Newsgroups: alt.philosophy.objectivism,sci.philosophy.meta,sci.anthropology,talk.politics.theory,soc.culture.intercultural,soc.couples.intercultural,alt.politics.reform,alt.politics.democrats.d,alt.activism,alt.discrimination,alt.politics.correct,alt.politics.usa.con,comp.ai.philosophy,comp.ai
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newsfeed.internetmci.com!howland.reston.ans.net!usc!news.isi.edu!gremlin!shomase!jbarnett
From: jbarnett@shomase.NoSubdomain.NoDomain (Jeff Barnett)
Subject: Re: Evidence FOR Racial Equality??
Message-ID: <DMq97I.Fsx@gremlin.nrtc.northrop.com>
Sender: news@gremlin.nrtc.northrop.com (Usenet News Manager)
Reply-To: jbarnett@charming.nrtc.northrop.com
Organization: Northrop Automation Sciences Laboratory
References: <31181a31.9028916@nntp.ix.netcom.com> <4farfl$t7k@pipe12.nyc.pipeline.com> <Pine.OSF.3.91.960207212912.32055E-100000@Bayou.UH.EDU> <4fe40k$dkn@pilot.njin.net> <4fnfat$cel@pheidippides.axion.bt.co.uk>
Distribution: inet
Date: Tue, 13 Feb 1996 18:40:29 GMT
Lines: 31
Xref: glinda.oz.cs.cmu.edu sci.philosophy.meta:24670 sci.anthropology:22035 comp.ai.philosophy:37709 comp.ai:37002

In article <4fnfat$cel@pheidippides.axion.bt.co.uk>, donald@srd.bt.co.uk (Donald Fisk) writes:

|> Getting back to my point about IQ tests being the ability to do puzzles,
|> I am confident that I could, on my own, develop a program which scored
|> genius level on IQ tests in about 5 years' full time work.   Programs
|> which could solve geometric puzzles were developed by AI researchers
|> in the 1960s (I can't be bothered to look up the original references,
|> but you'll find them at the back of any introductory AI textbook).   This
|> makes IQ tests practically useless as a means of measuring intelligence,
|> and why AI researchers don't waste their time with them, and until
|> relatively recently, used the Turing Test (assessment by interview)
|> instead.

Apropos of this: In the late 60's (early 70's) I proposed to some
friends at ARPA that we build an IQ test taker.  The idea was to have
Shakey the robot (an SRI project) waddle(roll) into a room be handed
an IQ test, use a TV camera to view it, send the images over the ARPA
net to a 100+ computers, and use a pencil to mark the answers when
the net sent them.  At the time, I thought that the AI/linquistics/
pattern-matching communities know enough to (1) parse the test questions,
(2) solve them, (3) do the necessary vision work to understand the
input, and (4) control the motor skills necessary to manipulate the
test and a pencil.

I thought that this project would be a good way to (1) consolidate the
current state of AI, (2) prove that computers where samrt by a commonly
use measure, and (3) teach folks that IQ test don't measure what they
thought was measured.  Unfortunately, we never tried to do the above.
It sure would have been a lot of fun.

Jeff Barnett
