Joho the Blog
|
|
|
June 19, 2003
Here's Ray Kurzweil on the nature of the self, in his generous and excellent article on Wolfram's A New Kind of Science:
From this Kurzweil seems to conclude — fallaciously — that the self is merely formal. That is, the substance is irrelevant. Therefore, other stuff with the same form is just as much the self. Thus, strong AI is possible. The fallacy is in thinking that if the stuff of X changes, all that counts is what remains constant, i.e. the pattern, and the pattern could be moved onto new types of material which would be an X just as real and fully as the original. Here's a counterexample: A restaurant is constantly bringing in, cooking, and serving new food-stuff. All that remains constant are the patterns: the menus and the recipes. Therefore, if we instantiated the same menus and recipes in non-food stuff, it'd still be the same great restaurant. (Remember, Kurzweil isn't talking about re-creating the pattern of person X in flesh but in an entirely different medium, a program running on silicon, where the pattern is actually even more abstract than a 1:1 relationship.) I'm perfectly happy to say that life is an emergent property of our carbon-based molecules. Consciousness, too. I am not an essentialist who thinks that somewhere there's a soul, ghost or life force that exists independent of the body. (Actually, I'm agnostic on the topic, mainly because a universe in which my wife's soul doesn't continue is not only unjust, it's just plain stupid.) But emergent properties often (always?) inhere in that from which they emerge: if, say, democracy emerges from the interaction of free individuals, you can't say that the same pattern when expressed on paper is a democracy; you need the free individuals for that. I can put this more simply: We are bodies. Flesh rulz. By the way, I'm going to go to the first day of Wolfram's 3-day conference on his new kind of science. Posted
by D. Weinberger at June 19, 2003 09:51 AM
TrackBackListed below are links to weblogs that reference Kurzweil on Self:
» Succint from Mercurial Tracked on June 19, 2003 11:46 AM
» Something that it is to be we... from Versions of Truth — blog Tracked on June 23, 2003 08:36 AM
» Scaffolding for the Church from Movable Theoblogical Tracked on June 27, 2003 12:42 PM
» Stray Thought from AKMA’s Random Thoughts Tracked on August 9, 2004 10:03 AM |
Comments
You make a good argument that flesh rules, or at least I agree with you that embodiment of some kind is the crux of our identities. :-)
I want to add that our future has never been in silicon, even though before the limits of silicon were realized, it seemed that way. The Future, even an instantiation of Kurzweils vision, is in carbon. Please read my blog entry on the subject.
Therefore, I think the reality of future intelligence will be a symbiosis between software (as Kurzweil describes a bit fallaciously), and hardware (i.e augmented brain structures via nanotechnology).
Posted by: Paul Hughes | June 19, 2003 01:40 PM
Great commentary, however, I have a slightly different interpretation of what Kurzweil is trying to say given the quote above. I do not see that he is arguing that the pattern is more important than the substance, but that it is the pattern and the substance that are important. Take water for example. You can duplicate the pattern of water with other substances but you won't end up with something you could call "artificial water".
The "what" of a thing is the pattern that emerges out of how the base substance(s) from which it is made interact with one another. As each substance interacts differently, the emergent patterns will be unique depending on what those substances are. Human intelligence is a unique emergent pattern that silicon-based substances simply cannot reproduce.
Posted by: James Snell | June 19, 2003 03:37 PM
Very interesting posting and comments. Bucky is one of my heros, and whether or not Kurzweil directly credits Fuller, or has even read this material from him, I bet Kurzweil admires Bucky as much as I do. Great link though.
I still don't understand the crux of the arguments against strong AI. Flores and Winograd in Understanding Computers and Cognition seem to be saying that there is something special in the lineage of living systems that cannot be achieved in synthetic systems. But following fuller, the essence of this particular "pattern integrity" is something related to emergence through evolutionary linking of system and environment; by what argument do you exclude this evolution based pattern of development and emergence from all possible complex synthetic systems?
Even so, it remains an open question whether you could ever "read" enough of the structure from a living brain, to be able to copy this to another substrate. Frankly, I doubt it on both practical and philosophical grounds. But machine intelligence, absolutely, but is it likely to be familiar or alien? Will it have a soul?
I'll be certain to check back about the Wolfram Conference. His principle of computational equivalence is significant, but I think far from fully developed. If I may speculate, the computational pattern complexity of a whole range of biological and formal systems do constitute a pattern integrity in the Fuller sense, but something is still missing. The missing part is the tuned evolutionary mechanisms that ride on top of this computational layer, and the possiblities of higher layers of order terminating at a Unity that you may as well call God.
Posted by: Gerry | June 23, 2003 01:23 PM