to simulate learning use genetic algorithm natural selection. for what types of problem are genetic algorithms suitable? Do they really learn? How can simple genetic algorithms be implemented

The approach to simulate learning is genetic algorithms, based on an analogy with biotic genetics and natural selection. The problem state is coded into separate parameters called genes. A random set of states is then allocated. The system then goes through a series of generations. In each generation, some of the individuals die, some breed and occasionally some mutate. The idea is that the individuals that survive will gradually become better, as measured by the same costs that drive the natural selection.You could address that following issues: for what types of problem are genetic algorithms suitable? Do they really learn? How can simple genetic algorithms be implemented?

I need a appropriate problem in this area .
A proposal should give a broad outline of the problem with
tasks to be completed and a schedule of work .