KNOWLEDGE MANIPULATION

by Rick Handrich

From DIMENSIONS, New Mexico Commodore Users' Group, Inc., July 
1988


One of the big buzzwords in computing these days is "artificial 
intelligence." Some people would argue that the words are a 
contradiction in terms, and that there could never really be such 
a thing. 

I think that argument is mainly about semantics. There certainly 
is a field of computing that calls itself by that name and whether 
or not it is true intelligence isn't really the point. This area 
is the future of computing to a large degree, and if you're 
interested in a career in data processing, you needn't be ignorant 
of it because you own an eight bit machine. 

Artificial intelligence works best in a narrowly defined area such 
as bridge building or medical diagnosis which has a lot of 
vocabulary and theory that applies only to that discipline. 

Programming for knowledge processing, as it's also known, involves 
presenting the computer with a lot of different options for the 
solution to a problem, and teaching it to make a "good guess" as 
to which solution to try. Chess, checkers, and other games of 
skill which the computer is good at are examples of this art. The 
computer doesn't just evaluate every possible chess move, then 
select the best one. If I remember correctly there are something 
like 10 to the 26th power possible moves in a chess game, and the 
sun would go out before your computer had time to analyze them 
all. 

This "good guess" technique is called heuristic programming. I'm 
not going to attempt to explain it here, but as a fan of good 
words, I like that one a lot. The most common programming language 
used in the field is LISP (List Processing). There are several 
versions of this language available for the C-64/128. I recently 
heard of a new one, which I believe is available on Compuserve, 
that is supposedly very good. 

Certainly such a language can't be developed for a 64k machine 
with all the power and flexibility of a mainframe application, but 
it is enough to get you started, to teach you the fundamentals and 
the "feel" of the language. There is also an ABACUS program 
available called X-PER which uses techniques similar to LISP. You 
could easily set up a file to, for instance, identify birds, or 
predict the weather. I don't know that it would forecast New 
Mexico weather any better than a meteorologist, but a well 
designed file would identify, say a spotted brown bird, 6-8 inches 
long with a reddish tail as some type of thrush. Then it might ask 
you if it displayed any characteristic movements, and when you 
said, "Yes, it picks its tail up, then slowly lowers it," the 
machine would say, "Aha! it's a hermit thrush." Setting up such a 
file for an area in which you have a special interest, would be 
rewarding for you both personally and very possibly 
professionally. That kind of knowledge base is the direction that 
data processing is headed, and if computing is more to you than a 
hobby, you should pick up the concepts wherever you can.

The Japanese have embarked on a national crash program to bring 
knowledge processing to the consumer market within the next ten 
years. If they are successful, you should be preparing yourself 
now to take advantage of these systems as they become available. 
Once again technology will have taken a quantum leap, and the 
people who have the conceptual knowledge to use it will be in 
short supply. You could be one, and your eight bit machine could 
be a strong contributor to your preparation. If you know of a 
serious program for any Commodore machine that is probing this 
area, I'd like to hear about it. Computing is serious business, 
and those who realize it the quickest, and capture the momentum 
early on will be the people with the power in the next decade.