True, genetic algorithms are only implied, but those implied gradients are used in the more successful evolutionary strategies. So while they might not look lik

True, genetic algorithms are only implied, but those implied gradients are used in the more successful evolutionary strategies. So while they might not look like it (because it's not used in a continuous descent) they still very much work like (although they represent a smoother function than) regular back-prop gradients when aggregated.