Listening to Steve talk, he's very upfront that there is no blueprint for creating a successful business/startup. His advice is not meant to be taken as "here's a list of steps, do them, and you'll be successful." His advice is more like a set of guidelines you'll likely want to pay attention to (or at least knowingly disagree with).

With respect to asking potential customers what they want, he's said specifically that you should do that, but don't take their answer and create exactly that. In his view, there's an "art" to startups that involves taking feedback and turning it into something that's useful. For example, the iPod/iTunes ecosystem wasn't something mp3 player users' would've come up with, but they probably would say, "I want it to be easier to get music my music on my mp3 player" and that's something Apple definitely knew.

There's definitely ambiguity around what you should asking potential customers and I don't think he'd claim there's a right answer about what to ask or not to ask. But it seems pretty likely that talking to people is worthwhile activity.

That said, he also admits there are exceptions to every piece of advice. You could avoid talking to anyone and still be very successful. It's just something he's seen correlated with success.

I'm not sure this is where I remember most of this from, but I think this podcast [1] has most of what I'm paraphrasing.

[1] http://startupsuccesspodcast.com/2012/02/show-134-prof-steve...