A Brief History About Artificial Intelligence

In this brief history, the beginnings of artificial intelligence are traced to philosophy, fiction, and imagination. Early inventions in electronics, engineering, and many other disciplines have influenced AI. Some early milestones include work in problems solving which included basic work in learning, knowledge representation, and inference as well as demonstration programs in language understanding, translation, theorem proving, associative memory, and knowledge-based systems. The article ends with a brief examination of influential organizations and current issues facing the field. The history of AI is a history of fantasies, possibilities, demonstrations, and promise. Ever since Homer wrote of mechanical “tripods” waiting on the gods at dinner, imagined mechanical assistants have been a part of our culture. However, only in the last half century have we, the AI community, been able to build experimental machines that test hypotheses about the mechanisms of thought and intelligent behavior and thereby demonstrate mechanisms that formerly existed only as theoretical possibilities. Although achieving full-blown artificial intelligence remains in the future, we must maintain the ongoing dialogue about the implications of realizing the promise.1 Philosophers have floated the possibility of intelligent machines as a literary device to help us define what it means to be human.

René Descartes, for example, seems to have been more interested in “mechanical man” as a metaphor than as a possibility. Gottfried Wilhelm Leibniz, on the other hand, seemed to see the possibility of mechanical reasoning devices using rules of logic to settle disputes. Both Leibniz and Blaise Pascal designed calculating machines that mechanized arithmetic, which had hitherto been the province of learned men called “calculators,” but they never made the claim that the devices could think.

Robots, and artificially created beings such as the Golem in Jewish tradition and Mary Shelly’s Frankenstein, have always captured the public’s imagination, in part by playing on our fears. Mechanical animals and dolls—including a mechanical trumpeter for which Ludwig van Beethoven wrote a fanfare—were actually built from clockwork mechanisms in the eventeenth century. Although they were obviously limited in their performance and were intended more as curiosities than as demonstrations of thinking, they provided some initial credibility to mechanistic views of behavior and to the idea that such behavior need not be feared. As the industrial world became more mechanized, machinery became more sophisticated and more commonplace. But it was still essentially clockwork. Chess is quite obviously an enterprise that requires thought. It is not too surprising, then, that chess-playing machines of the eighteenth and nineteenth centuries, most notably “the Turk,” were exhibited as intelligent machines and even fooled some people into believing the machines were playing autonomously. Samuel L. Clemens (“Mark Twain”) wrote in a newspaper column, for instance, that the Turk must be a machine because it played so well! Chess was widely used as a vehicle for studying inference and representation mechanisms in the early decades of AI work. (A major milestone was reached when the Deep Blue program defeated the world chess champion, Gary Kasparov, in 1997 [McCorduck 2004].)

With early twentieth century inventions in electronics and the post–World War II rise of modern computers in Alan Turing’s laboratory in Manchester, the Moore School at Penn, Howard Aiken’s laboratory at Harvard, the IBM and Bell Laboratories, and others, possibilities have given over to demonstrations. As a result of their awesome calculating power, computers in the 1940s were frequently referred to as “giantbrains.” Although robots have always been part of the public’s perception of intelligent computers,early robotics efforts had more to do with  mechanical engineering than with intelligent control. Recently, though, robots have become powerful vehicles for testing our ideas about intelligent behavior. Moreover, giving robots enough common knowledge about everyday objects to function in a human environment has become a daunting task. It is painfully obvious, for example, when a moving robot cannot distinguish a stairwell from a shadow. Nevertheless, some of the most resounding successes of AI planning and perception methods are in NASA’s autonomous vehicles in space. DARPA’s grand challenge for autonomous vehicles was recently won by a Stanford team, with 5 of 23 vehicles completing the 131.2-mile course.2 But AI is not just about robots. It is also about understanding the nature of intelligent thought and action using computers as experimental devices. By 1944, for example, Herb Simon had laid the basis for the information-processing, symbol-manipulation theory of psychology: “Any rational decision may be viewed as a conclusion reached from certain premises…. The behavior of a rational person can be controlled, therefore, if the value and factual premises upon which he bases his decisions are specified for him.”

[youtube=http://www.youtube.com/watch?v=P9ByGQGiVMg