Odds are low, but the frequency is high. An oft-cited IBM study from the 90s determined that memory will get a cosmic ray bit-flip once per 256MB per month. So, an 8GB system will see about 32 bit-flips per month. Probably more with modern memory. Of course, as you mention, it's not likely that several would occur at the same time in nearly the same place.
That vast majority will be in unused memory, executable code that never gets executed, or even in code or data that, while corrupted, simply doesn't have a noticeable effect.
But, what about a single bit flip of a parity bit? Does a good bit get "corrected" to an incorrect value? Serious question, as I really don't know enough about the specifics.