Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!bcm!dimacs.rutgers.edu!aramis.rutgers.edu!athos.rutgers.edu!nanotech From: Hanson@charon.arc.nasa.gov (Robin Hanson) Newsgroups: sci.nanotech Subject: Re: Down and Out in Nanoland Message-ID: Date: 9 Jan 91 22:08:08 GMT Sender: nanotech@athos.rutgers.edu Lines: 50 Approved: nanotech@aramis.rutgers.edu In geopi@hocpa.att.com (George P Cotsonas) writes: >In article , Hanson@charon.arc.nasa.gov (Robin Hanson) writes: >> I agree that there will be a significant time lapse between developing >> nanotechnology and technologies that require substantial understanding >> of how the brain works. >then contradicts it by saying >> However, it seems plausible that "uploading" >> will only require that we have a reasonable model of the signal > ---- >> processing capabilities of neurons and synapses, an understanding we >> seem close to today. ... >I question whether the nanotechnology required to analyze, dismantle, >and record neural nets would be "not particularly advanced." I apologize if I gave the impression that the technology required for uploading would be trivial - that would be an insult to the many researchers who have worked hard on it for decades. My main point was/is that the technology for uploading would probably be *easier*, and hence come sooner, than that for human-level AI or human "partials". Robin [Perhaps a bit more detail would help ground the arguments better. It is true that, for example, the ability to copy someone's voice (the phonograph) predated the ability to synthesize one by almost a century. What about uploading? For computer-oriented types like myself, a nice overview of some neurophysiology that is germane to the question is found in Ch. 20 of the "PDP" books (McClland & Rummelhart). At an absolute minimum we must consider 10 billion neurons, with an average 1000 synapses each, which can fire at rates of a kilohertz. It'll take at least a MIPS to simulate a synapse with any claim to fidelity, so we need 10 trillion MIPS (that's 10 million tera-ops or 10^19 instructions per second) to run the simulation. This can be compared with Moravec's estimate of 10^13 IPS (10 million MIPS) for human equivalence AI style (i.e. not simulating neuron by neuron). Now AI seems to be moving slowly if you're sitting behind it in traffic, but my own 15 year's association with the field convinces me that it's moving fast enough to keep up with the machines it has to run on. With the rules of thumb 1990=10 MIPS and a decade gives 1000x computing power, we get full AI in 2010 but don't have a machine you can upload into until 2030. (Nanotech doesn't change the rules of thumb, it simply helps them stay on the curve after electronics give out.) --JoSH]