First, read this article, “What Would the Average Human Do?” at The Outline. I’ll wait here.
So: it sounds scary, teaching some mindless AI program to execute an ethical routine that coldly calculates who should live and who should die, like a digital Unetaneh Tokef. But it shouldn’t, because this is exactly the same thing we do with children. We impart our cultural values and morals onto our offspring, and we hope that when the time comes, they will make the same ethical choices we would.
Yes, the ethical sampling pool in the above article may have been self-selected, but isn’t it always true that, when imparting values onto children, the pool is small? Who imparts ethical values onto children? It’s the ones closest to them: their parents, their closest friends, their teachers. Maybe a few significant persons in the community such as a rabbi, priest, or friend. All they are doing differently with AI is transferring that ethical knowledge onto an algorithm, for quick access.
The other important thing to point out is that in the milliseconds of reaction time a human being has to decide how to react in an accident, they are not pondering a long Talmudic list of ethical dilemmas, so that they may decide on the best course of action. They are slamming on the breaks, they are swerving, and reacting primarily from instinct. Yes, that instinct can be affected by one’s morals, but in general an AI performing this ethical calculation can do so much faster and in much more depth than any human could. By imparting our ethical codes into a machine, we actually make the world more ethical (and safer) than if a human was behind the wheel.
Yes, my ethical code may not overlap 100% with yours, and probably doesn’t overlap much at all with someone from ISIS, but in any society, there is usually a strong general consensus about what is considered ethical behavior. So for example, cars in Italy should have ethical sampling from mainly Italians. Cars in Saudi Arabia, from Saudi Arabians. Etc. At the end of the day, these machines will just be extensions of us. There is no cold, brutal calculation here. These are our values expanded into the world.
We should welcome this.