I am not an effective altruist: Morality is not a market

> Utilitarianism actually leads to repugnant conclusions everywhere, and you can find repugnancy in even the smallest drop.

The argument here is quite agreeable, that morality cannot be reduced to a scalar, universal morals don't exist and utilitarianism is not the endgame.

But what if we trained an AI to be moral? We wouldn't even specify the objective functions , but somehow let it figure it out for ourselves. Perhaps she could understand us better than we can hope to understand ourselves.