Thursday, October 15, 2009

An OT reply to a Motley Fool post

A reply to http://boards.fool.com/Post.aspx?mid=28031986&reply=true#reply

Darn it, man, I recieved my used copy of the Emperor's New Mind not two weeks ago and am already on page 14. You just made me surf all over the place on "Goedel Escher Bach strong AI" to find out what Hofstadter might think as this kind of thing matters to me very much these days. THe best I could find is "...his conviction that human-level consciousness/strong AI can appear in a machine."

I accept the Chinese Room and no longer believe you will get true AI by programming a computer. You MIGHT get fake AI, a simulation of intelligence which can do many of the things you think can only be done by someone/something intelligent. BUt they will not be intelligent, I now agree. A simulation of intelligence is no more intelligence than is Sim City a real city, World of Warcraft a real battle between monsters, or a simulation of an H=bomb likely to ever knock down a building. And classic AI is, I think, a simulation of intelligence. We keep looking to see what intelligences do, and where we can find patterns we code them up.

But can a machine have intelligence? Of course. What are we other than a wierd bioligical machine? Maybe we have wierd quantum stuff going on in our brains. But either there is a god with a personality that comes down and plugs souls into machines, in which case it is pretty much up to her mood whether an appropriately built machine gets a soul plugged into it, or it is something about the way the brain/body is built that gets it consciousness, in which case it seems we could eventually figure out how to build something like that, too.

I have to say the Chinese Room thing now seems obvious to me. A chess playing program is no more intelligent than is the calculator I use to figure my taxes. All the intelligence is in the programmer of the machine. Building real intelligence will require something very different from programming.