It may be more useful to reason about the consequences of the capabilities of any form of AI, than trying to box consciousness. Somehow connected to the above,

It may be more useful to reason about the consequences of the capabilities of any form of AI, than trying to box consciousness. Somehow connected to the above, one can use "axioms" to understand and predict minds. Here are a couple of examples: Axiom of need: "For any evolved entity that relies on X for its survival, if that entity has a sufficiently advanced mind, then there is a 'subjective' perception of the need for X that is deep and outside its conscious control. I.e., a feeling or intense craving." Example of the above: X=Food -> perception of hunger X=Procreation -> Love, lust X=Survival -> Good vibes from interacting with peers X=Peers -> I can attribute "consciousness" when I see it Axiom of fear: "For any evolved entity that relies on avoiding Y for its survival, if that entity has a sufficiently advanced mind, then there is a 'subjective' perception of the avoidance of Y that is deep and outside its conscious control. I.e., a fear or intense phobia." Y=Snakes -> Fear of snakes Y=arbitrary external threat -> Fear of discord, fear of alienation. Y=alienation -> shame I can think of a few more axioms, but this is getting long so I'll omit them. The axioms themselves may or may not be correct, but just as one does in mathematics, one can take them as true and see where the chain of thought leads. For me, the axioms above imply that we can create* entities that, as appreciated from outside, will seem entirely conscious to us. And I guess this is a procedural definition of consciousness. * Via synthetic evolution, e.g. gradient descent.