The Evolutionary Morality of Artificial Intelligence

Abstract

Evolution has been a constant source of inspiration to solve computational problems. In this paper, I use evolution to address issues of artificial morality instead of algorithmic computation. While understanding morality as a social software, I develop a theory of evolutionary morality of artificial intelligence (EMAI) and articulate neural norms for selfish algorithms and reciprocal altruism. I argue that if we consider algorithms as Artificial Moral Agents (AMA) then we cannot ascribe anthropocentric moral theories to them. The morality of conscious AMA would be a product of their evolution.