Most of the time you will set constraints to ensure that only valid inputs are chosen by constrained random, but you can send in bad data as well to see if it is handled correctly. What I am more concerned about is that you don't actually test anything at all, you just create a false illusion that you are.
Perhaps I have misunderstood, but isn't there a danger with what you're suggesting that you test a design to check it does what it should, but you're not checking to see that it doesn't do what it shouldn't? I can think of a couple of things that caught us out over the years where I thought - I wonder if random testing would have spotted that?
As we unveil EE Times’ 2015 Silicon 60 list, journalist & Silicon 60 researcher Peter Clarke hosts a conversation on startups in the electronics industry. Panelists Dan Armbrust (investment firm Silicon Catalyst), Andrew Kau (venture capital firm Walden International), and Stan Boland (successful serial entrepreneur, former CEO of Neul, Icera) join in the live debate.