Artificial intelligence

Computers have become part of everyday life. Today's schoolchildren carry around calculating power which would have filled a large room forty years ago. Computer scientists are now working on the next generation of computers; ones which will have true intelligence. The first step on the way is the development of 'expert systems'. An expert system is partially intelligent.

At Stanford University a computer has been developed for medical diagnosis. It stores 125 facts about each of 600 diseases. It can tell doctors which questions to ask, and which measurements to take. It can then analyse the result and make an accurate diagnosis.

At the University of California they have been working on computers which can write stories, while IBM has developed the Epistle program for business letters which can correct spelling, improve grammar and style and even check for mistakes by giving a list of words with similar sounds. One of the most interesting programs has been developed at Massachusetts Institute of Technology (MIT), and it is a kind of computer psychiatrist.

The program, which was developed by Professor Joseph Weizenbaum, is called Eliza. It uses a technique known as 'reflective listening'. The computer seems to understand, and to make sympathetic responses. In fact it does not understand a word that is being typed into it. Nevertheless, Weizenbauni's secretary fell under the spell of the machine. Weizenbaum says:

Well, I asked her to my office and sat her down at the keyboard, and she began to type and of course I looked at her to make sure everything was operating properly. After two or three interchanges with the machine she turned to me and she said, "Would you mind leaving the room please?"

Eliza responds sympathetically to what is typed in, but that it never adds new information, or gives an opinion. Ordinary psychiatrists use this reflective listening technique, too.

Here is an example of Eliza in operation: