Firstly, the subprograms I have described were chosen largely because the software was free and could be used without modification. This generally means that it was also old, and written as a demonstration of some concept rather than as a robust and reliable system for heavy-duty use. In some cases, there are relatives of these subprograms that do a lot better. For example, one can obtain heavy-duty ``industrial strength'' planners that have proved themselves indispensable for planning industrial assemblies.
Some of these methods - most of the ones used to implement expert systems, for example - only work within extremely restricted domains, but are useful as long as we recognise their limitations.
Note also that recent research is producing new methods which may succeed where the old ones failed. One example of an extremely useful application is ``inductive machine learning'', which involves constructing general logical rules from particular examples. This looks very promising for drug design, amongst other areas. If we're lucky, I may be able to provide some ILP software later in the course.
Secondly, as far as AI's contribution to psychology goes, it has made the subject more rigorous, by demonstrating how much is left unspecified in a theory which is not precise enough to be runnable on a computer.
It has helped make the idea of mental states respectable, shown that we should ask in detail how the mind does things as well as what it does, and demonstrated the kinds of language in which we can talk about mental processes.
It has given us a number of general ideas about how computational systems can be organised, and terminology for describing their properties. These include the ideas of representation, control, algorithmic complexity, and levels of description, as well as notions about how to connect small computational systems to build large ones.
It has offered us specific answers to some psychological questions, notably in early vision.
There are other views of AI's contributions. Sometime during the course, you should look at The Foundations of AI, edited by Partridge and Wilks (CUP 1990; PSY KH:P 025). It contains a few very technical papers on subjects like theorem-proving, but also a number of general ones by Marr, Boden and others, giving their views of AI.