X-Message-Number: 15918 Date: Thu, 22 Mar 2001 08:46:25 -0800 From: Kennita Watson <> Subject: Computer AI Thomas Donaldson wrote: > Finally, some time ago I raised problems about Turing machines as > models for us. One major problem was that of timing, since we are > highly parallel brains. Two people came back and said that this > did not declare Turing's ideas to be worthless. I should explain > myself here on this point: sorry, but it does cause problems. If > you want to accept a theory which suggests that in billiions of > years we can imitate the thinking of a human being neuron by neuron > for a perioid of 10 minutes, then go ahead and accept it. But in > practice the time it takes for computation will become MAJOR, > and anyone who wants to explain how human beings work "as computers" > needs to take that time into account. Moreover even imitation > needs some serious thought: our neurons are not all working > simultaneously in simultaneous steps. This makes even the theoretical > modelling of a brain with a single computer a hard problem: different > neurons will send out messages sometimes virtually simultaneously, > and affect others simultaneously too. It's not obvious how to > model such events with a single computer, even if you forget > how long the calculations would take. > I'm not sure if this has been covered, but for those new to the discussion: none of this means that computer intelligence won't work. We are making computers smaller and smaller and faster and faster (there's even been a start made on molecular computing), and we already know how to network computers so they can cooperate on solving problems in parallel (cf. the Internet). The number of neurons in a brain is finite, and the number of connections between them is finite, and chemical communication is actually slower than electronic communication, so I think a trillion networked Turing-equivalent molecular computers could do *much* better than a billion years per 10 minutes. It's a Simple Matter of Programming :-) . -- Live long and prosper, Kennita -- Kennita Watson | I vote Libertarian. | Find out why. http://www.kennita.com | http://www.lp.org/intro Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=15918