Any AI developer would actually know hard it is to predict and simulate human behavior patterns.
Most AI Algorithms that fall under the genetics category (Neural Net as an example), end up being hacked to work in simulations.
The amount of "garbage" and errors that would result from such algorithms is huge.
Consider life simulation games
http://en.wikipedia.org/wiki/Life_simulation_game
These cost a lot to make and "should" as initial requirement simulate life.
The equation parameters are huge.
How would one simulate nature? One of the fundamentals principle of nature is that nature is optimized. That said, any selection made by nature guarantees an optimal result [or optimal way to get a result]
http://en.wikipedia.org/wiki/Natural_selection
a genetics algorithm for natural selection can be easily implemented, but it depends on a "function" which is what favors certain traits over others? how would one determine adaptive evolution if one does not fully simulate the "environment" that the generations need to adapt to?
This topic exceeds our perception as scientists, for one reason, we do not have enough power to simulate real nature, we fake it (video games for example). Once we can simulate a natural environment, we can simulate the creation of a universe [let alone humans].
But again this is not a simple 1+1 formula. The small microscopic difference is what makes uniqueness, randomness, and thus individuality [as mentioned above].
