zarnuk@caen.engin.umich.edu (Paul Steven Mccarthy) (04/08/90)
>(Fergal Toomey) writes: >Perhaps I should have chosen a better example. The novice+Kasparov >idea was originally intended to parallel the computer+programmer idea >in computer science. If you argue that the system under consideration >is not just the novice, but the novice *plus* Kasparov, then what can >we say about an apparently intelligent computer program? Must we say >that it is not the computer that is intelligent, but the computer >*plus* its programmer? How about Kasparov *plus* his tutors, parents, collegues, forebearers, etc... ? > [... modified example with a "programmed" novice -- cannot think, > but can remember every possible line of play as dictated by > Kasparov... Novice can still win at chess, but cannot _explain_ > why/how it was done... Conclusion: no ability to explain => no > understanding.] Two points before proceeding: 1. If you require the system to be able to explain its actions to establish your belief in its understanding, you are using a _functional_ definition of understanding. (Which is fine by me. :-) 2. Although many people use this as a criterion for "true" understanding, it seems like a pretty weak criterion to me. Even your "brute force" algorithm could have some extra code added which would maintain lists of the reasons behind the various decisions that it made. It's just extra overhead (although psychologically soothing) -- it doesn't improve the decisions that are made. It only makes the decisions visible. >My point is that the system "novice+list" has no understanding of the >game, although it can beat you every time. My point (happy to differ :-) is that the "novice+list" system has a better understanding of the game than I do if it can beat me every time. >The list corresponds to >a computer program, the novice to a computer. Saying that we must consider >the "Kasparov+novice+list" system instead is equivalent to saying that we >must always consider the "Programmer+computer+program" system instead of >the "computer+program" system. This argument is demonstably false, in >that chess-playing programs have been written which consistently beat >their authors. If it is demonstrably false, then you have not supplied an adequate demonstration. How is "programmer+computer+program" supposed to be compared with "programmer"? I would like to thank you for bringing it up, however, since it sheds a nice light on viewing the "computer+program" system as understanding chess better than the author of the program. BTW I have no problem expanding the focus of attention to "Kasparov+novice+list" or "programmer+computer+ program" ... in fact I would extend them even further! Knowledge is _cultural_. Kasparov was not born playing chess at the international grand master level. Somebody had to _teach_ him how to play. After that, I'm sure he spent a lot of time studying classic matches of former masters. Sounds vaguely reminiscent of extended book openings for computers... > Clearly these programs were not simply programmed to play >"brute-force" chess, they were deliberated programmed to analyse the >game, to think up new moves, ... to "understand". Oh? While I may be inclined to share your hypothesis, I don't see sufficient evidence to draw a firm conclusion. I have written a chess program myself which I can beat blind-folded when it runs on my microcomputer, but I imagine it would "blow me off the board" if I could get my hands on a Cray. > Unfortunately, >there is no way of distinguishing between a "brute-force" program >and an "understanding" program if you consider only their behaviour: My point exactly. >you have to look at the algorithms themselves. If two systems perform comparably, one that uses a "brute-force" approach and another that "analyzes" the game, are you willing to claim that the latter "understands" the game, while the former does not -- even though the systems perform at the same level? Personally, I prefer a more pragmatic approach -- it's results that count. Everything else is just marketing. >>Certainly the algorithms employed by a system affect the level of >>"understanding" that the system can display, but the idea that there >>is a "magical" algorithm that represents "true" understanding, while >>all other algorithms represent ignorance is absurd. Even the idea >>that there is a fuzzy boundary between algorithms which display >>"understanding" and algorithms that do not is highly suspect. >IMO, there is a fuzzy boundary between the two, and I do not see >at all why this should be highly suspect. People make all kinds of >distinctions between algorithms already, for example between >"brute force" algorithms and "insight" algorithms. Why shouldn't >we be able to say: >"This computer has been programmed to play chess, and this computer >has been programmed to *understand* how to play chess (and in fact >the first computer has beaten Kasparov, but the second computer >has given Kasparov some hints to improve his game next time)." Ah, I don't think Kasparov (bow! ye mortals) would be Kasparov if he didn't learn from his own mistakes, and unfortunately, I doubt that significant amounts of "new insight" would be derived from a system that couldn't beat him. >>To my mind, there can only be a _functional_ definition of >>"understanding". How do you know that Kasparov uses a different >>algorithm for playing chess than the computer? Better yet, how >>do you know that Kasparov "understands" chess -- except by virtue >>of his behavior? >I don't. Thank you. >... Before the arrival of computers, the question would never have >been asked, everyone simply assumed that other people thought in the same >way. I can't open up Kasparov's head and have a look at his algorithms, >and until I can, I'll simply choose to give him the benefit of the doubt. But, oh no! he might be using a "brute force" algorithm! I would not like to be the one that had to tell Him (bow! ye mortals) that he didn't "understand" chess. (I liken it to telling Nelson Rockefeller that he didn't "understand" how to make money! :) Honestly though, it is not being strictly rigorous to make assumptions about system A, and then prove that system B doesn't cut the mustard -- just because we have the advantage of peeking inside of system B and criticizing what we find. Peeking inside of system A? ... Don't be surprised if some of those assumptions you made turn out to be false! There are already chess-programs that play at the national grand master level. As long as we don't have "heads-in-the-sand" international grand masters, I predict that we will have a computer playing at the international grand master level within two decades. Probably with specialized hardware for implementing "brute-force" algorithms. I have trouble "understanding" how (second prediction) people will still claim that the system which defeated the reigning world champion doesn't "understand" how to play chess. --- Paul... (P-K4, P-QB4 -- Sicillian Defense...)
jep@mtiame.oz (Jesper Peterson) (04/09/90)
A left field question from someone not involved in AI: Is the ability to apply _understanding_ of one system to a similar system a feature of understanding? e.g. Brute force program v. Kasparov in game of chess. The judge says for this match only, the board will be 8x9 squares. Kasparov undoubtedly uses _brute force_ for his openings but will now have to rely on _true_ understanding. The program is hopelessly lost and dumps core. Perhaps *I* don't understand :-) Jesper. -- -------------------------------------------------------------------------------- ACSnet: jep@mtiame.mtia.oz "Prediction is very difficult, UUCP: ...!uunet!munnari!mtiame.oz!jep especially of the future." PHONE: (03) 699-1022 - Niels Bohr (Danish physicist)
zarnuk@caen.engin.umich.edu (Paul Steven Mccarthy) (04/09/90)
In article <201@mtiame.oz> jep@mtiame.oz (Jesper Peterson) writes: >Is the ability to apply _understanding_ of one system to a similar >system a feature of understanding? > >e.g. Brute force program v. Kasparov in game of chess. The judge says for >this match only, the board will be 8x9 squares. Kasparov undoubtedly uses >_brute force_ for his openings but will now have to rely on _true_ >understanding. The program is hopelessly lost and dumps core. > >Perhaps *I* don't understand :-) Good point. However, 8x9 boards are not chess. (And changing the board even in this apparently cosmetic manner makes fundamental differences in the applicable strategies and tactics of the game.) But as long as we're studying the "variations" (couldn't resist), how would an "insight" program perform any better? --- Paul... (o-o-o -- Queen's side castle)