How Are Robots’ Reasons for Action Grounded?

Abstract

This paper defends the view that (non-conscious) robots’ reasons for action can only be grounded externally, in the qualitative character of the conscious affective experience of their programmers or users. Within reasoning, reasons for action need to be evaluated in a way that provides immediate non-inferential justification of one’s reasons to oneself, in order to stop a potential regress of whys. Robots devoid of consciousness and thus incapable of feeling emotion cannot process information about reasons for action in a way that is subjectively meaningful. Different types of grounding will be discussed, together with the question of relativism about fundamentality in the context of grounding. The concluding discussion will consider the case of hypothetical conscious robots with internally grounded reasons for action, arguing that it would be unethical for such robots to be created, as they would either effectively be brought into slavery or, if developing AI rather than human-centred values, would potentially represent a threat to human life.