f.Christian, genetic programming would have been invented even if evolution was rejected. It's an useful technique, and it does work in certain situations.
The two limiting factors are (1) hardware.. going through many levels of recursion with potentially large populations... and (2) it's hard to define a fitness function (e.g., a measure of how "good" the result is that the computer can "understand".
For example, we can't use genetic algorithms on recipes because there's no computational way to measure how good something will taste. But, for a robot throwing a ball, circuit design, and
finding the world's only working program written in Malbolge... GP is great.
In a way, it's a very advanced trail-and-error technique. There's no religious implications about it. I don't even know what AndrewC is trying to accomplish in arguing here--I'm just defending GP as a legitmate, useful technique.