How to Assign Weights to Values?

Abstract

In this paper, one of my primary objectives is to analyze why adopting particular machine-learning techniques and using a moral AI as an adviser is an insufficient condition for eradicating racist human attitudes. By outlining some difficulties in justifying what artificial “explicit ethical agents” in Moor’s sense should look like, I explore why, even if the development of machine-learning techniques can be accepted in epistemic terms, it does not follow that the techniques in question will have a positive impact in changing immoral human behavior.