People don't get confirmation of those details normally unlike the ML algorithm. Of course we don't lock people in boxes with a stack of training set photos for a period of time equivalent to ML training.