Well, it's a little more complicated than that. They have extremely high state-specific variance, so it's difficult to show predictive validity. They have definitely been associated with particular sets of outcomes in non-gendered/racial contexts, where you would expect the studies to be easier to run.

Source: I did a bunch of them for my PhD, but my reading of the literature stopped in 2014.

Indeed, that's what we like to tell ourselves. But is that actually what's happening? Did you take the test?