Computer smarts changing business.

Computer smarts changing business

Expert systems are computers that solve problems by applying simply reasoning skills to a store of knowlege. A two-year study of the use of expert systems in business reveals an explosive growth in the number of expert systems employed, and shows that companies using such systems reap tremendous gains in productivity.

In 1981, companies used only two expert systems; today about 2,000 different expert systems are used worldwide, reports one of the study's authors, computer scientist Edward Feigenbaum of Stanford University. Examples of productivity gains range from the 10-fold decrease in the time it takes to design a camera lens to a 300-fold decrease in the time it takes computer companies to organize the components of a large computer system. The time difference between walking and flying a jet also is 300-fold, Feigenbaum notes.

Time savings come, for instance, through the ability of lens designers to give the expert system a design and ask it to minimize the size of the lens. Time is money, and Feingenbaum reports that expert systems now save Digital Equipment Corp. $70 million to $100 million a year.

Furthermore, such benefits result from only the "first wave" of expert systems, using artificial intelligence technology of the 1970s and personal computers, Feigenbaum says. Far more impressive gains will come when research of the 1980s creates business applications in mainframe computers with large data bases of knowledge, he adds.

"Expert systems carry the potential for significant changes in the economy," Feigenbaum says. National wealth is tied to the nation's productivity, and the only way to increase productivity is to "work harder or work smarter," he says.

Feigenbaum, Stanford researcher H. Penny Nii and freelance writer Pamela McCorduck report the results in a book, The Rise of the Expert Economy, scheduled for publication in late September.