Mutations can be random, survival of the fit is less random.
Mutations may not be as random as you think, if a gene is predisposed to mutate in a certain way. In fact, we already know this to be the case with certain genes.
Such a mutation could in fact be far more impactful on the gene pool, even if having a net negative impact on survival, than other mutations that provide minute positive impacts on survival. All that is required is that the math be weighted in that mutations favor. If the mutation is a 1/10,000 chance, but only causes 1/100,000 individuals to fail, then it will continue and increase. Scientists shouldn't preclude that every change is an adaptive necessity for survival.
I saw a youtube video of a wild goat of some type. It had unbelievably long horns and was using the tip of its horn to scratch its butt over its back. A handy use of the horns no doubt, but would you conclude that those monstrously long horns were a survival adaptation that came about out of a usefullness for butt scratching? I would not. Yet, biologists would happily conclude that they know the precise purpose that the long horns were evolutionarily adapted for. Perhaps that purpose is no different from the butt scratching, i.e. the goat developed progressively longer horns and found a few good uses for them.