Mind / Iron
by Bryan Bergeron, Editor ;
Published Monthly By
T & L Publications, Inc.
430 Princeland Ct., Corona, CA 92879-1300
FAX (951) 371-3052
Webstore Only 1-800-783-4624
Toll Free 1-877-525-2539
Outside US 1-818-487-4545
P.O. Box 15277, N. Hollywood, CA 91615
VP of OPERATIONS
Tom Carroll Kevin Berry
Dennis Clark R. Steven Rainwater
Jaidyn Edwards Collin Berry
Daniel Albert Thomas Henry
Gordon McComb Camp Peavy
Theron Wierenga Pete Smith
David Geer Jeff & Jenn Eckert
Copyright 2014 by
T & L Publications, Inc.
All Rights Reserved
All advertising is subject to publisher’s approval.
We are not responsible for mistakes, misprints,
or typographical errors. SERVO Magazine assumes
no responsibility for the availability or condition of
advertised items or for the honesty of the
advertiser. The publisher makes no claims for the
legality of any item advertised in SERVO. This is the
sole responsibility of the advertiser. Advertisers and
their agencies agree to indemnify and protect the
publisher from any and all claims, action, or expense
arising from advertising placed in SERVO. Please send
all editorial correspondence, UPS, overnight mail,
and artwork to: 430 Princeland Court, Corona,
Printed in the USA on SFI & FSC stock.
ERVO FOR THE ROBOT INNOVATOR
6 SERVO 09.2014
Passing the Turing Test,
but Not the Sniff Test
As I'm sure you're aware, the program “Eugene Goostman” managed to fool 10 out of 30 judges in a subjective Turing Test. Although we can argue about whether the bar was set to low or
whether the judges were actually paying attention to the answers
provided by the program, it's clear that passing the Turing Test is
necessary but insufficient in approximating human intelligence.
Don't get me wrong — a textual exchange in which a computer
program mimics the pattern of a typical human is an impressive feat.
However, there's more to human intelligence than the ability to text
message. For an extreme example, consider a "sniff test" in which a robot
or program has to associate a fragrance with an emotion, past event, or
the sex of an individual.
A necessary step would be to enable the program to sense the range
of fragrances: from "fresh," "woody," and "ambery," to "leathery," and
"citrus" for masculine fragrances. Then, there's the sensor apparatus — a
bionic nose of sorts — and an associative element that associates, for
example, the amber-spicy scent of Old Spice with an elderly gentleman.
A female or teenager wearing Old Spice should trigger some sort of
cognitive dissonance alarm, in that the combination isn't an expected
There's also the issue of color. What should a program infer from the
color and design of a dress, for instance? A little black dress might be
appropriate for an evening out at a fine restaurant, but not for
McDonalds. The ability to sense and make inferences from such details
could be significant in tracking a suspected criminal or estimating the
efficiency of a UPS delivery driver.
In robotics, parallels to human intelligence aren't necessarily tied to
speech and conversation. The ability to locate a person trapped in the
rubble of a collapsed building — say, using audio, visual, and smell
sensors — can easily require at least average human intelligence. I don't
know of a Turing Test for robotics, but I do know that passing the sniff
test together with a minimal set of sensory and motor skills should be
part of the exam.
Celebration is certainly in order. However, we should be redoubling
our efforts to make robots at least a little more like humans. SV