A model for regulating of ethical preferences in machine ethics

Abstract

Relying upon machine intelligence with reductions in the supervision of human beings, requires us to be able to count on a certain level of ethical behavior from it. Formalizing ethical theories is one of the plausible ways to add ethical dimensions to machines. Rule-based and consequence-based ethical theories are proper candidates for Machine Ethics. It is debatable that methodologies for each ethical theory separately might result in an action that is not always justifiable by human values. This inspires us to combine the reasoning procedure of two ethical theories, deontology and utilitarianism, in a utilitarian-based deontic logic which is an extension of STIT (Seeing To It That) logic. We keep the knowledge domain regarding the methodology in a knowledge base system called IDP. IDP supports inferences to examine and evaluate the process of ethical decision making in our formalization. To validate our proposed methodology we perform a Case Study for some real scenarios in the domain of robotics and automatous agents.