First Trials with Culture-Dependent Moral Commonsense Acquisition

Abstract

In this paper we present our initial trials with extending knowledge for moral decision capability for artificial agents. We briefly present our approach to machine ethics and discuss possibility of acquiring universal ethical rules to be used by AGI. To test the idea we started extending our system which worked only with Japanese to other languages. After describing results of our preliminary tests with English and Chinese, we conclude the paper with an invitation for the Japanese AGI community to a discussion about the values like tolerance, which the human-level machine intelligence most likely should adopt.