Computing Machinery and Intelligence

Computing Machinery and Intelligence is a seminal 1950 paper by the British mathematician and computer scientist Alan Turing. Published in the journal Mind, it introduced the now-famous Turing Test as a criterion for machine intelligence and systematically addressed philosophical objections to the concept of artificial intelligence. The paper is considered a foundational text in the fields of computer science, cognitive science, and the philosophy of mind, shifting the debate from abstract definitions to a concrete, operational test.

The Turing Test

The central proposal is an operational definition of intelligence, where a human interrogator engages in natural language conversations with both a human and a machine, hidden from view. If the interrogator cannot reliably tell the machine from the human, the machine is said to have passed the test. Turing framed this within the context of mid-20th-century technology, referencing early digital computers like the Manchester Mark 1 and the theoretical foundations laid by Alonzo Church and Kurt Gödel. The test deliberately avoids metaphysical questions about consciousness or mind, focusing instead on observable behavior, a pragmatic approach influenced by the behaviorism prevalent in the work of figures like B.F. Skinner.

The Imitation Game

Turing introduced his concept by first describing a parlor game, the "imitation game," involving three participants: a man, a woman, and an interrogator. The interrogator's goal is to determine which participant is which through written questions and answers. Turing then modified this game, replacing one of the human participants with a computer, thereby transforming it into a test for machine intelligence. This framing connected his ideas to earlier explorations of machine capabilities, such as those in Ada Lovelace's notes on the Analytical Engine and the logical machines of Charles Babbage.

Arguments Against Machine Intelligence

The paper is structured as a dialogue, anticipating and rebutting nine major objections to the possibility of thinking machines. These range from theological arguments, suggesting intelligence is a gift from a divine soul, to mathematical objections based on the limitations described in Gödel's incompleteness theorems. Turing also addressed the "head in the sand" objection, which he dismissed as an emotional preference to believe humans are unique, and the argument from various disabilities, such as a machine's inability to enjoy Strawberries or make a mistake, which he argued were not essential to intelligence.

Learning Machines

A pivotal section argues that instead of programming a machine with all possible knowledge, one should construct a relatively simple "child machine" capable of learning from experience. Turing proposed that such a machine could be educated through a process akin to that used in a child, utilizing rewards, punishments, and principles of associationism. He envisioned this as a more promising path to intelligence, an idea that foreshadowed the development of machine learning and neural networks, fields later advanced by researchers at institutions like the Massachusetts Institute of Technology and the Stanford Research Institute.

Digital Computers and Intelligence

Turing provided a concise description of the nature of digital computers, explaining their operation through the manipulation of discrete states according to stored instructions. He emphasized their universality, a concept formalized as the Church–Turing thesis, which states that any computable function can be computed by such a machine. This universality, he argued, is what makes them suitable vessels for potentially hosting intelligence, distinguishing them from single-purpose machines like the Pilot ACE or mere calculators.

Objections to the Turing Test

While systematically countering objections, Turing also acknowledged critiques of his own test. He considered but rejected the idea that extrasensory perception, such as telepathy, could invalidate the results. More enduringly, the paper planted the seeds for future philosophical debates, including John Searle's Chinese room argument, which challenges the test's sufficiency, and concerns about simulation versus genuine understanding, later explored by thinkers like Hubert Dreyfus and institutions like the University of California, Berkeley. The test's focus on symbolic manipulation also set the stage for the symbolic AI paradigm championed by early AI labs like the one at Carnegie Mellon University. Category:Philosophy of artificial intelligence Category:Computer science papers Category:Works by Alan Turing