Random Variable
A quantity whose value is decided by a random process, described by the distribution over the values it can take.
A random variable is not a variable holding an unknown number — it is the whole mapping from outcomes to values, together with how likely each one is. Writing for a die roll means all six faces and their probabilities, not one particular roll.
Discrete random variables take countably many values and get a probability per value. Continuous ones take values in a range and get a density instead, which is why the probability that a continuous variable equals exactly 3.0 is zero while the probability it falls between 2.9 and 3.1 is not.
Nearly everything in machine learning is one of these once you look. The label given an input is a random variable, and a classifier's job is to model its conditional distribution. So is a model's prediction under dropout, the loss on a randomly drawn batch, and any metric computed from a sample — which is why a single benchmark number without a spread is an incomplete report.