Thank you for the link, and the ideas. Obviously the whole thought process in the brain is heavily influenced by the signals it receives from the body. But if someone loses a limb, he still remains a human, for all practical purposes. If that person keeps losing body parts, he is still a human, especially if they are replaced by artificial protheses. (Certainly his behavior will change, but that is not the point).
Sure, agree, on the arm. But the timing is important, as is the degree of loss, or variation from the “human prototype”. An amputee is wired differently in the neuronal connections that formed from someone with a genetic glitch that left off an arm. I don’t suppose that either of those cases (no arm from birth, or amputation) would make a person non-human, but if you take body parts away until such point that they don’t have the sense of touch, then I think you very much have made “human” problematic as a label, just because that sense is so integral not just to interacting mentally with the world, but also to how we think.
Maybe a better example is the enteric nervous system, the dense mesh of neurotransmitters in an around you gut, that Michael Gershon calls “the second brain”. It’s what’s firing away when you have “butterflies in the stomach”, and other strong anxiety-laden reactions. I won’t bother with all the dependencies and interactions, which you may already be familiar with, but the chain nets out to this: if you are going to think like a human (as silicon), you have to feel like a human does, and this is impossible in the casing of 4U server rack. You have to be hooked up to a “gut” (among many other things) to have the kind of chemical and electrical interactions that would approximate how humans think as a mix of logical, emotional, and other physiological factors.
Which is just to say that as we make (slow) progress toward truly compelling human “emulation”, we learn how important the physiology of the rest of the body is to our cognition. Bummer for us, as that makes things hundreds of times more difficult, if we are chasing ‘human AI’.
I like to concentrate on the Turing test. Suppose that one is able to conduct a sufficiently long conversation with an entity on a phone line. If the human side is unable to distinguish the other entity from a human, then it is a human. Could be built on a hardware or wetware platform, it does not matter. The only thing that matters is the performance of the activity.
Agree on the principle. I’m not convinced that the Turing Test, especially the “phone line” version of it, is a sufficiently strong test for intelligence (see Ned Block’s “Blockhead” objection, for example), but I agree that some test that fulfills the principle of the Turing Test is what matters: performance abstracted from the question of water-vs.silicon.
Suppose the human party tells a joke, and the other party transmits a laughing sound. Then the human can ask, what did you find funny about it? The other party says: “I just found it funny”. Is that an indicative that the other party is a computer? Not necessarily, since most humans don’t have the foggiest idea, why do they find a joke funny. The point is that we cannot capture consciousness, or conceptualization in a test tube. All we can do is infer these from the responses of the other party. Paraphrazing Forrest Gump: “human is as human does”.
Yes, a professor I had way back in school always held that humor would be last hurdle for strong AI and passing the Turing Test, even over tests for artistic creativity. I think he may be right, if only because humor is so mysterious and intractable for us in analyzing ourselves as humans. It’s hard to model in passable fashion what you don’t understand at the source. The famous “Lovelace Objection”, which denies that AI can truly match human intelligence in creative originality, may well be conquered long before we overcome the “Seinfeld Objection”, where a program would have to provide a passable set of reactions to listening to a Seinfeild standup routine (without the audience laugh track for cues, of course).
By the way, all this is not my idea. My concepts were very heavily influenced by Stanislaw Lem, who was the greatest thinker of our time - in my opinion.
The replacement idea is a ubiquitous one, and once it’s brought up, it seems obvious, but I hadn’t considered the strength of the “incremental replacement” idea. It really goes to the basis for many objections, I think, which is that the the “digital brain” just can’t work because it’s… radical. The incremental strategy “deradicalizes” it and makes it much harder to resist, as there’s no (clear) step where one could say “it stopped being human right there”, as there clearly is in the case of a single-step brain switch from wetware to software/hardare.
-TS
PS: Here’s my sticking point with the Turing Test. No matter how perfect or sophisticated our software
algorithms are (and that is what we typically mean by “thinking”, right – a process/heuristic/algorithm as opposed to the (name removed by moderator)ut data?), the Turing Test very quickly becomes an examination of experience. That is, if I am serious about identifying the machine on the other end of the phone line, I don’t focus on grammar, or analytics, or the algorithm at all; I focus on the person’s history, experiences, his/her
data. I believe I’m much more likely to smoke out the impostors there, as having inadequate data/history, than I am to find a problem with the algorithm – the “way it thinks” is not its Achilles’ Heel, but “what it has to think
from” is. The Turing Test gets twisted in a data-accumulation game, rather than a quest for algorithms, doesn’t it?
. Body and soul dualism doesn’t figure in the faith, or at least in my branch, we have a very minimalistic theology compared to you sophisticated types. I’m quite content that God created a physical universe and can’t see the rational for needing more. I mean I try really hard to understand why people want more, along with such things as why anyone wants or needs a proof for the existence of God 
, but it goes straight over my head.
