Computers And Society

David D. Thornburg, Associate Editor

Expert Systems And The Mass Market Micro

A few months back I raised the controversial notion that, contrary to the views of the pundits, the home computer industry hadn't died. In fact (and here's where the criticism came), I argued that the true "home" computer market had yet to be born.

My point was that personal computers are unlikely to penetrate much more than 10 percent of American homes until they become as easy to use as much of the other technology we find in our homes. I agree it is silly to expect a personal computer to be as easy to use as a clothes dryer. On the other hand, why should it be any harder to use than a video disc player, especially when the video player contains more complex technology than is found in most personal computers?

Before home computers become commonplace, we must also create home software applications that will allow the computer to advance beyond the stage of being a tool for writing or record-keeping. Imagine sitting at your computer in the middle of the night and engaging in the following dialog with your home computer:

What seems to be the problem?

My child just woke up and says he itches. He has red splotches on his face.

Red splotches can indicate many things. Please call your child's doctor now. If the doctor has to call you back, please continue to answer my questions while you are waiting for the call to be returned. Are the red splotches located on other areas of the body as well?

Yes, there are some small ones on the chest.

Did your child eat any of the following foods in the last eight hours: oranges, strawberries, chocolate?

Yes, we had ham with an orange peel sauce for dinner.

Has your child ever displayed an allergic reaction to orange juice?

[And so on.]

Applying Artificial Intelligence

This hypothetical interaction could help save a family member's life, or at least reduce discomfort. Such programs—which can help solve problems ranging from diagnosing an illness to selecting the correct wine for dinner—are called expert systems.

Expert systems are one of the current commercial applications of research in the field of artificial intelligence. Up to this point, most expert systems have been run on fairly large computer systems, and they have been applied to massive computational tasks such as choosing the correct location for offshore oil wells.

Creating an expert system requires close interaction with human experts who are able to express their own decision-making process in terms of rules. Each rule is generally expressed in the form: IF (conditions are true) THEN (result is likely). Many of the more sophisticated expert systems have ways of dealing with imprecise information—assigning likelihoods to various results depending on the certainty with which the conditions are known.

The expert system program contains this set of rules (which can range in number to well over 100). The program also contains another part called an inference engine. The inference engine decides which rule to apply to the various information that has been entered, knows when to ask for more information, and infers a result. Once an expert system has drawn a conclusion, the user can usually enter a command such as WHY?, and the system outlines the various rules and information it used to get its result. Mathematically speaking, such programs can prove themselves.

Ferns Versus Roses

Anyone can probably list several applications for expert systems: home medicine, car problems, plant choices for the garden, choosing the right stereo, and even picking the right computer!

As home computers acquire more memory and disk storage, expert systems will become commonplace. Artificial intelligence languages such as PROLOG and LISP are now being readied for the personal computers many of us already have, so there is no technological reason that expert systems won't become a reality in our country in the next two years.

Consider that in Britain, PRO-LOG is available on the inexpensive Sinclair Spectrum + for under $40. According to Sinclair, this language is selling well, and is being taught to school children who are using it to build their own expert systems. The current interest in teaching database skills in our country is a refreshing step in the right direction.

The birth of the true consumer market for computers (a market in which computers will become as commonplace as televisions) will come very soon. We who have used personal computers since the 1970s and early 1980s will be fondly remembered as the pioneers of the true information revolution.

And, to the extent that we create useful applications for these machines, we may become shapers of the revolution as well.