For others reading, there are still important biases these models pick up (for a variety of reasons I won't go into). The takeaway here is that some of the evidence you've heard about isn't really evidence, not that the biases aren't there.