Expert Systems in Decision Support

R.B.Jones, 82/11/1


The writing of this paper was started as part of an attempt to secure myself a job which would exercise my interest in Artificial Intelligence. It was written without benefit of a significant amount of research, entirely in my own time.

My qualifications for being considered an authority in this field are insubstantial, I have no practical experience in Expert Systems and little experience of the superstructure (or other) products most likely to be affected by such developments, but my interest in Artificial Intelligence spans a period of more than fifteen years. I have found writing this paper an interesting and stimulating exercise, and have had no difficulty in persuading myself that some of the ideas presented have merit. I hope therefore that it may prove of interest.

It is tedious and space consuming to preface each assertion with indications of uncertainty and incompetence, and I have therefore written the rest of the paper in an unwarranted but brief tone of authority.

The paper consists of three main sections. The first (section 2) was written in deference to some doubt on my part about whether I and the rest of the company understood the same thing by the term 'Expert System' (occasioned in part by the blank looks I sometimes get when I suggest that an understanding of logic insight be relevant).

Section 3 was written not intentionally but compulsively. I am persuaded that the techniques involved in expert systems will be very important in the medium, to long term future of computing, both in influencing, and in permitting the exploitation of new architectures. I believe that success will not be a matter of having the best semiconductor (or superconductor) technology, nor of making the largest investment in 'the fifth generation'. It will require a clear vision of how to exploit effectively the techniques developed and developing in Artificial Intelligence labs throucthout the world. ICL could be in the forefront of these new developments.

Section 4 deals more directly with the problems involved in effectively integrating expert systems into our products.

Frequent reference is rode to the objectives, and the methods proposed for their attainment, of the Japanese 'Fifth Generation' project. In doing so I do not wish to imply that we should be engaged in slavishly following the Japanese. Simply that they should not lightly be dismissed. There are few other sources of serious (whether credible or not) suggestions on how we can exploit the advances of technology.

A 'conclusion' is supplied, to prevent the unwitting but persistent reader from falling off the end of the paper.

up home © RBJ dated 1982/11/1 HTML 1996/6/6 edited 1997/4/18