~
I think the problem with uneven distribution might actually come down to a small sample size.
It’s proven that under certain specific conditions (in particular, if the sample is random, meaning the observations are independent) the sample mean converges to the true expected value of a random variable as the sample size (number of observations, trials, measurements) approaches infinity.
Simply put, the longer the game runs (the more generated card deals), the closer the outcomes will align with the expected value – and the more often pocket AA will beat pocket KK, as they statistically should.
But those are mathematical laws – and I’m more inclined to agree with them than with the claim that the issue lies in selective memory.