Theoretically maaku is right. With infinite RAM and CPU you would just have to execute every random string of bytes until one of them happens to be strong AI. Of course, there's probably less than 1% chance of it being friendly...
Optimization power, possibly divided by available resources.
In a game of Chess, only a narrow set of moves will let me steer the future into a winning state. Well, the same is true for the Game of Life (the real one, not Conway's): we humans are intelligent because we're able to steer the future through probability to a-priori incredibly unlikely outcomes. Compare walking vs moving your limbs randomly.
It was a rhetorical question. The point was once you have a definition of intelligence - yours is one of many fine definitions - you can refine that into a metric for comparing different intelligences, and then you have a way for a comparison function given two program descriptions to determine which is "more intelligent".
Building artificial intelligence then reduces to undirected search, assuming infinite CPU and RAM.
