The programs don't make "intuitive" moves, but nor do they exactly perform brute search - the search tree is far too wide.
They actually use a mixture of techniques. For example, in the beginning they may select moves from fuseki databases of accepted good play, and have some knowledge of joseki for plays in the corner.
During the very end game they may be able to brute force search and make sensible decisions about the order of play - however there is also a book about using simulated annealing with neural networks to decide the order moves should be played. <a href="http://www.amazon.co.uk/Mathematical-Go-Chilling-Gets-Point/... Go: Chilling Gets the Last Point</a>.
There have been attempts to code "rules of thumb" into the Go algorithms, such as knowledge of bamboo joints, table shapes, etc., but the recent improvements have largely been down to techniques such as evaluating the position after allowing two consecutive moves and even simply trying various random plays rather than an exhaustive breadth-first search.
There certainly used to be a mailing list for people interested in the development of Go-playing programs. I can't remember the address but if you broaden your search terms to include "Go, wei-chi, weiqi, baduk" you should be able to find it.