As long as we're talking about, as you say, significant help rather than solving the whole problem, about what can be done without having full natural language - then I think this is one of the more promising areas of AI research for the next couple of decades.

I talked a few months ago to somebody who's doing biomedical research - one of the smartest guys I know - asking what AI might be able to do to make his job easier, and his answer was that the one thing likely to be feasible in the near future that would really help would be better text mining, something that could do better than just keyword matching for e.g. flagging papers likely to be relevant to a particular problem.