Currently the term "ArtificialIntelligence" is used almost exclusively as a marketing buzzword. In so far as it has any meaning it means either:

1. the development of programs and computers that act like people, a la the TuringTest. Work on this has been somewhat haphazard, but various attempts have given us Prolog, SymbolicManipulation, NeuralNetworks and other techniques. After 30 years the easiest way to build a machine with human intelligence takes 9 months...
2. anything a computer does for the first time. The first Accounts Receivable system was ArtificialIntelligence, as was the first Compiler, the first WebBrowser and the first MTA.

The definition of Artificial Intelligence changes as programs get more sophisticated - mostly because things that we think require intelligence eventually prove to be do-able with heuristics and/or brute force and/or new algorithms. For example, machine vision (recognising objects, driving cars, etc) used to be thought to require intelligence, but now computers can do these things fairly easily, without what we would call "intelligence".