It has to be an interaction between representatives of the general public—constituencies that can be identified in advance as having problems—and scientific researchers. If you're trying to match research projects with human needs, you'll need to know what the scientists envisage they are going to be able to do and what the public needs.
I've written at some length about this. Chapter five of my book Science in a Democratic Society is about what I call “well-ordered science”. That is when this match is worked out and tries to come together.
That's the best I can do in a minute or two.