The question seems a bit misleaded!
GA is nothing but a technique (some would say algorithm but that's a bit strong) to optimize a function! That is not in any sense a limitation. What that optimization is all about is pretty much limited to the imagination of the user. You could use it to optimize social behaviour (what actions to perform in a given context) or why not optimize the movement of you army? But in many cases a GA is much to ineffective to use for optimization and instead an assimilated annealing approch will help you get a more effective optimization (to the price of accuracy). Put try to not think of GA as more as a technique influenced by genetics for optimization/classification.