Once people find out that you build software, they are (in my experience, at least) all too happy to tell you what kind of app/website you should build.
The bad news is that most of the ideas they have are unworkable, unmarketable, or just plain stupid. The good news is that underneath it all, there is often a kernel of a good need that can be solved in a better way than they are expecting and/or applied more generally.
Over the holidays, my sisters told me about an app I should build - it seemed silly and unmarketable to me at first, but as I thought about it more, I saw a lot of potential uses (many of them applicable to a commercial context) for a more general solution to their problem. I'm working on that app now on the side.
The other thing to think about is the way you approach people for ideas - don't ask them what app you could build to help them, ask them what their problems are. Even better if this is in the context of their job - this makes it far more likely that you could charge enough money for your software to make it worthwhile.