What is Artificial Intelligence?

Artificial intelligence (An is the process of programming computers to carry out tasks which simulate the intelligence or thoughts of humans. A computer chess game, which beats many good chess players, is an example of artificial intelligence. The computer simulates certain thought processes of humans by acting and reacting to each move. If the computer could "think" of a good trading strategy in the same way it beats chess players, there is promise in using AI. Two areas of AI being aggressively pursued in trading are natural language systems and expert systems.

A natural language in computing is more closely related to English, unlike the typical programming language which can be difficult for many people to comprehend. Many programs that test trading systems employ natural languages to assist the trader in creating and analyzing a strategy.

An expert system is another area in AI which is being actively investigated. An expert system learns information in terms of rules or factual information from experts and then attempts to make decisions from what it has "learned." A thermostat is a simple example of an expert system in that it has learned if the temperature goes below a certain amount it will decide to turn on the heater.

The knowledge base of an expert system are the rules, facts, and other information which it "knows." The inference engine interprets the information to arrive at conclusions about the problem. An expert system will learn all the trading methods of a good trader and then try to make trading decisions from this information.

Heuristics refers to programming a computer to think about how it might solve a problem, such as a simple one of determining how to playa computer game. The computer will try to get the most points, and, through trial and error, eventually arrive at a way to win.

Neural networks involve having the computer learn the different patterns of the market and then arrive at decisions to trade. It is different than expert systems in that the computer does not have the expert information "told" to it. The system arrives at a trading decision based on "learning" pattern recognition and the rules it develops. Neural networks come much closer to learning, the same way we believe we did as children (unless you had know-it-all parents, in which case it would be an expert system).

Much has been written about artificial intelligence and the promise it holds for mankind, but specifically for our purpose which is trading. AI can be valuable in many areas of trading and is aggressively being pursued in areas pertaining to pattern recognition. This area is a natural match where much money and time is being spent. But, there is still a long way to go before AI can be used practically by the basic computer user. The promise is high but the road is long.