Turing-Test

The Turing test was proposed in 1950 by Alan Turing to determine whether a machine has an equivalent man thinking skills. The time from the beginning of the computer science portion Artificial Intelligence coming and since legendary test helped to revive the ancient myth of the thinking machine for the computer age.

Test procedure

The impairment test involves a human questioner via a keyboard and a screen leads without sight and hearing contact with two unfamiliar interlocutors a conversation. One interlocutor is a human, the other a machine. Both try to convince the questioner assumes that they are thinking people. If the questioner can not say clearly after the intensive survey, which is accessible from both the machine, the machine has passed the Turing test, and it is assumed a man equal rank mind of the machine.

Criticism

There is a number of arguments have been put forward that view the Turing Test to be unsuitable for the detection of Intelligence:

• The simulation of an interpersonal conversation was less than intelligence and so most only a part of what is meant by ( human ) intelligence.
• A machine can be intelligent without that it can communicate like a human.
• Many people, such as small children or people with mental health problems are not constituted the Turing Test, although they were intelligent.
• People who do not cooperate with the test conditions would not pass the test. This is also conceivable for the postulated intelligent machine. The non-cooperation does not, it lack of intelligence.
• Since the machine being compared with a human and humans are very different, the test was not standardized.
• The Turing test just check on functionality, not on the presence of intentionality or consciousness. This argument has been elaborated among others by John Searle in his thought experiment of the Chinese room. Turing was aware of these issues when formulating his test, however, was of the view that this could be regarded as evidence of a consciousness. Searle rejects this, however.

Predictions and Results

Turing suggested that it would be possible by the year 2000, computer programming so that the average user have a more than 70 - percent chance of man and machine to successfully identify, after he has "spoken" five minutes with them. The fact that this prediction has not been fulfilled, many see as a testament to the underestimation of the complexity of natural intelligence.

Programs such as ELIZA test subjects over briefly appeared as a human, without being able to pass the Turing test formally. In its response strategy they only went on a seemingly their counterparts; the subjects were not aware that they might have to do with non-human interlocutors.

In October 2008, the 30 - percent mark was in an experiment at the University of Reading, in which took part in six computer programs that just missed. The best program managed to fool 25 percent of human test participants.

On September 3, 2011 AI web application Cleverbot took along with real people in a formal Turing test the technical Festival 2011 part of the Indian Institute of IIT Guwahati. The results were announced on September 4. 59.3 % of 1334 persons held Cleverbot for a man. The human opponents, however achieved 63.3%. "What quite surprising result. Even higher than even I would have expected or even dreamed of, "said the British computer scientist Rollo Carpenter during a lecture at the Technical Festival. " Yesterday's results exceeded 50%, and one might think that Cleverbot has passed the Turing test (...). "

Practical significance

For spam protection, it is necessary to distinguish automated input from those derived from humans. The CAPTCHA method for frequently used derives its name from the Turing test from (Completely Automated Public Turing test to tell Computers and Humans Apart). Human Interaction Proof ( HIP) is another name for this method.

Loebner Prize

The Loebner Prize is announced in 1991 and will be awarded to the computer program that is first an extended Turing test, which also multimedia content such as music, speech, images and video to be processed. The award is named after Hugh G. Loebner and endowed with $ 100,000 and a gold medal, a silver medal and $ 25,000 is available for the existence of the written Turing test. However, no computer program could fulfill the necessary conditions. Furthermore, a Loebner Prize is awarded annually to the computer program that comes a human conversation at the next. This is worth $ 2,000 and a bronze medal.

Cultural references

• Philip K. Dick used in his 1968 novel Do Androids Dream of Electric Sheep? (1982 filmed under the title Blade Runner ) the so-called Voight - Kampff test, a variant of the Turing test. In a not distant future (2019) where artificial humans, called replicants that physically the people of the same, an empathy test are subjected to the checks her emotional response through long surveys and to bring forth, whether they are human or replicant.
• When 2K BotPrize examine human tester in the computer game Unreal Tournament 2004 Bots on human responses. The aim is a bot program so that it is indistinguishable from a human player.
• In Ian McDonald science fiction novel River of Gods (2004; German: Cyberabad, 2012), a world is designed, are prohibited in the artificial intelligence from a more sophisticated, the people comparable level. The Turing test to examine the degree of AI is here rejected the argument that a sufficiently sophisticated intelligence could provoke their own performance in the test yourself.