that these have the ability to generate almost human responses. However, this is a concept that was developed some time ago when the Turing test was created.
The Turing test, also known as Turing test, is a kind of experiment that seeks to measure the responsiveness of a computer, In this way, a human maintains conversations with a machine and with another human, without knowing who is giving the answers, that is, they do not know who the machine is. This test was developed to measure the potential of artificial intelligence in the 1950s to see if an AI has the ability to generate human responses.
What is the Turing test used for?
The objective of the Turing test is to make a measurement with which an evaluation of the behavior of a computer is made, this in order to know if said behavior resembles that of a human. The test is run using the language displayed by a human and the language used by a computer that was designed to generate responses.
In the test, the machine and the human are in separate places, while the evaluator communicates with both to do the test. The medium used for the conversation is textual, so in both cases a keyboard and a monitor are used, so that the evaluator does not know what the machine’s responses are.
Once the evaluator asks the questions, he will have to submit the answers to an evaluation, in which he seeks to determine whose answer each is, thus, if the evaluator cannot identify the origin of the answer, then it is described that the computer passed the Turing test. At the beginning of his research, Alan Turing described that the computer had to convince the evaluator in at least 5 minutes.
It is important to note that the Turing test does not measure how much knowledge a certain computer has, since they work with artificial intelligence and learn automatically. On the contrary, the test seeks to assess the ability to generate answers and whether the way in which the questions are answered is similar to the actions that a human performs. In other words, behavior is compared in order to know if the computer can act similar to a person.
What are the Turing test questions?
The main question on which the Turing test is based is ‘can machines think?’However, the verb to think is very broad and requires many mental processes. So Alan Turing decided to change that term and express the question of whether there are digital computers that perform well in the imitation game.
The Turing test process is simple, since it only requires answering some questions made by the evaluator, these questions will be on a topic that the interrogator chooses and may vary depending on the communication situation. Those involved will be a computer and an anonymous person. Both separated from each other and from the evaluator. In this way, the questions are asked through a device that has a keyboard and a monitor, so there is no visual or auditory contact.
What is the history of the Turing test?
At the University of Manchester, Alan Turing worked with a vacuum tube computer called the Manchester Mark 1. Similarly, he wrote a scientific paper which he titled ‘Computing machinery and intelligence’ or computing machinery and intelligence. Said article was published in the magazine ‘Mind’ where he exposed the experiment that today is called the Turing test.
This famous computer scientist called his experiment ‘imitation games‘ or imitation game in Spanish, where the main objective was to observe the communication between humans and machines. Thus, Turing defended that a program or algorithm that causes a computer to give responses similar to those of humans. So, the evaluator could not distinguish between who is the machine and who is the other human.
Turing established two versions of the test in the scientific paper, although there is a third that is described as the standard mode. Some authors wrote that Turing was amazed that a computer could think like humans, so he created this system to determine if it was true. The method establishes that a person must carry out a set of questions to a machine and another human, then you must analyze the responses.
The imitation game is one of the versions in which the experiment or test can be carried out, where a role-play is done to determine if the computer can imitate a situation. Turing describes a game in which there are three players ‘A’, ‘B’ and ‘C’, where A is male, B is female, and C can be either sex.
In this game, character C acts as an interrogator trying to find out the gender of the other players, so he can communicate with them only through cards. Player A must try to trick the evaluator and make it choose the wrong option, while player B must help the evaluator with his answers. Here, the tester must determine if the computer can act like a human, whether it is player A or player B.
From there, the test evolved and was mixed with the vision of other authors such as Sterret, so there are several confusions with the main theory of Alan Turing. In the game, the important thing is not to know who is the man and who is the woman, but to identify the human or machine. If the third participant, who is the interrogator, cannot be sure which is the computer, then the machine will pass the test. It is important to highlight that this test seeks to measure the way artificial intelligence acts, to see how much it resembles natural human actions and responses.
Who created the Turing test?
Alan Turing was the person who developed the Test that measures the responsiveness of artificial intelligence in a computer, which is why the test bears his last name as a hallmark. He was a British mathematician, logician, and computer scientist who created this test in the 1950s when he was working in Manchester, England. Alan Turing focused on artificial intelligence work and the world’s criteria towards it.
The first program with artificial intelligence that provided responses very similar to those of humans in interrogation situations was ELIZA, which pretended to be a therapist, this program was developed in 1966 by Joseph Weizenbaum. For its part, in 1972 PARRY was created, a artificial intelligence program that he can act like a schizophrenic person.
There is an award called loebner awards where the program with the greatest number of human characteristics is evaluated. There they evaluate the algorithm that generates the most intuitive responses and that have natural interactions with humans. The KUKI program has the world record for passing Turing tests, since it won this recognition on four different occasions, so it is described as the best artificial intelligence program with human responses.
The Turing test or test had a great impact in the scientific world, since it helped to develop many computer programs and in the advancement of artificial intelligence. However, nowadays there are many critics that call into question the real operation of this test in an algorithm as advanced in technology as those that exist today. Well, there are people who believe that Turing imitation games cannot prove intentions or the intellectual capacity of the machine.