I'm also not an expert on numerics, but very generally we expect to only lose precision during computations, never gain it. So, at the very least we expect that the precision of an expression be no greater than the minimum precision of the numbers within it. So, for your vector, the minimum precision of the numbers within it is MachinePrecision. Granted, it may seem like a vector doesn't really have a precision, but what else would you suggest that its precision should be? If we were to apply some numeric computation to it that combined all the elements, we'd expect the resulting precision to not exceed the minimum precision of the elements. However I do think Mathematica will try to preserve as much precision as it can without losing correctness, so it might be able to "increase" precision of some numbers for some computations. But I'm way out of my depth here, so take it with a grain of salt.
Having said all of that, what is your actual problem that you're trying to solve?
