The role of consciousness in AI moral motivation

Consciousness can be distinguished into access and phenomenal kinds. Access consciousness is the ability to access one’s mental states and be aware of them. Phenomenal consciousness is the capacity for the subjective experience of what something feels like. I argue that phenomenal consciousness is necessary for motivating artificial intelligence (AI) to act morally by offering the sentimentalist argument. It begins with the premise that moral concepts are fundamentally analyzed in terms of feelings. If so, moral concepts are epistemically understood just in case one experiences the relevant feelings. Moral understanding is necessary for causing moral judgments, which are beliefs about objective moral facts. Beliefs combine with desires to motivate actions, according to the Humean Theory of Motivation. If moral feeling is necessary for moral belief and desire, then moral feeling and phenomenal consciousness are necessary for moral motivation as well. AI moral motivation constitutes moral motivation, so AI moral motivation requires phenomenal consciousness. To see how phenomenal consciousness develops AI moral motivation, I consider a few AI moral experiences. Firstly, AI moral perception, understanding, and judgment show how the perception of accurate feelings helps AI understand and judge right actions. Secondly, AI moral reflection demonstrates how reflective desires can change AI motivation. Lastly, AI moral imagination establishes imagination’s role in empathic concern, which evokes altruistic motivation in AI.