Author: Dan Homan
Date: 10:35:23 12/16/99
Go up one level in this thread
On December 16, 1999 at 12:23:28, Tom Kerrigan wrote: >On December 16, 1999 at 10:51:52, Dan Homan wrote: > >>That is like saying that it takes intelligence to solve a system of partial >>differential equations therefore algorithms that can do so are >>artificially intelligent. It is all how you define intelligence. If >>intelligence is simply the ability to solve a particular type of problem, >>then lots of things have intelligence (including your washing >>machine). > >No, I think nobody would argue that it takes intelligence to wash your clothes. Perhaps you mean any problem that *requires* intelligence for a human to solve. If such a problem is solved by artifical means, then the artifical means is by definition intelligent. Let's use this as your definition of artificial intelligence. A human can certainly use intelligence to wash clothes (identify the dirty spots, scrub them approriately, decide when they are done), but I suppose it is not required, a human could just move the clothes around randomly rather like a washing machine. Couldn't a human just follow a brute force algorithm for playing chess? That wouldn't seem to *require* intelligence, in fact the human wouldn't even have to know they were playing chess. My point is that every individual problem that we think requires intelligence (our generalized problem solving ability) could be solved by other means - so how do we decide when something is truly intelligent and how intelligent that thing is..... I think the definition of intelligence we use for human beings is a good start - I outlined it a bit in my previous post - just think that there are lots of extremely intelligent people who play terrible chess. Take someone who doesn't know how to play chess at all. By your definition they are not intelligent. By my definition, their ability to learn is what makes them intelligent - not how well they play. > >Before computers could play chess, everybody thought that chess required >artificial intelligence. Now nobody seems to care. Same goes for optical >character recognition, handwriting recognition, speech recognition, etc. All >problems that supposedly required AI--until they were solved. Now people think, >"Of course computers can recognize handwriting. That's just some algorithms." > >That's why a friend of mine once defined artificial intelligence as "something >computers aren't allowed to have." :) Computers are allowed to have intelligence.... (I think that some programs, like KnightCap, show intelligence in their ability to learn from experience and generalize that learning to new experiences) I just think we need a better definition than finding very specific solutions to very specific problems. - Dan > >-Tom
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.