Now that AI are starting to make moral decisions, programmers may have to decide which morals.
MIT Tech Reviews talks about applying a moral study to Machine Learning decisions like how to drive a car during an accident. If there is no other paths, should the AI aim for a large crowd of people or a single pedestrian. If it is single pedestrian, should you aim for larger or smaller person? Older or younger? These are definite moral decisions the developer will have to program into the AI.