Genetic algorithms are useful when the landscape is awkward
Genetic algorithms are useful when the search landscape is awkward.
If the objective is discontinuous, noisy, constrained, or difficult to differentiate, population-based search can explore candidate solutions without requiring clean gradients. The tradeoff is efficiency and tuning effort.
They are a practical search tool, not a universal optimizer.
Related: