Stephen Downes

I like the overall framing of this discussion of AI ethics as a 'three body problem' involving the training data, the outputs, and the users (it could well have involved a fouth body, the algorithm, and maybe more). I would have preferred that the ethical analysis moved beyond a risk analysis, however. Ethics is about more than just risk avoidance (especially when one person's risk is another person's opportunity). Best line of the piece: "I personally don't think we evolved for truth but for survival, meaning that shared stories and beliefs, rather than rational thinking, were critical to human survival." Via Miguel Guhlin.