In a genetic algorithm, fitness is computed by some expression involving the characteristics of the system you are trying to optimize. In biology, fitness is computed by examining the relative numbers of offspring. If you tried to write a genetic algorithm with the biological version of fitness, it simply (and obviously) wouldn't work.
I’d say that fitness in biology is based upon the relative number of offspring that survive. Species and bloodlines that get too inbred can have lots of offspring, but they do not survive. I contend it is still a matter of fitness of the solution to the problem-set.
Additionally, many approaches with GAs use the approach of the fittest “parents” get extra children, while the worst parents get no children. Weight the reproduction of entities to fitness.
But see, here we’re discussing applications of evolution. NOT intelligent design! For evolution can be tested, modified, checked. ID simply can’t. You can’t even have this discussion about ID because it is fundamentally a supernatural basis. It is because it is.
THAT is what makes ID fundamentally not scientific.