Artificial Intelligence

The first recorded use of the term artificial intelligence (AI) was in the proposal for the Dartmouth Summer Research Conference on Artificial Intelligence in 1956. This conference—essentially a month-long brainstorming session—is usually considered the birthplace of modern AI.

However, AI has many elements that predate this. Indeed, if one takes the possibility of the mechanization of thought as the key philosophical notion underlying AI, then this notion can be found in early works by Chinese, Indian, and European philosophers. Modern AI frequently acknowledges its debt to earlier thinkers, although often in a highly selective manner and frequently as part of factional disputes.

There have been many debates on the precise definition of AI in the years since 1956. The closest thing to an agreed definition of AI within the field is attributed to Marvin Minsky, an instigator of the Dartmouth Conference: Artificial intelligence is the science of making machines do things that would require intelligence if done by men.

This definition may be taken as a starting point and expanded. Even if we allow no sexist implication, many researchers would not wish to be restricted solely to human intelligence. Biological research on animal behavior during the 21st century strongly suggests that intelligence has evolved more than once on Earth. Many AI researchers are influenced by these discoveries in biology and might therefore prefer a definition that included things that require intelligence if done by animals.

It is important to distinguish the term artificial intelligence from the endeavor that is Artificial Intelligence. This is because much of AI research, AI technology, and AI theory may use different names. For commercial or political reasons, the term artificial intelligence has sometimes been thought to convey a negative image, so the endeavor has often been rebranded or referred to solely under the name of a component technology, such as machine learning, artificial life, robotics, or advanced computer science.

The endeavor that is AI has turned out to be as much engineering as science, if not more. Many of the technological successes of AI have come from building systems that solve specific problems, rather than from first developing an understanding of the scientific principles—particularly in the many areas where human intelligence is poorly or not understood. The attempt to reproduce human intelligent behavior in machines has often revealed a partial or complete ignorance of the way in which humans achieve such behavior. In this respect, AI has prompted and pushed the science of human intelligence rather than followed its development.

A better, expanded definition would therefore be as follows: AI is the study of intelligent behavior (in humans, animals, and machines) and the attempt to find ways in which such behavior could be engineered [Page 39]in any type of artifact. It is important to include “any type of artifact” because some intelligent behavior is produced by the right sort of computer software, some by the physically right sort of robot, and some by a combination of the two. Indeed, much successful AI has only a virtual existence as portions of computer code within larger systems.

...