An abridged version of this paper appeared in R. Ennals and P. Molyneux (eds.), Managing with Advanced Information Technology. London:Springer-Verlag. 1993

4. Conclusion

What else does AI currently offer? The emerging applications of AI are too numerous to catalogue, so rather than list the applications I shall summarise the basic approaches to using AI that are open to business and industry:

  • The simple and cheapest solution may be the `off-the-shelf' knowledge-based system for supporting fairly routine judgements and decisions in certain types of generic task. Accountancy, taxation and tax analysis, assessing a company's financial future, and so on, are areas where the required knowledge is of a standard kind that can be built-in to the system, the complexity of the underlying technology being hidden from the user.
  • Otherwise, there are semi-custom applications. Some generic knowledge applicable to particular application domains is already built-in to the system; but recognising that most companies have quite specific personal requirements of the technology, it allows for the end-user to add -- for example, from a menu -- the remaining knowledge specific to a company's problem.
  • A third option is application development tools, for example, expert system shells such as XI Plus, Crystal, Instant Expert, and so forth. These are available for micro-computers, and are by and large sensibly priced.
  • A more serious investment, in terms of cost but also of potential pay-offs at the end, is in knowledge engineering and development tools like ART, KEE, KAPPA, ...; environments such as POPLOG. Unlike the former, these very often require the use of some AI-specific knowledge and skills to be used successfully.
  • The final option is to use one of the few AI programming languages (LISP, PROLOG, POP-11, SCHEME, OPS5, SMALLTALK, ..., & cet) to craft your own bespoke product. The advantage is transparent: you have a unique application customised exactly to your personal needs. The investment, however, is likely to be high, in particular if it requires importing the programming expertise from outside.

In Britain, more than anywhere else, the platform for AI and expert systems development has been largely micro-computers. A broad array of good quality AI software is now available for IBM PC and Apple Macintosh computers, with most basic software sensibly priced below 1,000.

What will be the training needs? Will the company investing for the first time in knowledge-based systems need to engage new skilled personnel? This will all depend on its AI requirements. If a simple expert system shell will do the job, every major computer exhibition now has stands demonstrating commercial shells—such as Crystal—that can be learned and used in a very short time. For more advanced technologies, the need for training of key IT personnel will become more likely.

But I must end this chapter on a quieter note. Writing in January 1983, in a book charting the emergence of `Fifth Generation' computing initiatives world-wide, Edward Feigenbaum and Pamela McCorduck soberly note that

Britain is the only major Common Market country to have experienced a decline in privately funded research and development between 1967 and 1975—down 11 percent in those years. It's largely thanks to this indifference on the part of private industry that the percentage of Britain's GNP spent on basic research also fell from 2.32 percent in 1964 to 2.09 percent in 1975. (Feigenbaum & McCorduck, 1984, p.212)

"No one", they conclude grimly, "expects any change in these trends. ... the British ... have demonstrated how to turn a nation from a winner into a loser". This is a harsh judgement, but hardly unjustified. The economic future of any nation today depends crucially on its investment in the new technologies, since they will quite clearly have implications for every other area of the economy: telecommunications, retailing, manufacturing, agriculture, you name it. This often means taking gambles—such as the gamble the Japanese took ten years ago when they announced their Fifth Generation project—looking not for the `quick and dirty' short term pay off, but with an eye to the future. Anyone can make a fast buck—there's usually nothing especially clever in that; only nations with vision, however, can seed the buck that will grow.