36_5130@uwovax.uwo.ca (Kinch) (07/05/88)
In article <58.22CCF148@isishq.UUCP>, doug@isishq.UUCP (Doug Thompson) writes: > > > J> From: jimmyz%oak.dnet@VLSI2.EE.UFL.EDU (Anubis The Psychic Chaos > > J> What leads you to believe the human mind is not a complex computer? > J> Computers of are current time have a level of complexity no human > J> device has > J> EVER achieved before, and they are just a granule of what the > J> human mind is. > J> Or rather, the brain. THe brain is undoubetedly a complex computer, > J> but the > J> mind is a non-tangible thing. Just as this VAX 8600 I am using > J> now is a complex > J> (supercomputers and the like aside) computer, but the programs > J> I am using to > J> send this message have absolutely no physical substance. > J> I'd say there is a heck of a lot of similarity. > J> > > Agreed, there is similarity. In the same way there is similarity between > the sun and a lightbulb. We have been making bigger and better lighbulbs > for quite a while now. Can we make one just like the sun? What does "just like the sun" mean? > > Well, in this case we happen to *know* there are differences as well as > similarities. We can't assemble enough matter, at least on earth, to > build something just like the sun. Our lighbulbs, though similar, do not > use the same sort of natural processes as the sun. Maybe not but "OUR" H-bombs come pretty damn close for my likeing. > > It is not a logically sound argument to say that because our technology > is advancing it will ever reach any given goal. The evidence of > advancing technology does not *prove* anything at all. True but to ignore that evidence is rather silly. > > It is my hypothesis that there are fundamental differences between the > way organic thought operates in a human creature - that is to say, human > intelligence, and the totally logical, mathematical, effective > procedures which by definition are machine intelligence. completely agree, here here. > > To build a good human intelligence out of silicon we would have to > minimally understand human thought and intelligence quite well. This is > really one of the more itneresting parts of AI research today, because > the understanding of human intelligence is not really a "mechanical" > problem. Heck, I don't even understand *myself*! > > A good example is provided by chess programs. The intelligence required > to play chess is among the most mechanical and methodical. Machines can > play very good games of chess too. But they don't do it the same way > people do. They do it by making millions of calculations, and we *know* > that is not how people do it. People use something like intuition and > pattern recognition which we know to be quite intependent of any > numerical analysis or number crunching. Computers play chess by doing an > immense amount of arithmetic. As a calculator, the human brain is really > quite slow. Something else is going on. > > Thus a quantum leap in technology is needed, a different kind of > computer, to even begin to process data of any sort (even mathematical > data) the way the human mind processes data. Or at least a different way to program the same old computers. I think that your "quantum leap in technology" is a quantum leap in Logic. But I certainly agree that a new way to mimic thinking is needed. > > We know that any AI problem is highly dependent on input. Now the input > into the "computer in my skull" comes through my eyes and my ears and my > fingers and toes, my nose and my mouth. To call an intelligence human, > it would need the same input spectrum. Or a subset of these, many humans work on relatively small subsets of these inputs. > Of course people are working on > computer smell and tactile sensors, and you might one day mimic the > whole human sensorium, and create a mechanical copy of a human's entire > experience. You might get a computer to react just like a man to a > beautiful sunset, a starving child, or a girl in a bikini. You might get > a computer pondering ethical problems and answering questions about the > relative merits of marxism vs capitalism as a social order. You really > might one day be able to do that (though I seriously doubt it), but God > forbid that anyone would *want* to! Really! Why should GOD forbid this? > > I've confused the issue between the can do and the should do. But the > should do speaks to the can do. What is a man? What is a machine? Do you > really honestly think that the indisputable similarities add up to a > potential identity? It strikes me as preposterous, and I am searching > for a language to articulate why that is. But you obviously have not found that language. > > The very fact that making mechanical men is something we should not be > doing from a moral and ethical perspective suggests to me that we > probably can't. Another huge leap, you didn't show me that we should not be doing this from either a moral or ethical perspective, but even if you had how this shows or even suggests that we proably cant is beyond my understanding. I do not even think that they are connected. We should not be murdering other humans from a moral and ethical perspective but this obviously does not mean that we CANT. > Why is that? It's hard to be precise, but think of the > sort of human intelligence manifested as a mother holds her newborn to > her breast to suckle. Think about all the complex web of social, > emotional, political, relational and economic input into the behaviour > involved, (and there are two human intelligences to take into account in > this behaviour which is highly relational in nature) and then try to > think of a way to program a computer of any hypothetical power to mimic > it. > So are you saying that because YOU cant think of a way to program these types of behavior into a computer it means that this is an impossible task? I am not an AI expert or really an expert in anything but I think that I can spot faulty logic and poor reasoning when I see it, this is the result of my spying your posting. Dave Kinchlea CCS Program Consultant University of Wester Ontario London, Ontario, Canada.