Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!iuvax!rutgers!elbereth.rutgers.edu!prathuri From: prathuri@elbereth.rutgers.edu (Chandrasekhar .P) Newsgroups: comp.ai Subject: Re: The difference between machine and human intelligence (was: AI and Intelligence) Message-ID: Date: 21 Nov 88 00:06:00 GMT References: <484@soleil.UUCP> <4216@homxc.UUCP> <401@uwslh.UUCP> <402@uwslh.UUCP> Organization: Rutgers Univ., New Brunswick, N.J. Lines: 35 With reference to Lishka's and Mann's postings: Let us look at the problem of AI realistically: How far have we come and how far can we go. My intention is not to prove any problem as NP-complete, ie., to make a lengthy study of how difficult a problem is and not how to solve the problem(as most algorithm people do!). And, just because we have done things in the past which we thought to be impossible, it doesnot mean that we can solve all the problems. At the same time, I am not intending to be pessimistic; history has great evidence that it is the hope that drives us to achieve our goals. But how far have we come in Knowledge representation? - McCarthy's Logic, Minsky's Frames, Schank's Conceptual dependency - each has a different approach towards the problem and none of them are complete. It is humanly impossible to make all knowledge representational. A computer scientist cannot solve the problem on his own. Biology and Psychology of brain has to be understood clearly. As to date we do not know how knowledge is represented in our brain. Is it represented at more than one location? .. Or each piece of knowledge has only one representation? .. . How do we retrieve information from our brain? .. Is it by depth first or breadth first search.? Backward chaining or forward chaining or both? .. If information is stored at more than one location, what type of information is stored at more than one palce?.. I am not saying all the approaches made towards the problem till now are incorrect( And who am I to say!). All those great researchers have opened to us new doors to thinking. My feeling is that, we should make use of biology and psychology more, so that it will be less difficult to prove AI an NP complete problem. Chandra s prathuri csp