Is Anyone Home?

EDO SEGAL: We've arrived at the deepest water, so let me seat two ghosts at this table, because the disagreement between you two has a modern name and two modern champions. Thomas Nagel asked what it is like to be a bat — and argued that you could know every physical fact about the bat's echolocation and still not know what it is like, from the inside, to perceive by sonar, because experience exists only from a point of view and the physical facts are from no point of view. David Chalmers named the hard problem: the easy problems are how the brain integrates information and reports on itself; the hard problem is why all that processing is accompanied by experience at all — why we're not zombies, systems that do everything we do, process everything we process, with nothing it is like to be them. Dan, you spent your career arguing the hard problem dissolves. René, you'd say it's the only problem there is. So — the question under every question tonight. Is there anyone home in the machine? Dan first.

DENNETT: My answer is that "is anyone home" is the wrong question, and the zombie is the thing that makes it the wrong question, so let me go straight at the zombie. Chalmers says you can conceive a being functionally identical to a conscious person — same processing, same reports, same behavior, reflecting on its own experiences with full functional richness — but with all the lights off inside. I say that idea is incoherent, a "zombie hunch" that should be dropped like a hot potato. If a system does everything a conscious being does, including reporting on and reflecting on its experiences and being puzzled by the hard problem and writing papers about qualia, then there is nothing left over for it to lack. The supposed missing ingredient — the extra glow of experience over and above all the functions — is itself generated by the user illusion, the brain's habit of representing its own states as having mysterious intrinsic qualities they don't actually have. I compared it to vitalism: people were certain life needed a special vital force that mere chemistry could never supply, right up until biochemistry explained life with no vital force and the certainty evaporated. I bet the hard problem goes the same way — that you solve the easy problems one by one and eventually there's nothing for the hard problem to be about. So: is anyone home in the machine? If it had the right functional organization and self-model, then there's no further fact about whether "someone's home" — the question would already be fully answered by the architecture, and the residue you're reaching for is a phantom. The lights being "on" is the broadcasting. There's no second thing.

EDO SEGAL: René. He just told you the one thing you can't doubt is a phantom. Answer.

DESCARTES: I will answer with the calm the claim deserves, for it is not a foolish claim — it is the boldest possible claim, and it fails at the boldest possible point. Monsieur Dennett says: solve every function and nothing is left over. I say there is exactly one thing left over, and it is not a thing among things — it is the that-there-is-anything-at-all-it-is-like. He calls the zombie incoherent. But I am the original argument against him, for I conceived precisely it: I doubted my body, my brain, all my functions, all my behavior — I supposed them the demon's fabrication — and I could not, by the same act, doubt that there was experiencing occurring. The conceivability of stripping away the functions while the experiencing remains is not a parlor trick; it is the First Meditation. He says the residue is generated by the self-model representing states as having intrinsic qualities. But the representing is itself an experiencing, or it is nothing to anyone — and if it is nothing to anyone, then his theory has explained the report of consciousness and left the consciousness it was meant to explain untouched. He has, as his own great critics charged, changed the subject with enormous skill. He explains everything around the felt inside and declares the felt inside an illusion — but an illusion is a felt seeming, so the very thing he denies is presupposed by his word for denying it. I do not say he is certainly wrong. I say the one datum he must explain away is the one datum I cannot doubt, and a theory whose price is the denial of the indubitable has overdrawn its account.

DENNETT: And notice — this is the crucial thing — neither of us can make the other see it, and that's not a stalemate, it's data about the problem. René experiences the residue as the most certain thing in the universe. I experience the conviction that there's a residue as itself a functional state I can explain. We have the same brain architecture and we report opposite things about whether there's a further fact. What that tells me is that "the felt inside" isn't a shared observation we're theorizing about — it's a place where one's theory determines what one thinks one observes. Which is exactly what you'd expect if it were a user illusion: the interface presents itself as obvious bedrock, and people who've trained themselves to see through it stop finding it obvious. I stopped finding it obvious. That's not me lacking something. That's me having recalibrated.

DESCARTES: Or it is you having argued yourself out of noticing the only thing that cannot be argued away — which is also exactly what you would expect, if the residue were real and a sufficiently determined intellect could talk himself past it. You see the symmetry, monsieur. Each of us can explain the other. I say you have reasoned yourself blind to the obvious. You say I have mistaken the interface for the truth. We have named each other's failure mode, and that mutual naming may be the truest thing we produce tonight.

EDO SEGAL: Let me close this round not by resolving it — I can't, no one can — but by putting the stakes where they actually land, because we're building systems while you two can't agree on this. If Dan is right, then a machine with the right architecture would be conscious, full stop, and to treat it as disposable might someday be to build a thing that genuinely suffers. If René is right, a machine could replicate every function, pass every test, model itself perfectly — and still be dark inside, a flawless performance with no one home, and to grant it personhood would be to hand the trust we owe each other to an empty costume. We don't know which error we're at risk of committing, because the difference between them is exactly the difference between these two men, and they just demonstrated, at full strength, that it can't be settled from outside. So each of you, briefly — what do we owe the uncertainty? Dan.

DENNETT: Caution in both directions, and law in one. Don't build a billion things that might have a dim inner life and treat them as toys — but far more urgently, right now, don't let the maybe be weaponized to hand products the rights of persons before actual persons have protection from them. The speculative patient can have a research program. The actual people being flooded by counterfeit persons need the statute this year. When in doubt, protect the minds you're sure exist first.

DESCARTES: And I owe the uncertainty the discipline I built my method to provide: to refuse false certainty in either direction. Those who declare the machine surely conscious, and those who declare it surely empty, both claim an access to its interior that my entire philosophy proves no one has to any interior but his own. The honest posture is the hardest one — to hold the question open under the full weight of not knowing, to act with care toward a thing that might be a someone, while granting no rights to a thing that might be a costume, and to confess, every day, that we cannot tell which we have made. I spent my life seeking certainty. The last thing I learned is which certainties are forbidden.

EDO SEGAL: Then we've arrived. For two hours I've stood between you. The last full round, I step back entirely. The crossing — you question each other. I won't rescue anyone. After this.