Their example about the algorithm that predicts if you’re a criminal by looking at your face seems... poorly thought out. Note, it’s not facial recognition. It’

Their example about the algorithm that predicts if you’re a criminal by looking at your face seems... poorly thought out. Note, it’s not facial recognition. It’s prediction. They make the claim that it’s all about jury perception. Maybe it is, but far more important is that it’s likely just using an estimate of race and wealth. I’m annoyed that they didn’t talk about the bigger red flag: “90% accuracy”. Maybe they should just teach stats... Edit: pre-empting a nitpick: when I say “wealth” I mean largely as proxied by hairstyle