Newsgroups: comp.ai,comp.ai.philosophy,comp.ai.games,comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!vlsi_lib
From: vlsi_lib@netcom.com (Gerard Malecki)
Subject: Re: Loebner Prize $2000 and a medal
Message-ID: <vlsi_libD6Kx7q.BGo@netcom.com>
Reply-To: shankar@vlibs.com
Organization: VLSI Libraries Incorporated
References: <3ls4nh$qkn@hopper.acm.org> <cmckin.309.000F6593@mbnet.mb.ca> <3lutqk$5nf@acmex.gatech.edu>
Date: Wed, 5 Apr 1995 20:35:50 GMT
Lines: 37
Sender: vlsi_lib@netcom18.netcom.com
Xref: glinda.oz.cs.cmu.edu comp.ai:28822 comp.ai.philosophy:26589 comp.ai.games:828 comp.ai.nat-lang:3160

In article <3lutqk$5nf@acmex.gatech.edu> gt4767a@prism.gatech.edu (Kramer) writes:
>In article <cmckin.309.000F6593@mbnet.mb.ca>,
>Christopher McKinstry <cmckin@mbnet.mb.ca> wrote:
>>As to the 'x' on multi lines... It is a fair question. If a human can see an 
>>'x', so should the ai be able to.
>
>I think that humans have an unfair advantage, even in normal "typewritten"
>communication, as long as we use text "windows" of any sort. (i.e. CRT
>monitors, X-windows, or whatever) With these devices, we can see groups
>of characters and easily recognize such patterns. We're not looking at the
>individual characters, but their relative positions on the screen. The
>text window is oriented visually to accomodate our usual assumptions
>about such concepts as up, down, left, and right.
>
>A text-based program
>reads in characters one-at-a-time, and strings them together into tokens,
>and the tokens get combined into patterns (often sentences). The equivalent
>situation for humans would be a little LCD display that shows one character
>at a time (including spaces, punctuation, and carriage-returns). The display
>might be triggered to go to the next character in the sequence by pressing
>a button (a getchar() function).
>
>This just underlines the fact that human reading is _visually_ based. While
>you are reading this message, your eyes are scanning a few characters ahead.
>This process is unconscious and allows you to better understand the words
>that you're reading. We are accustomed to seeing our text-based comm. on
>a "page" consisting of one or more long lines of characters. At least, this 
>format is prevalent in Western languages.

Agreed that the human eye does not scan text strictly serially. However,
there is no reason why the same cannot be done by an AI program that,
once it reads in the text file, forms an internal 2D representation of the
same, like a video buffer, and uses an 'eye' to scan the 'image' in 
whichever way it wants.

Shankar Ramakrishnan
shankar@vlibs.com
