Current AIs face the challenge of filtering bias out of the corpus they are trained on. As AI researchers have pointed out in, for example, predictive policing projects, the danger is mainly in propagating and automating previous biases.<p>An AI that isn't biased is likely to please Elon even less.