I was speaking to a guy with a PhD in mechanical engineering and a background in optimization and the topic of genetic algorithms came up. He dismissed them as "BS", saying that the dependence on randomness to optimize a certain situation is not mathematically sound. I am not familiar with other optimization techniques so I understand that the field of optimization is something I am highly unaware of, but this still confused me.
If genetic algorithms are not mathematically sound, why/how are they used? I saw this wikipedia page on the many uses of genetic algorithms and it seems quite extensive. Is it dismissed by theorists but used in industry because it gives an acceptable answer efficiently / is easier to encode?