| > how does any of that show that “minds arise from chemistry”? That's too long a story for an HN comment (which is the reason I referred you to an entire field of study) but the TL;DR is that the only reason we have to believe that minds exist at all is the I/O behavior of things that purport to have them (i.e. people) and that I/O behavior can (as far as we can tell) be completely accounted for the the behavior of neurons, which can be completely accounted for by chemistry. > Replace the person with a powerful CPU. That completely eviscerates the experiment. The whole point of the Chinese Room is that there is a conscious person inside who does not speak Chinese. Without that, the Chinese Room is just a run-of-the-mill AI. > Showing that it is not a philosophical zombie would call for some conclusive evidence showing that the phenomenon of consciousness is caused by whatever entities feature in models from today’s natural sciences—so that manipulating them in a particular way is enough to cause consciousness to magically arise. Where is your "conclusive evidence" that this "phenomenon of consciousness" actually exists? If an AI exhibits I/O behavior that is indistinguishable from a human (i.e. can pass the Turing test) then on what basis can you call one a "philosophical zombie" and not the other? > they always strike me as inelegant and needlessly contrived What is your alternative? |
Thing is, this can be explained the other way around. If neurons & chemistry were merely how conscious phenomena appear (a map of the territory), the observed outcome would not change. (Most of chemistry, physics, etc. all work equally well in that scenario, by the way, but there may be implications in other fields.)
By the way, reducing everything to I/O behaviour is also a philosophical position, I believe it’s called behaviourism.
> The whole point of the Chinese Room is that there is a conscious person inside who does not speak Chinese.
Neither does a powerful CPU/an LLM—the point of putting a slow person that doesn’t speak the language is to illustrate on an intuitive level what happens with a fast program that does the same, just in the blink of an eye.
> Where is your "conclusive evidence" that this "phenomenon of consciousness" actually exists?
You want to attribute me a claim I do not make. There is no conclusive evidence either way, and it could be impossible to obtain any (at least within the framework of scientific method). However, a theory where it does not exist has major logical flaws in my view.
> If an AI exhibits I/O behavior that is indistinguishable from a human (i.e. can pass the Turing test) then on what basis can you call one a "philosophical zombie" and not the other?
Hinges on the hard problem. If you claim consciousness does not exist, then you have your answer and I have mine, but I would object to treating it as a fact.
> What is your alternative?
I would not claim to have my own, but variants of monistic idealism as I understand them presuppose the objective existence of consciousness and go from there. I find that way we may have to magically conjure out of nothing much fewer entities and arbitrary rules, and don’t have to explain away the only phenomenon we have direct access to.