Abstract

M ost definitions of artificial intelligence in the standard textsare overly complex for a general survey of the field, so hereis a simple one that will suffice instead: Artificial intelli-gence is the science of mimicking human mental facultiesin a computer. Pinpointing the beginning of AI research is tricky. George Boole (1815-1864) had plenty of ideas on the mathematical analysis of thought processes, and the field still retains several of his ideas. However, since he had no com-puter, my simple definition rules out Boole as AI’s founder. Just as historians on either side of the Atlantic have different opinions about who built the first programmable computer, they also diverge over AI’s origins. British historians point to Alan Turing’s 1950 article that defined what is now known as the Turing test for determining whether a computer displays intelligence.1 American historians prefer to point to the Dartmouth conference of 1956, which was explicitly billed as a study of AI and is believed to have been the source of the term “artificial intelligence.” A SPECTRUM OF INTELLIGENT BEHAVIOR A difficulty with my simple definition of AI is that it leaves the notion of intelligence rather vague. Figure 1 helps clarify the matter by showing a spectrum of intelligent behaviors based on the level of understanding involved. The lowest-level behaviors include instinctive reactions, such as withdrawing a hand from a hot object or dodging a projectile. High-level behaviors demand specialist expertise such as in the legal requirements of company takeovers or the interpretation of mass spectrograms. Researchers have developed conventional computing techniques to han-dle the low-level decision making and control needed for the low end of the spectrum. Highly effective computer systems exist for monitoring and con-trolling a variety of equipment. For example, Figure 2 shows the Advanced Step in Innovative Mobility robot