Artificial Intelligence?
The term artificial intelligence' was introduced in 1956 (McCorduck, 1979) and is now in general use, although some prefer to call the discipline cognitive science'. Its goal is to understand the nature of intelligent behaviour. It is often loosely defined as encompassing all those activities of programming computers to perform tasks that we generally would say require intelligence' when people perform them.
Within computer science, AI has a somewhat more technical characterisation. AI is considered the branch of computer science that studies the representation and use of symbolic information (as opposed to using numbers only) and studies heuristic processes (as opposed to algorithmic processes in which there is a guarantee of success in a finite time). Chess, for example requires reasoning about plausible moves and requires representing symbolic features of a game, such as threats and control.
AI programs have been criticised because they lack common sense'. People use propositions to guide them through many situations, which AI lacks. How can computers do something similar to using common sense'. Common sense reasoning may be captured as a very large set of facts together with their exception conditions, ie; "Water runs downhill, UNLESS its 1. under pressure 2. in non liquid form
In the development of the INTERNIST program Dr Jack Myers added about 100,000 common sense' facts of medicine to improve the programs performance (Miller et al, 1984).
The default-reasoning' is one such mode in which a program assumes the general fact holds in a situation unless it is contradicted by a more specialised piece of knowledge. Taxonomic hierarchies play an important role in default reasoning because general rules can be written once to apply to all members of a class, with special exception clauses being attached to subclasses or individuals. For example, we would not need to represent the fact...
View Full Essay