X-Message-Number: 12570 Date: Fri, 15 Oct 1999 10:35:04 -0700 From: Peter Merel <> Subject: Cut & Paste Olaf Henny writes, >Conjecture: >You stack one billion of Ralph's 1MB computers into 1mm^3, the >size of a small sugar grain, and you have a one million GB >storage memory. How many of these little things can you stick >into my cranial cavity without cramming the brain, and how many >would it take to hold all the information contained today in the >world's libraries. Now if you can refine the computer-mind >interface, which has so crudely demonstrated by NASA, to make all >that information directly available to me, I would present quite >a challenge to overcome by an artificial intelligence. You're forgetting speed and effectors here Olaf. To really compete we'd need to ramp up our poor old neurons, which operate at a snail's pace. Then we'd need sensors and manipulators on the scale of foglets or else we'd be comparatively crippled and blind. But I confess I think you're on the side of the angels still. Human engineers are lazy sods. Laziness is one of our three classical virtues - the others defined by Wall being hubris and impatience. We don't like to reinvent things when we have an adequate implementation handy. Instead we adapt what we've got. It seems obvious that a human order of reasoning-intelligence isn't easy to design. We only have one example in nature, and that one is very fragile. Just a little change sends it reeling away from rationality. All attempts at constructing even the most abstract and limited simulacrum have failed. It seems highly likely then - I'd suggest a certainty, though you never know - that AI in the pulp fiction sense will first be human minds uploaded. It's simply much easier to cut and paste than to design ab initio. Perhaps the uploaded minds, vastened by new abilities, will construct classical AI. What's key, however, is we'll get there first, and any AI we construct won't have potentials we won't possess ourselves by then. There's still a chance that reasoning-intelligence isn't actually that hard to construct, that for some peculiar reason evolution usually selects against it. But I think most of us find this an implausible notion. There's hubris at work. Peter Merel. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=12570