Feats without heroes: Norms, means, and ideal robotic action

Abstract

Moral competence is an increasingly recognized challenge and goal for human-robot interaction and robotic design. For autonomous robots, the question is how they can arrive at and execute the best action in a certain context. This paper explores how a computational system could best decide and act given the practical, logistical, and cultural constraints involved. We argue that in ethically charged situations where certain forms of information are more limited than normal, a robot may use certain norms in order to adjudicate and plan an action. What is more, an autonomous robot's provisional reliance on a norm, due to the robot's distinctive abilities and lack of patience, could fulfill those norms in unusual ways. While those extraordinary aspects to the robot's action - what makes it a feat one might say - may carry associations with virtue or heroism (as these actions might be viewed if performed by human beings), the objective for computationally rendered norms is to yield the best actions in an accountable fashion.