This is a bit of a tangential ramble on why diversity might be kind of a good idea.

Different evidence accrues to people with different experiences.

A Bayesian agent who goes through an upbringing as a boy and one who goes through an upbringing as a girl will probably not possess identical beliefs about society, the world, humanity, and so on. This is not because one has been held back or misled, nor because one is less rational than the other ... but because two different partial explorations of the same territory do not yield the same map.

This does not mean that "men's truth" and "women's truth" (or "European truth" and "African truth") are different truths. Nor does it mean that any map is just as good as any other. Some people really do sit down and scribble all over their map until it is useless.

But since nobody's map is equivalent to the territory, overall we can expect that we will navigate the territory better if we can get help from people whose maps are different from our own.

That means that if we spend our time hanging out only with people whose experiences are a lot like our own, and going all Robber's Cave on anyone whose map doesn't look like ours, we are probably going to end up kinda ignorant. At the very least we will not have as complete a picture of the landscape as a group who has shared maps from lots of different paths.

This matters if we care about possessing accurate maps; and it also matters a great deal if what we are trying to map includes things like "the good of humanity" or "coherent extrapolated volition of humankind" or things like that.