u-dmfloy%ug.utah.edu@wasatch.UUCP (Daniel M Floyd) (03/23/89)
In article <1368@hub.ucsb.edu> silber@sbphy.ucsb.edu writes: >In dreaming, ... [things are different]. >I am interested in any NET-speculation re: this phenomenon and its >interpretation from the standpoint of a-i. Dreaming is a strange thought process. (I am not trying to define it explicitly, only generally at best.) With all this Total-Keyboard-Language-Understanding-Turing test going on in this group, dreaming is a welcome topic. Do computers dream? Can they? Could they? Should they? If they do do they then posess intelligence? And, prehaps more directly related to AI, if we model intelligence, particularly that of a human, must we also model the dreaming aspect of thought? I think that if we *broadly* define dreaming, a computer can dream, some do, have, and should? But, we have to define dreaming properly to follow that line of thought. I suppose that sleep disorder researchers would love to have a computer that modeled human thought patterns while dreaming - a dream simulator. Maybe we could make ELIZA dream. Wouldn't that be fun. Dan Floyd 8<D=
hillman@cpsvax.cps.msu.edu (Thomas Hillman) (03/29/89)
In article <1406@wasatch.UUCP> you write: >In article <1368@hub.ucsb.edu> silber@sbphy.ucsb.edu writes: >>In dreaming, ... [things are different]. >>I am interested in any NET-speculation re: this phenomenon and its >>interpretation from the standpoint of a-i. There are two types of dreaming: day dreaming and nocturnal dreaming. They both tie together with the idea of imagination. To be able to imagine something real or unreal is a very powerful ability. Should computers be able to imagine? What does that mean, imagine? I was recently at a lecture given by Herb Simon where the notion of visualization to perform reasoning was discussed. Here was an example that he gave: Image a rectangular box. Now draw a line from one side of the box to the other at the midpoint of the longest side. Next draw a line from one corner to another. Question: Do the two lines intersect? Can you imagine that they do? Can you prove it on paper? --Tom-
murthy@tut.cis.ohio-state.edu (Murthy Gandikota) (03/29/89)
In article <2293@cps3xx.UUCP> hillman@cpsvax.cps.msu.edu (Thomas Hillman) describes a problem Herb Simon has posed: > Image a rectangular box. Now draw a line from one side of the box to >the other at the midpoint of the longest side. Next draw a line from >one corner to another. > Question: Do the two lines intersect? Can you imagine that they do? >Can you prove it on paper? > > --Tom- If the problem is as vague, then all I could do is guess or speculate even with my 3-d geometry background. But I am sure this problem can be solved using the line equations (3-d or 2-d depending on whether the two lines are on the same plane), if the meanings of "one side to the other" or "one corner to another" are made clear (mathematically or in some formal language). If someone doesn't want to give away the meanings, then all possible combinations of sides and corners can be considered. However, if the intent is to show that semantics of natural language are inadequate for certain kinds of proofs, I agree with a reservation. This should not be a case for mental visualization. Because if you change dimensions to 4-d I wonder what a "corner" means there...and what kind of visualization happens?...I suppose only symbols make sense. On the other hand if I've to visualize mentally "The Gardens of Eden" or "Hamlet's Physical Form" then you've a case. --murthy -- "What can the fiery sun do to a passing rain cloud, except to decorate it with a silver lining?" Surface mail: 65 E.18th Ave # A, Columbus, OH-43201; Tel: (614)297-7951