Preliminary Thoughts on Defining f(x) for Ethical Machines

Abstract

There is a growing literature in machine ethics attempting at creating ethical machines through AI and machine learning. Although many concerns with respect to such attempts have been raised, including the difficulties regarding the gathering of relevant contextual information as well as solving ethical dilemmas, it appears that many fundamental ethical notions have been overlooked in the implementation of normative theories to machines. This paper provides a preliminary analysis of important aspects that need to be taken into account in the attempt of defining so called ethical machines.