714

Modern generative AI is sort of John Searle's Chinese Room come to life, where he postulates being inside of a machine that can pass a Turing Test and see that it is only manipulating syntax, and has no consciousness, intentionality, or actual understanding of semantics. One big opponent of his theories are the Functional Theories of Mind. People like Hilary Putnam will say that if a machine can be build that does all the functions of a mind, then it is a mind. Anything like consciousness will come along with those functions, and if Searle's room is indistinguishable functionally from a human brain, then the room itself does in fact understand as we do.

Of course they have not exactly been proven wrong either with generative AI, because while they can pass simple Turing tests they fall apart very quickly under real scrutiny, precisely because they lack the things Searle suggested. Searle believed we needed to understand how the brain creates consciousness more before trying to replicate it with machines, because what the machines are doing now is clearly quite different. As of yet, we have not created any hardware that even attempts to do this, because we simply don't know what it is. Our conscious minds do clearly seem to do something other than "make a statistical guess at what other people might have said in this situation" though.

Modern generative AI is sort of John Searle's Chinese Room come to life, where he postulates being inside of a machine that can pass a Turing Test and see that it is only manipulating syntax, and has no consciousness, intentionality, or actual understanding of semantics. One big opponent of his theories are the Functional Theories of Mind. People like Hilary Putnam will say that if a machine can be build that does all the functions of a mind, then it is a mind. Anything like consciousness will come along with those functions, and if Searle's room is indistinguishable functionally from a human brain, then the room itself does in fact understand as we do.

Of course they have not exactly been proven wrong either with generative AI, because while they can pass simple Turing tests they fall apart very quickly under real scrutiny, precisely because they lack the things Searle suggested. Searle believed we needed to understand how the brain creates consciousness more before trying to replicate it with machines, because what the machines are doing now is clearly quite different. As of yet, we have not created any hardware that even attempts to do this, because we simply don't know what it is. Our conscious minds do clearly seem to do something other than "make a statistical guess at what other people might have said in this situation" though.