Computers Are Limiting Our World

Everyone agrees that computers highlight the achievements of 20th-century science and technology. With so much obvious success it is very easy to forget the limitations. Are we making serious mistakes because science is leading us in the wrong direction? I believe that we are. Because computers are designed by scientists, they reflect the scientific ethos. A “good” scientist observes a narrowly defined phenomenon and explains its behavior with simple generalizations. A “famous”scientist finds particularly elegant laws. It is therefore not surprising that computers are designed to work at their best when processing a set of precise rules (laws), using well-ordered data (observations). So what are computers bad at? Handling poorly structured and open-ended tasks with no clear general rules. I wonder why? How many famous scientists can you name who became famous for unsuccessfully trying to solve such a problem? Computers are bad at doing the very things that the well-educated scientist is trained to avoid if he wants to have a successful career. Rather surprisingly, commercial data processing produces a similar effect. Large companies (and these are the ones that matter in economic terms) use computers to regiment their clerical procedures with an almost fascist precision. A neat and tidy system, working precisely to Head Office’s requirements, is the organization man’s dream. The ideal computer is one that handles well-specified tasks to clearcut, predefined rules. As long asthe economy is booming, it is profitable to concentrate on the majority of conforming” customers who are prepared to accept the limitations of mass-produced uniformity for a few cents’ savings. Once hooked on the technology, the companies become trapped. For instance, the current investment in programs written in old-fashioned languages such as COBOL is so vast that movement to newer and more powerful tools must wait until the old systems become obsolete. Is there an alternative? What about the vast research effort being put into fields such as artificial intelligence and expert systems? Surely the people who work in this area have avoided this trap? Unfortu nately not. Early AT research was heavily oriented toward highly formal tasks with well-defined rules, such as chess playing. It then became a cult based on obscure, mathematically oriented languages and very specialized techniques. The result is a self selecting “elite” whose profes sional ethos is completely out of touch with the needs of mere mortals, for whom mathematics is a distinctly unfriendly language. Even the expert system experts are caught in the trap. Look at the most successful systems and you will find that the application reflects the scientific ethos that has caused the difficulty in the first place. So is it possible to have an information processing system that handles messy data? Of course. But we have to stand back from the problem to be able to see the real-world wood from the mass of scientific trees. Werner Heisenberg and Kurt Godel have shown why complete knowledge of the universe is theoretically impossible. Of what can be known, humanity knows only an infinitesimal part. An adult knows a microcosm of this, and a baby starts knowing nothing. We have evolved to survive in a sea of virtually total ignorance. Everybody’s knowledge is incomplete, ambiguous, often wrong, and in places contradictory. Formalized thinking, of the kind that underlies computer technology, is a highly unnatural artifact of civilization. Even elementary mathematics represents a major mountain of formalism—one that many children fail to climb, despite thousands of hours of study. So it is clear that people process information in a very different way than do computers. There is no point in developing even more sophisticated computer-based information-processing tools if we are asking the wrong question. Common Sense In fact, it is very important that we ask the right question. Railways, and later cars and aircraft, have influenced the locations where people live and work by altering the attitudes to travel over the years. Computers, as our first real information processing machines, are influencing our attitudes to information, and hence the way we think. If we want to become more like computers, the uncritical worshiping of the current technology is the quickest way to do it. What we need is a modicum of that rare commodity, “common sense.” If we want to build an information processing system to handle openended, real-world problems, our starting point should be ignorance—not knowledge. It can be done. Every human baby demonstrates its potential. Perhaps they know “the answer” because they have not yet been trained in the finer points of scientific research! Chris F. Reynolds is managing director of CODIL Language Systems Ltd., Tring Hertfordshire, U.K.

Computers Are Limiting Our World

Interested in reading more?

Become a Member of