It can hardly have escaped anyone’s notice that artificial intelligence technology has dismally failed to live up to the high hopes held for it a decade ago – and, reports Microbytes, a University of California at Berkeley professor reckons he knows why: Professor L A Zadeh believes that problem is that the technology has been hitched to the wrong kind of logic, namely, predicate logic, which is highly intolerant of imprecision, whereas almost all common sense is based on imprecise assertions that things are assumed to be the way they usually are – grass is green (look at a cricket pitch), snow is white (unless it happens to be laced with soot) where predicate logic requires a list of exceptions that can never be exhaustive; he therefore proposes a the adoption of what he calls dispositional logic, which quantifies the usuality of an assertion in a similar manner to the way probability quantifies the frequency of a particular occurrence, and combines the concepts of probability and usuality; the McGraw-Hill newswire says that Zadeh and some collaborators in Japan and Poland are working on the idea of dispositional logic, and while he hasn’t developed an application or language using dispositional logic, he hopes his idea will be taken up by the artificial intelligentsia.
THE FAILURE OF ARTIFICIAL INTELLIGENCE
