@Shamar I remember doing that thing a while back and it's pretty fucked up because, among other things, it assumes a self-driving car would somehow be able to determine a person's social standing and therefore "know" whether someone is a doctor vs a criminal vs a homeless person (not that I think we should be using those criteria to value human life).
@Shamar The way they choose to interpret the data is, as you say, completely dependent on their own moral frameworks. I just did it such that the car always kills itself & it's passengers when given the option. When forced to choose between killing 2 groups of people, I made it always just go straight, regardless of who/how many people that would kill. The results appear that I really care about saving people of high social standing, despite that not factoring into my decisions.
Indeed my own personal feeling is that the #MIT's #MoralMachine experiment is not #SocialScience, but #propaganda.
