I remember when bias in machine learning was supposed to be an existential problem. We'll see how big of a problem it is when it's the "right" kind of bias.