Artificial Intelligence on any computer is a tricky subject, as technically it is not achievable - at least not using the accepted definition of AI, propounded by Alan Turing in the forties: if you can't tell whether it's a computer or a human beyond the keyboard then it's intelligent (since we assume that most humans are intelligent).

We are still a long way from the artificially intelligent machine, more than 30 years after Turing's death, so the rider to the title - make your micro think - is a little dishonest.

However, although the Brains have not given the Dragon artificial intelligence, they have produced a book of interesting ideas about AI. Unfortuantely, it gets off to a bad start with a couple of pages on science fiction robots and computers, moving on to reality via a series of tired jokes.

Chapter two blows the gaff: "As your computer is actually totally unintelligent..." but from then the book deals with things which make the Dragon respond in a more reasonable (if not intelligent) way, whetting the appetite with talk of natural language, problem solving and learning.

The book culminates with a large example (five pages of flowcharts, seven of program) which simulates a computer salesman, and does have qualities of cleverness - it looks to see if you can afford to buy the machine (A vision of the future?), whether the sales commission is high enough, etc - but no real intelligence. A two page section-by-section explanation is essential to understand the program.

A book full of interesting ideas, but the humour does not work well. The most encouraging feeling I got from it was that if computers are to be inelligent, they must also be user-friendly (or just friendly).

Piers Letcher