Earlier quoted context omitted.
As an example programmers can understand, machine learning is a better model for this than a logistic map. Natural selection is basically gradient descent with random walk as the "step" algorithm rather than back propagation. At each step, you randomly adjust the weights by a small amount. You do this with like 100 candidates and pick the best one that lowers the cost function the most. This is one generation, and yo…
You're talking about evolution here, but I was never talking about evolution; nor was the original article.
Reread my post initial post with this in mind, because you don't get it. I've been speaking about evolution from the beginning and you've been missing the point since the beginning.