Author: Jeremiah Penery
Date: 22:43:56 11/20/00
Go up one level in this thread
On November 20, 2000 at 18:25:21, Bruce Moreland wrote: >On November 20, 2000 at 03:02:48, Andrew Dados wrote: > >>On November 19, 2000 at 20:11:59, Bruce Moreland wrote: >> >>>On November 19, 2000 at 10:25:21, Hermano Ecuadoriano wrote: >>> >>>>There has been some discussion here about holding some >>>>"exhibition" variations on the Turing test. >>>> >>>>This must be done eventually. >>>>If successful, it would be epoch-making. >>>> >>>>I think the year 2001 is fantastically apropo, >>>>promotionally speaking! >>> >>>I will look this up later, but until then, does anyone know of a good definition >>>of the Turing test? I would prefer Turing's. >>> >>>Is it posted on the web someplace? >>> >>>bruce >> >>Btw.. a computer program passed 'composer' Turing test a few years ago. A piece >>written by a program was performed against original Bach piece. All the audience >>knew was that one of the two was written by a human and the other by a computer. >>Majority of spectators decided computer's composition was genuine Bach. >> >>The program was sort of neural network fed with several hundreds of genuine Bach >>to train on.... >> >>-Andrew- >> >>P.S. I always suspected J.S.Bach was not a human...:) > >I printed out the article. It is pretty dense stuff, and it's 26 pages long, so >I haven't gotten through it yet. Which article is this? (In other words, where did you find it? :) >I would think that faking Bach would be pretty easy. To the average person, it would be almost too easy. But I think for someone who has studied Bach's music, it could be very difficult. Probably the same for chess-playing: to someone who doesn't seriously study chess (specifically computer chess), it will be far easier for a computer to pass such a test; but let it play against someone who has extensive knowledge of how computers play chess (Bob, for example, who has said he can probably pick out a computer after a few games easily enough.) and it will most likely have a very difficult time passing the test. >Imagine faking latin text. I think you could do this pretty easily, if your >audience couldn't speak Latin. You can make something that looks Latin enough, >by taking Latin text, partially digesting it, and vomiting it back out. A >friend of mine made a program that did this, and he barfed Latin all over us. The same applies here...Give this to someone who has studied Latin (perhaps seriously studied it), and the program's output may look halfway like gibberish. >This sounds similar to the Bach thing. > >bruce
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.