Lee Spetner makes the distinction between random and non-random mutations. He says that there are no examples of random mutations increasing information in the genome. However there are inbuilt mechanisms that can cause non-random genetic changes to respond to environmental events.
Consider a coin flip...
If we knew all the information about the forces that were involved in flipping it and causing it to tumble around in the air, we could accurately predict which side it lands on because, technically, it is not random.
But from our perspective it doesn't matter, we don't actually know which side it is going to land on so it appears random to us and gets the job done.
It's similar for mutations. The physical organism is what interacts with the environment and the mutations are to the genome that is buried deep within. The mutations may not, technically, be random in that there are causes for them, but from the perspective of the environment and its interactions with the physical organism, the mutations have the appearance of being random from the perspective of the environment.
The environment is blind to the alleles and only sees the phenotype. From that perspective, the mutations are random even if they do have predictable causes at the genetic level.
Mutations are random "with respect to fitness", not purely random as in a completely non-predictable manner.
Make sense?