The Turing Test
Overview of Turing’s Theory
The Computer Encyclopedia defines the Turing test as:
The “acid test” of true artificial intelligence, as defined by the English scientist Alan Turing. In the 1940s, he said “a machine has artificial intelligence when there is no discernible difference between the conversation generated by the machine and that of an intelligent person” (2011).
Alan Turing referred to his test as “The Imitation Game” in his famous paper, where a human observer is supposed to differentiate between a human being and a computer based on answers to typewritten questions (Turing, 1950). He then basically refuted the arguments of his time which asserted that machines cannot develop the ability to think – the following objections: theological, “The Heads in the Sand,” mathematical, “Lady Lovelace’s”, consciousness, “arguments from various disabilities,” “arguments from informality of behavior,” evidence for Extrasensory Perception, and “arguments from continuity in the nervous system.”
Abramson considers the question of the possibility of a machine being able to be built that passes the Turing test to be partly a philosophical question of the mind as well as the philosophy of computer science (2011, pg. 203–204, 217).
A lot of the dispute over the possibility of true artificial intelligence (AI) centers around the definition of an algorithm (Hoffman, 2010, pg. 205–206). Turing refers to a learning machine, starting with a child-like “mind” to achieve AI or a thinking machine (Turing, 1950). Hoffman argues that the definition of algorithms are too general.
An infinite class of problems for which it has been proven that there does not exist an algorithm, e.g. the decision problem of first order predicate logic. Actually, there is no empirical evidence available for the claims that humans do better than algorithms. Despite the fact, that this kind of objections to the enterprise of AI demand more from machines than from the human mind it appears also quite irrelevant for the practical aspect of AI systems whether a system is able to respond properly to an infinite number of conceivable requests with which the system will never be confronted. Since in practice always only a limited number of requests will be addressed to a system it is sufficient if the system can respond to those as desired (2010).
Another objection by Hoffman involves thinking creatively and the difference in the pattern of thought that cannot be reduced to an algorithm. Hoffman’s counter-argument is weak here, stating the fallacy with this argument is a particular rule scheme is not given governing human thought.
Diane Proudfoot asserts the importance and the still current relevance of the Turing test in evaluating AI, which has been discredited by various arguments of AI researchers (2011, pg. 951). She points out the tendency to anthropomorphize AI systems skews researcher’s logic and that Turing test is immune to such conceptualization (pg. 955). It has also been suggested that perhaps the gold standard for intelligence is not human intelligence, obsolescing the Turing test and IQ tests in favor of a “universal intelligence test” (Biever, 2011, pg. 42–45). Biever makes a good point that perhaps an IQ test would not be applicable to a machine or even an alien life form.
Completely Automatic Public Turing test to tell Computers and Humans Apart (CAPTCHA)
Since my major is web development, CAPTCHA has great significance to me, as CAPTCHA is a barrier to automated “bots” that exploit contact and comment forms as well as forums on websites. Most commonly CAPTCHA is in the form of a distorted image that makes it difficult for a computer to recognize the characters. There are other methods such as audio (speech recognition), a puzzle or math question, as well as correctly identifying pictures with their corresponding labels (Lee & Hsu, 2011, pg. 81). Lee and Hsu conducted a usability study with various methods of text distortion to determine the difficulty for actual humans to discern the characters.
Today the significance and relevance of the Turing test is under fire from many in AI research. Within the confines of human behavior mimicked by a computer, it is still very relevant and should still be considered a standard when considering the “intelligence” of a machine.
Abramson, D. (2011). Philosophy of mind is (in part) philosophy of Computer Science. Minds & Machines, 21(2), 203–219. doi:10.1007/s11023-011‑9236-0
Biever, Celeste. (2011). Ultimate IQ. New Scientist, 211(2829), 42–45.
Hoffmann, A. (2010). Can machines think? An old question reformulated. Minds & Machines, 20(2), 203–212. doi:10.1007/s11023-010‑9193-z
Lee, Y., & Hsu, C. (2011). Usability study of text-based CAPTCHAs. Displays, 32(2), 81–86. doi:10.1016/j.displa.2010.12.004
Proudfoot, D. (2011). Anthropomorphism and AI: Turingʼs much misunderstood imitation game. Artificial Intelligence, 175(5/6), 950–957. doi:10.1016/j.artint.2011.01.006
Skyttner, L. (2005). General systems theory: problems, perspectives, practice. Hackensack, NJ: World Scientific Publishing Co. Pte. Ltd.
Turing, A.M. (1950). Computing machinery and intelligence. Mind, 59, 433–460, retrieved September 22, 2011 from http://loebner.net/Prizef/TuringArticle.html
Turing test. (2011). Computer Desktop Encyclopedia, 1. Retrieved from EBSCOhost.