How disingenuous is modern ML? Since it's all stats couldn't I very easily create a misleading model. I.e. let's say I have a tranny detector that takes random images from the population and labels them as yes or no. Since trannies are only one percent of the population, couldn't I just make a program that says no regardless of input and it will generalize to 99% of a random and representative population? How else can ml be misleading?