Every now and then I see a quant problem solving question with one of the answers options being "Cannot be determined". In all these cases, that has not been the answer choice. I am wondering, is that ever the correct choice or can I pretty much write it off as a guessable choice?

how many cases have you seen? where have you seen them?
i've only seen 3-4 such problems, and that's from an extremely large sample of problems.

even if the correct answers were chosen completely at random, there would only be a 1/5 chance that "cannot be determined" is the correct answer. therefore, unless you've seen a LOT of these (like 15-20+), it's not wise to jump to that sort of conclusion.
analogy: imagine a spinner that can spin numbers 1-5. even if you spin the spinner 5-10 times, there's still a decent chance of not getting any "5"s.
on the other hand, if you still haven't gotten any after 20+ spins, then you can start to conclude that something is wrong with the spinner.

--

ON THE OTHER HAND:

remember that a VERY common theme of the gmat is writing problems whose solutions aren't what they seem to be.

so, if you have a problem that seems to give very little information, and offers up "cannot be determined" as one of the answer choices, then you can be virtually certain that that's the wrong answer.
ironically, they try so hard to be unpredictable that they occasionally become ... predictable. this is one of the ways.

--

also:
you won't see too many problem solving questions on which "cannot be determined" is a choice, because the other half of the quant section (IE data sufficiency) is already completely dedicated to this exact concept.