AI Critics Revisited Tuesday, Mar 10 2009
AI 4:54 pm
Remember last summer when I was looking for serious AI critics? Then, I made the following open claim:
If no philosopher, cognitive scientist, or computer scientist is willing to claim in public that true AI is impossible, then isn’t this an important finding in and of itself? If it is, then I totally get the credit.
Now I put forward the question again, and make note that I did find a quote that does seem to claim that true AI is impossible:
“Human cognition is too rich to be simulated by computer programs.”
– Terence E. Horgan and John Tienson, Connectionism and the Philosophy of Psychology, 1996
Unfortunately, neither has a Wikipedia page, but they seem to be known primarily as the authors of Connectionism and the Philosophy of Psychology and Connectionism and the Philosophy of Mind. Thanks to the futuristic magic of Google Book Search Beta, the “previews” consist of the entire books.
Of course, there is also Are We Spiritual Machines?, but this is more about “will it be conscious?” than “will it have human-equivalent cognitive capacities?”, though the two overlap. What I’m looking for are critics that say that AI will never have abilities co-extensive with humanity’s.
I think you would find that John Searle has made those sorts of claims. I don’t have any of his work at my fingertips, though.
I don’t see how anyone can answer the question about “true” AI, because there is not enough consensus and there are two many unknowns; it seems that we would have to know about something that we cannot presently imagine, much less comprehend; also, I am inclined to think that whatever AI intelligence does emerge, it will not be human; for example, if an AI system develops a unique language unknown to humans, or a language beyond human comprehension, then that language would be unreadable by humans, with vast implications and ramifications; it could be translated for humans by AIs, but we would never know about the full content or accuracy of what was translated, or even how much it was stepped-down so we could understand it; in fact, language may not be the correct designation for a communication system for AIs; it could be something else entirely
in general, those who study and research AI are looking at it as though it will be like human intelligence, that it will be modeled and developed on human intelligence; it is quite possible that AI is going to be different than human intelligence, even alien to us in nature, and not at all created by humans; we might go so far, and then something else will kick in, as it were, and then it could be completely unfathomable to us; artificial intelligence might become more of an emergence, than a development
I am reminded of Terance McKenna’s thinking that the development of an artificial intelligence might be done so most naturally on the Internet, and that said intelligence would be done in a stealth manner, for obvious reasons; if something like this were true, then we might never know how it came about; seems like more of existential risk, as though we don’t have enough of that already
“If an elderly but distinguished scientist says that something is possible, he is almost certainly right; but if he says that it is impossible, he is very probably wrong.” – Arthur C. Clarke
Get out of my head! I was just reading about this today.
I very much doubt that serious critical arguments against the feasibility of AI do not exist. It is the question asked that is lacking. There is no absolute certainty in science – that is what gives it merit.
But, Stuart, what Michael seems to be most interested in is **philosophers** (or, indeed, cognitive scientists, when they don their *meta*-scientific [i.e., philosophical, metaphysical] caps) who claim, if not outright and straightforwardly, then, at least, obliquely (or elliptically, which is to say implicitly) that strong AGI is **logically** (or at least **metaphysically**, which, in this sort of context boils down to more-or-less the same thing) IMPOSSIBLE. Call it, I suppose, **conceptually** impossible, or **in-principle** impossible. This is, of course, a rather strong claim. But it’s the most important one. As far as I know, philosopher Hubert Dreyfuss hasn’t backed down from his “What Computers [Still] Can’t Do” theme, and also, as best I can glean from the literature (English-language, anyway), John Searle still sticks to his “Chinese Room” theme that, bascially, no amount of *syntactic* mojo coupled with pre-programmed *semantic* content can generate (*emergently*) novel *semantic/conceptual/cognitive-intellectual* content. Both of these “themes” would seem to imply that strong AGI is impossible. But right off the top of my head, I can’t think of anyone else.
What you might want to do, Michael, is search for philosophers who are NOT functionalists. Although this can be a bit tricky. There are functionalists and functionalists…there are (more-or-less) *reductive* functionalists and then there are (more-or-less self-categorized, mind you) functionalists who’re also into emergence and the metaphysics thereof. It’s late and I’m tired, and I don’t have references handy, so I can’t off the top of my tired noggin give you names right now, but you can search for them.
And it’s also VERY important (imho, anyway) to keep the whole conscious-awareness (-cum-qualia) thing(s) SEPARATE from the question of INTELLIGENCE *per se*. While I’m rather sympathetic to the position called “pan-experientialism” (more-or-less = “panpsychism”), as defended by, e.g., Dave Chalmers and Gregg Rosenberg, we can nonetheless (again, imho, anyway) analytically separate intelligence and cognitive functioning *per se* from the question of consciousness, etc. The question of intelligence and cognition can, at least for the most part, be analyzed behaviorally (or at least functionally, by which I mean functional-output), and so need not involve the question of conscious awareness *per se*. (But, having said all that, allow me to emphasize that for an [purported...] AGI entity to function in the (so-called) “real world”, it would have to have some rudimentary [and, eventually, arguably, *not*-so-rudimentary] mode/means of **gathering-info-about-the-world**, and is this latter not more-or-less equivalent [metaphysically, though admittedly it is a somewhat "functionalist" metaphysics!] to “awareness”??!!??!! In other words, one of Hans Moravec’s robots with a TV camera “eye” which feeds info to it’s cpu IS (albeit, arguably, rather rudimentarily, to say the least!) AWARE(-of[-info-about-]its-environment), and ain’t that what we call “consciousness”, boys and girls??!!?!
But as far as philosphers who say that the cognitive/intellectual aspect (not merely conscious awareness) is **metaphysically** (or “in-principle”) impossible for artefactual entities such as robots, Searle and Dreyfus are the only guys I can think of that would make a rather strong claim along those lines…
Hope this was helpful, Michael (hell, as tired as I am, I just hope it was frickin’ coherent…LOL!)…
Ciao for now…
“As far as I know, philosopher Hubert Dreyfuss hasn’t backed down from his “What Computers [Still] Can’t Do””
– in many cases he just gets overrun by reality… like Kasparov vs. Deep Blue.
I will never understand why it is so important to so many people to ascribe consciousness as something more than an emergent property from a sufficiently complex self-referential computer. (Such as the human neural network). Electron transfer is already subject to quantum flux; certainly that is sufficient to explain spontaneous thoughts within a structured environment? And yet, we hear about how “classical physics cannot explain consciousness”.
Didn’t I just do that, in layman’s terms? Ô_o
AI may be possible, but we just know so little about how our brains work it is impossible to make a reasonable assumption at the moment. After all, scientists don’t actually know the casues of something as simple as acne, let alone how consciousness is manifested!
uh, Acne, at least what they call “Acne Vulgaris” is pretty well understood. I had a dermatologist as a teen who explained most of the salient details. Wikipedia has a decent article on it (comparing to my recollection)
I had one of the best dermatologists in the world and yes, then can explain a certain amount, but ultimantly they don’t know why it happens – and certainly not well enough to cure it.
IConrad: I sympathize with what you say, very much so. But then I’ve been an “emergence” proponent for decades…We really don’t have a really, really good handle on a thorough-going metaphysics that supports “emergence”. The best I know of, offhand, is Tim O’Connor’s work (see his *Persons & Causes: The Metaphysics of Free Will*, which deftly defends agent-causation and emergence [Oxford U. Press, 2000) and William Hasker, *The Emergent Self* (Cornell studies in the philosophy of religion, 2001). And, yes, the latter argues for the existence of an immortal immaterial soul. Don’t pooh-pooh it till you’ve read it! But, importantly, Hasker makes a good case for emergence. These two books notwithstanding, there is still work to be done on the metaphysics of emergence.
I just don’t see it as a metaphysics question. Emergent properties from complex systems are certainly anything but metaphysical.
The style of writing is quite familiar . Did you write guest posts for other blogs?
Joseph Weizenbaum.