|Alan Mathison Turing|
Alan Mathison Turing was born June 23, 1912, in London and died June 7, 1954, at his home near Manchester. He suffered the conventional schooling of the English upper-middle class, but defeated convention by becoming a shy, eccentric but athletic Cambridge mathematician. The Second World War transformed Turing’s life by giving him a crucial role in breaking German ciphers, with particular responsibility for the Atlantic war.
Thereafter Turing led the design of electronic computers and the program of artificial intelligence. In 1950 he began another career as a mathematical biologist, but was assailed by prosecution for homosexuality. His last two years, though overshadowed by punishment and security risk status, saw vigorous and defiant work until his death by cyanide poisoning.
Turing’s paper Computing Machinery and Intelligence appeared in 1950. This, his only contribution to a philosophical journal, was to become one of the most cited. He considered the question “can a machine think” and gave an argument that broke with all previous speculation about homunculi and robots, and from all earlier discussion of mind, matter, freewill, and determinism. It was based on his own elucidation of mathematical computability, as achieved in 1936. It also reflected his unique experience with practical computation.
Turing’s computability arose from the long search for a logical basis to mathematics, in which Bertrand Russell had played a prominent part. In 1931 Gödel showed that no formal proof scheme such as Russell had envisaged could encapsulate mathematics. In 1935 Turing seized on the further outstanding question, of whether there could be a definite method for deciding whether a given proposition was susceptible to formal proof.
The question turned on finding a definition of “method,” and this Turing supplied with his “Turing machine” construction. This was mathematically equivalent to the definition of “effectively calculable” offered by the logician Alonzo Church a little earlier, but Church accepted that Turing’s argument gave it a natural and compelling rationale.
Their assertions, taken together, are referred to as the Church-Turing thesis. On accepting this thesis, it follows that there is no effective method for deciding provability. Many other mathematical questions of decidability have likewise been resolved.
Turing’s thesis was based on analyzing the actions of a human mind when following a rule, and translating it into formal actions of reading and writing. More generally, Turing’s formalism was intended to capture what could be carried out by a “purely mechanical process,” interpreting this as one that “could be carried out by a machine.”
Thus Turing found a new connection between the mind and the material world. On the one hand, he gave a new logical analysis of mental operations, but on the other hand, the criterion of “effectiveness” implied something that could be implemented physically.
As mathematics, Turing’s argument meant encoding operations on symbols by symbols, rather as Gödel encoded theorems about numbers by numbers. Turing exploited this by describing a “universal”machine, which could do the work of any Turing machine.
|mathematics of uncomputability|
This concept led directly to the modern computer in which program and data are stored and manipulated alike as symbols. In 1936 Turing had no technology for implementing this idea. He did further important work exploring the mathematics of uncomputability, which touched on the role of human intuition in mathematics.
He also discussed the foundations of mathematics with Wittgenstein. But then six years of war work brought him back to the “universal machine.” He had gained the experience of advanced electronics and hence the means of putting his idea into practice.
Turing’s central interest in computing lay in its role in investigating the nature of the mind. He described his post-war computer plan as “building a brain,” and asserted with increasing confidence that any action of the mind, including creative acts, could be described as computable operations.
Turing’s sophisticated cryptanalytic work had impressed him with the apparently limitless scope of the computable. He now discounted arguments derived from Gödel’s theorem suggesting a noncomputable aspect to the human mind. He emphasized that any computable operation could be implemented on a single universal machine: the computer. Hence, the computer could rival human intelligence.
Turing’s 1950 paper summarized these arguments for a wide readership. His underlying view assumed a physical basis for Mind, but rather than argue for this he appealed to an argument from external observation. He held that a computer exhibiting the appearance of intelligence should be credited with intelligence.
He thus avoided discussing the reality of consciousness, and sought to sidestep its traditional philosophical primacy. Instead, he illustrated his “imitation game” with a provocatively wide view of “intelligence,” and took pleasure in playing the role of a new Galileo, defying orthodox belief in the uniquely human nature of mind.
This “imitation game,” the so-called “Turing Test” for intelligence, was not the only content of this paper. He also sketched a constructive program for Artificial Intelligence research, which he saw as a combination of “top-down” methods by programming and “bottom-up” methods using networks capable of developing functions through training.
Turing saw self-modification in machines as a key analogy with human mental development. His doubts and reservations centered on the question of defining a valid line separating the mind from the external world with which it interacts.
Turing made a prophecy of progress within fifty years, which though cautiously expressed, still proved over-optimistic. Some artificial intelligence protagonists have come to see Turing’s ambitious goal as a distraction from systematic research. But many thinkers have found it vital to continue Turing’s arguments.
Lucas revived the objection from Gödel’s theorem, which Turing had dismissed. Hofstadter and Dennett then vigorously defended Turing’s view. A new argument was made by Penrose. This shares with Turing a wholly materialist viewpoint, but holds that there must be uncomputable elements in the physics of the brain, arising from the reduction process of quantum mechanics.
A late talk given by Turing indicates that he, too, considered this question, but death cut off the physical investigations he undertook in 1953 and 1954. The relationship of computability to physics, in particular to the material basis of mind, is the central question left by Turing’s work.
As a human being Alan Turing was highly willful and far from soulless, yet he sought to mechanize will and mocked the concept of soul. He was highly original and resisted social conformity, yet attempted to explain creativity as a process of learning. Truthfulness was paramount to him, yet he committed himself to state secrets and defined intelligence by imitation. The paradoxical life and death of Alan Turing continue to fascinate.
|process of learning|