No, Netflix uses collaborative filters (not feature learning) for recommending shows, and does not collect feature data on individual "consumer" preference (at least not publicly).
What my company does uses sensory and preference data at the individual level, and projects for demographics and populations. Good food doesn't feel fake. Good drinks don't feel fake. In the long run, few products will be made by averaging tastes for the mass market - that's what coke and pepsi are.
Also, creating art is not as simple as entering the infinite realm of possible songs (or paintings, or meals or...) and just choosing a set of parameters. Again, a broad set of parameters, sure. I'm going to write a rock song. I'm going to cook a boeuf bourguignon. But how to capture the weird unsystematic idiosyncrasies of any given individual, and do so in a way that doesn't feel soulless and artificial?