> This is also why using raw machine learning for automatic driving is a terrible idea. Agreed. Train a billion-parameter model on million-feature sensor inpu

> This is also why using raw machine learning for automatic driving is a terrible idea. Agreed. Train a billion-parameter model on million-feature sensor inputs from "normal" conditions, and it will drive really well until its massive over-fitting runs into slightly unusual circumstances, like a tumbleweed rolling across the road. Then it will do something completely unpredictable, and people will probably die. ML plus pervasive surveillance can automate a lot of routine work, but it has a serious problem with outliers.