Artificial intelligence has been a buzz word for many years.

The term was created by John McCarthy in 1956, when trying to find a catchy motto for the Dartmouth (U.K.) Conference, which was the first conference to determine the premisses to build a decision-making machine. It was not until 1960 that the first attempts to implement "intelligence" on a computer had a small success with the completion of IPL, the first artificial intelligence language application.

The main difference between AI languages (such as IPL, LISP or PROLOG) and a standard language is the execution structure. Most computer languages separate procedures from data, while AI languages consider data as procedure. This sounds a little complicated now, and surely trying to explain the complete semantics would make a book (and there are many good books on AI on the market).

Artificial intelligence is divided into several fields, such as:

• Robotic, which tries to create a machine capable of autonomously fulfilling a task and overcoming difficulties not previously foreseen.
• Expert Systems, a vast data bank automatically capable of determining causes by inputting symptoms.
• Theorem Prover, mostly used in mathematics to prove the veracity of a formula under more than one circumstance.
• Language Comprehending Systems, machines that receive their commands by natural spoken language and that are capable of understanding all implications.
• Picture Interpreting Systems.

Taking all the above fields together and combining them into one computer may make a passable machine, but at this point it would be far from intelligent. Many attempts have been made to create a machine model capable of learning and using knowledge in an intelligent way have failed because the program got too vast for one machine. It is said that even a Craig computer has an IQ of less then 20, but the speed at which it can perform repetitive tasks makes it valuable (the original quote is "... they are very gifted morons").

So far, many demonstrations of AI have been less than satisfactory, even though some programs could actually fool the casual observer to believe that he/she is interacting with an intelligent being. The most successful probably was a program called Eliza (by Prof. J.Weizenbaum) that emulates a non-directive psychologist. Most non-psychologists that had a session with the machine were absolutely convinced that they had "talked" to an intelligent being. Prof. Weizenbaum was not amused and never tried to improve the program.

If you think that your computer could use some intelligence I suggest you try PROLOG to program your first steps. There are several PROLOG compilers that can be downloaded from the Internet, there is SWI PROLOG (developed by the University of Amsterdam, the Netherlands) as freeware, Strawberry PROLOG as Shareware and Visual PROLOG as a commercial application.

All AI languages are very complex, not due to the number of "commands" you have to learn, but due to the multiple implications of each of them. This can be overcome with a good reference book, obtainable at your local bookstore, at via Amazon.com or Borders.com (if you happen to live on an island like me and all your local bookstore carries are school and novels).

And don't fail to notify me if you succeed in making your computer "think." After all, the lure of an intelligent machine is what got me into computers some 30 years ago ... and mine is as stupid as ever.

*******************************************************************************

Username: Alf B. Meier
UserEmail: kafeniocom@rho.forthnet.gr
Comments:

Hi, I would appreciate if you included a working link on Thinking Computers if you want to use it