AI Consciousness, Pluralism, and Anthropocentrism

Abstract

In the current debate about AI consciousness, philosophers tend to agree that the potential for AI consciousness raises profound moral questions — for example, some philosophers think that we may imminently create systems that deserve rights similar to those of humans. But it is also widely agreed that it will be very difficult to tell whether an artificial system is conscious, so we may be ignorant of a fact that makes a profound moral difference. In this paper I offer an alternative, deflationary understanding of these issues. I don’t think there is a profound metaphysical question of whether an AI is conscious, or that we are condemned to ignorance about AI consciousness in any interesting sense. I do think potentially conscious AI systems could raise difficult moral and political challenges, but not because we are ignorant of important facts about them. The difficulties rather have to do with extending our moral and psychological thinking into uncharted waters for which it was not designed. This is particularly true if we are committed to avoiding anthropocentric bias in our ethics — and I explain why I think that even those taking rights for AI seriously are guilty of a covert anthropocentrism in their thinking.