On Cheap Artificial Consciousness

A natural way to determine whether an artificial system is conscious is to consult our best theories of consciousness and check whether the system meets their conditions. The problem is that these conditions seem far too easy to meet and could be realized in systems nobody would think are conscious. This is the problem of cheap artificial consciousness. It indicates either that our theories are all wrong, that functionalism is wrong, or that consciousness comes cheap. I argue that none of this follows. The problem results from a confusion between core and total realizers of a function—a distinction I clarify. Whether a system is conscious depends not only on its instantiating the core realizer of consciousness but also on a set of enabling conditions. I show that identifying the enabling conditions for consciousness turns on philosophical debates that no theory of consciousness resolves for us—such as the mark of the mental, or the nature of qualitative character. The upshot is that even the true theory of consciousness would not tell us which artificial systems are conscious so long as the relevant philosophical debates remain unsettled.