In undergrad I feared a feeling of locked-in-ness, and ditched my intention to do a PhD in math (which I think I could have done well in) partly for this reason, though it was also easier for me because I hadn't established close ties to a particular line of research, and because I had programming background. I worked a couple of years in programming, and now I'm back in school doing a PhD in stats, because I like probability spaces and because I wanted to do something more mathematical than (most) programming. I guess I picked stats over applied math partly out of the same worry about overspecialization; I think stats has a bigger wealth of better-integrated more widely applicable concepts/insights.

I'm a mathematician, and I had similar problems in the past. Here are some things that have worked fantastically well for me.

First of all, just start talking to people in other fields at your university. Start with "nearby" fields (like theoretical physics) and progress from there. See what they're up to; you might find opportunities for fruitful collaboration. These people need not be professors; indeed, since they're used to being pestered by eager students trying to climb the social ladder, you might find professors unable or unwilling to give you their time unless you can convince them it'll be to their benefit. Fortunately, talking to students can often be just as worthwhile.

Other things you can do: read first-year textbooks. In the fields that you find interesting, read more advanced textbooks. This can be highly enjoyable in its own right; I like to savour the different kinds of argumentation and notions of salience that get used in different fields. You can also attend conferences and that sort of thing.

If you're interested in escaping the bounds of academia, of course, you should also talk to people who aren't academics. You'll still want to find people doing something worthwhile, of course: I find I get along best with artists, radical/activist types, social workers, etc.

One caveat in all this: approach unfamiliar beliefs and methodologies with openness and respect. Often it takes awhile to really grok why things are done in certain ways, and moreover, people tend to take offense (rightly!) at newbies who are too quick to attack what they don't yet understand. If something seems silly or irrational to you, it may very well be; but people are on the whole more rational than the LW crowd tends to give them credit for, and so it's just as likely that you simply aren't familiar yet with their motivations. Ask lots of questions, but ask them with the appropriate attitude, and actually listen carefully to the answers you get.

I know multiple instances of math majors switching to non-math fields and easily getting good positions. I myself was doing fairly good research (improving on the current best technique for solving a problem that people care about) within 8 months of switching from math to robotics (although I also spent a year doing computational cognitive science in the time between math and robotics).

It sounds like the real issue is figuring out what field to go into. Personally I just tried a few different fields until I started to get a better idea of what was going on, and finally was able to narrow down to a shortlist of problems that I consider really important (probably similar to paulfchristiano's, it looks something like synthetic biology, brain scanning / simulation, machine learning, environmental engineering, clean energy, and materials science).

If you ask yourself what you would like the world to look like, it is usually easy to at least differentiate between things that do a lot to move you towards that goal, and things that do very little; maybe at the top it is hard to figure out which research programs will be the most effective, but it is probably not difficult to find something that is more effective than what you are currently doing, which should be good enough from the perspective of deciding to switch.

I was in a PhD program for Physics, then left it for Operations Research (where I will likely get a masters then exit, but may stick around for the PhD). In some sense that is the field of applied rationality, so you might want to check it out (it may be called Industrial Engineering where you are).

Did I feel locked in? A little. I was leery about committing to a research group for several years, as I hadn't come across any groups that I was really interested in or ideas I really wanted to pursue. But I always had very broad interests (I have programming skills, employment experience, finance knowledge, and know a fair amount about math), and so while I eventually realized I was sticking with my plans out of inertia I had several other options that I was ready to jump ship for.

I ended up deciding on OR rather than economics much for the same reason I decided on physics over economics- economics is what I get most fired up about (well, efficiency in general, but it's more fun to argue than do), but I wanted to make something besides policy papers. I was already planning to take my physics degree and do industrial research / invent something / go into consulting, and so I decided to change my "invent something" desire from physical things to software and scrap the industrial research plan in order to get into consulting in 2 years rather than 5-6.

You may want to look at being an actuary. They draw a rather significant income using skills you have most likely already developed.

When I said "most fields" I really meant "almost every field" (including math).

There are a lot of problems that need solving which will still get you publications, even in math. The issue is that there are also a lot of problems which don't need solving which will nevertheless get you publications (and that telling which are which is difficult). Given the number of people who seem to be somehow engaged with theoretical research here, it seems like a topic that could be profitably discussed.