Martin Banks, Personal Computer World 10/89 - checked

October 1989

One of the reasons given by the experts as to why we need expert systems is that they can help capture precious knowledge in industry and commerce. Then, we can all take advantage of it.

The theory is wonderful, but the practice may be a little flakey. Take, as an example, the following situation. Hands up everyone who has gone to their bank only to find that the wise, kind and above all lenient old bank manager has retired, to be replaced by a thrusting, brash, opinionated and above all stupid young bank manager?

Wouldn't it be wonderful if the latter could have access to an expert system which has managed to glean all the wisdom accumulated by the former. Unfortunately, I suspect that the former's wisdom is based upon factors which are difficult to capture in a systematic fashion, such as the realisation that life is about more than making numbers add up.

It is possible to speculate, however, that the attitudes of the young, brash, thrusting etc bank manager are much more in keeping with a systemisable approach to life, making this 'expert's' knowledge the stuff which can be captured.

In other words, though the idea of an expert system is sound, it begs the question of whether it is workable. And it begs the question because there has to be serious doubts as to whether the wisdom of real experts can be captured in any coherent form that computers can exploit.

This, in turn, begs the supplementary question of whether we will inevitably be led towards expert systems that are used to capture the knowledge of only those 'experts' that have their knowledge in a form which can fit the 'system'?

It has been my own experience that, having seem someone I denote as an expert do something I would like to learn, and having asked them to teach me, they say something like the following: "its easy, all you do is this."

The 'this' in question is almost always totally unfathomable, and being told it's easy is not the most helpful tip one can ever receive. I for one would not like the job of trying to capture such knowledge for an expert system.

I would not like it for two reasons. Firstly, the experts in question often have no specific idea of the 'how' of their knowledge; they just do it. Secondly, if they can't work out what they are doing, there is not much chance that I will be able to, either.

This does, in practice, pose some interesting side issues to the question of expert systems capturing important knowledge for future use. The success of any knowledge base thus built up will, for example, be dependent upon the capabilities of the system to capture the information; on the ability of the expert to, effectively, 'teach' the system the knowledge; and the ability of others to verify that knowledge in some way.

Could this be a tall order? I hope not, but the practical achievement of a truly expert, expert system-generated knowledge base must be suspect, at least for now.

I hope it is not a tall order because I am aware that too much human knowledge has been lost or destroyed already. Sadly, much of that knowledge has been purposely destroyed in the name of other, supposedly 'higher' knowledge causes, such as religion, but that is another story. It does, however make one imagine a possible future where all that expert knowledge is stored somewhere, on tape or disk. Just think of the havoc one individual could wreak with a dipole magnet?

But, leaving that aside, expert systems have another aspect which I feel is worth consideration (especially if you're into paranoia). Let's suppose they work?

What type of future might we find if expert systems can actually capture knowledge and wisdom and allow us to exploit it fully?

Well, I suppose the most obvious suggestion is that it would be a much better world than it is now. The illogicality of much that we do would be stopped for the arrant nonsense it is, and even young, brash bank managers might learn a little humanity.

But that raises an interesting issue: would they learn, really learn, or would they just understand that their job involves pressing certain keys and following the displayed answers to the letter?

Would we reach a point where the knowledge base is so complete that, for all practical purposes, we don't need to know anything ourselves? If all the answers can be found in 'the system', would there be a need for us to think at all?

In systems terms, of course, that might be 'a good thing'. If we have a greater degree of uniformity in thinking, or more specifically non-thinking, then we can become more 'standardised' in our approach to life and work. As we all know, standards generally ARE a good thing, especially in computing terms, for it means that we get much greater benefits from the intellectual efforts that have gone in to creating the machines in the first place.

So maybe a standardised human run by expert systems might be a better bet than what we have now.

The real question, of course, is only going to be answerable in hindsight. One of the perennial accusations levelled at computing and its practical utilisation is that it is, by its very nature, de-skilling.

It has to be said that, while computers can be accused of moving the skills base around amongst the population, skilled people do not necessarily have their skills exorcised. For example, a DTP system may allow a 'secretary-type' to layout and print a brochure or pamphlet, shifting a skilled task from a 'graphic artist-type'. But that is not necessarily de-skilling the latter. That same DTP package can be fully exploited by the latter actually extending their skills base. Then, print output which is unachievable in any other way can be obtained.

The doubt is that if expert systems do achieve their full paranoid potential, we may no longer have the skills to recognise they have done so.

end