ABSTRACT
In this chapter, the author argues that research in Artificial Intelligence (AI) was based upon mistaken assumptions about the nature of human knowledge and understanding. He explains why he believe that even the cautious and guarded optimism which he once had with respect to certain isolated areas of AI research was unjustified and, ultimately, mistaken. The early stages of AI research were characterized by overly ambitious goals, wishful rhetoric and outlandish predictions. The basic project of AI research is to produce genuine intelligence by means of a programmed digital computer. This requires, in effect, that human knowledge and understanding be reconstructed out of bits of isolated and meaningless data and sequences of rule-governed operations. The mental processes of the novice are easily imitated by the digital computer. Since it can use more rules and consider more context-free elements in a given amount of time, the computer typically outperforms the novice.
