Artificial Intelligence: Arrogance or Ignorance?

W ojciech Cellary's remarks on decision making and intelligence in Com-puter's September 2003 issue (" The Profession's Role in the Global Information Society , " pp. 124, 122-123) brought to mind Bob Colwell's " Engineering Decisions " column (Computer, Aug. 2003, pp. 9-11) and my disappointment with Adrian Hopgood's article, " Artificial Intelligence: Hype or Real-ity? " (Computer, May 2003, pp. 24-28). Shortly afterward, under the Artificial Intelligence heading in the book review section of the May 2003 issue of American Scientist, I found descriptions of two books on the chess-playing computer, Deep Blue, as well as a book on the " sociable " robot, Kismet. When the September 2003 special issue of Scientific American titled " Better Brains: How Neuroscience Will Enhance You " arrived the same day, I felt pressed to move in a predestinate groove. Although I previously had resolved not to revisit the topic I covered in " The Myth of the Intelligent Computer " (Computer, July 1997, p. 8), the omens seemed to dictate an about-face. The term artificial intelligence suffers from many problems beyond the ugliness of its initialism. Hopgood found their nub with his observation " If AI were named 'nifty computer programs,' it would surely be hailed an unqualified success, " but treated it as a mere throwaway line. The term's adoption specifically implies that we see ourselves as creating in machines an intelligence roughly equivalent to natural intelligence. The many aspects of natural intelligence that dictionaries define boil down to its being an essential element of successful behavior, especially social behavior. As computing professionals, we have a moral duty to maintain and promote a distinction between the machines we use as tools and the people whose purposes and well-being we support. People behave or misbehave. Machines function or malfunction. Functioning does not exhibit intelligence. Any inferred intelligence should be credited to the machine's designers. Society judges our competence as computing professionals by the claims we make for ourselves. If we claim to be creating intelligence, people will assume we claim also to understand what intelligence is. Yet the evidence shows that we don't understand it. This last point affects three groups of people differently. The computing profession in general, unsuspecting of its self-deceit, will have unrealistic expectations and transmit them to the other groups. People who have not thought much about intelligence will be led into unreasonable hopes and fears. People …