I have been to a lot of seminars over the years. One of my favorite parts of seminars is the question session at the end. It is fun to interact with the person about their work and see how their interpretation fits in with your perspective of the results. It is also a lot of fun as a speaker, IMO. Almost always someone will ask a question which you, the speaker, won't know the answer to. It may be that there is no answer, or it could be that YOU just don't know what it is. Either is OK. There are graceful ways out of this situation. I think the best option is to start with "I don't know" and then expand on either what you know is NOT the answer - based on experiments you've done or other published work - and/or discuss ways that you could address the issue. These are also good strategies for grad students giving a qualifying exam, by the way. A skilled (non)answer makes it clear that you are well-read and knowledgeable, because you are able to understand the question, but recognize the limits of what you know.
BUT this assumes that you have been asked a question that is interesting. This is not always the case. I ask a LOT of questions at seminars. It is possible to ask misguided, out-of-context, or just plain ridiculous questions. I know, I have done it (NOT ON PURPOSE!). And this is why I find it SO irritating when speakers start every answer (or just answers to the questions that they don't know the answer to) with "that is an interesting question!" or something similar. Because it is NOT always an interesting or even good question.
It is OK to not answer a question that is not interesting, and it is possible to do it while being pleasant and diplomatic. If you are really good, you may even be able to twist the ridiculous into something interesting. YAY!
But please, don't pander to me.