For diversity and inclusion in AI papers, they bring up facts like "All of the top 10 most-cited AI researchers are men" and then posit this as bad, and in support of a 50%-50% gender representation in AI. Like in: The field still has a long way to go. Not: we judged these extremely rare geniuses on their gender, and when you do this, you get bad diversity-and-inclusion optics.

And how to solve that perceived problem? Judge other academics on their gender, place them in a victimized group, and then compose your reference list with that in mind? The top 10 starts doing exactly that, because the guilt they feel is real. So they say nothing when privileged researchers can use the LatinX-in-AI, Women-in-AI, Black-in-AI, Gay-in-AI, Jewish-in-AI backdoors to get in a prestigious AI conference publication.

Variation hypothesis actually makes sense from evolutionary point of view: you need as many women as you can get to produce offsprings, but only small number of men suffices to impregnate them all. So men are disposable, therefore they are where the mother nature does its risky experiments.

This may be true, but I suggest you keep this to yourself. We also see that college majors with the highest average IQ (like Physics, Engineering) have the least women: http://www.randalolson.com/2014/06/25/average-iq-of-students...