The Turing Test: The Elusive Standard of Artificial Intelligence

276 Pages·2003·5.54 MB·English

by  B. Jack Copeland (auth.)

THE TURING TEST
STUDIES IN COGNITIVE SYSTEMS
VOLUME 30
EDITOR
James H. Fetzer, University of Minnesota, Duluth
ADVISORY EDITORIAL BOARD
Fred Dretske, Stanford University
Charles E. M. Dunlop, University of Michigan, Flint
Ellery Eells, Univeristy of Wisconsin, Madison
Alick Elithom, Royal Free Hospital, London
Jerry Fodor, Rutgers University
Alvin Goldman, University ofA rizona
Jaakko Hintikka, Boston University
Frank Keil, Cornell University
William Rapaport, State University of New York at Buffalo
Barry Richards, Imperial College, London
Stephen Stich, Rutgers University
Lucia Vaina, Boston University
Terry Winograd, Stanford University
THE TURING TEST
The Elusive Standard of
Artificial Intelligence
Edited by
JAMES H. MOOR
Dartmouth College, Hanover, U.S.A.
SPRINGER-8CIENCE+BUSINESS MEDIA, B.V.
A C.I.P. Catalogue record for this book is available from the Library of Congress.
ISBN 978-1-4020-1205-1 ISBN 978-94-010-0105-2 (eBook)
DOI 10.1007/978-94-010-0105-2
Photograph of "The Universal Electronic computer", reproduced by kind permission of
P.N. Furbank, Alan Turing's executor.
Printed an acidlree paper
Ali Rights Reserved
© 2003 Springer Science+Business Media Dordrecht
Originally published by Kluwer Acadcmic Publishcrs in 2003
Softcover reprint of the hardcover lst edition 2003
No pact of this work may be reproduced, stored in a retrieval system, or transmitted
in any form or by any means, electronic, mechanical, photocopying, microfilming, recording
or otherwise, without written permission from the Publisher, with the exception
of any material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work.
CONTENTS
Origins of the articles vii
Preface ix
1. HISTORY
B. JACK COPELAND, The Turing Test 1
AYSE PINAR SAYGIN, ILY AS CICEKLI, V AROL AKMAN,
Turing Test: 50 Years Later 23
2. INTERPRETATION
SUSAN G. STERRETT, Turing's Two Tests for Intelligence 79
SAUL TRAIGER, Making the Right Identification in the Turing Test 99
GUALTIERO PICCININI, Turing's Rules for the Imitation Game 111
3. CRITICISM
SEAN ZDENEK, Passing Loebner's Turing Test:
A Case of Conflicting Discourse Functions 121
BRUCE EDMONDS, The Constructibility of Artificial Intelligence
(as Defined by the Turing Test) 145
EDMUND M.A. RONALD, Intelligence is not Enough:
On the Socialization of Talking Machines 151
VI CONTENTS
4. DEFENSE
WILLIAM J. RAPAPORT, How to Pass a Turing Test 161
LARRY HAUSER, Look Who's Moving the Goal Posts Now 185
JAMES H. MOOR The Status and Future of the Turing Test 197
5. ALTERNATIVES
SELMER BRINGSJORD, PAUL BELLO, DAVID FERRUCCI,
Creativity, the Turing Test, and the (Better) Lovelace Test 215
GERALD J. ERION, The Cartesian Test for Automatism 241
S. HARNAD, Minds, Machines and Turing 253
ORIGINS OF THE ARTICLES
Selmer Bringsjord, Paul Bello, David Ferrucci: Creativity, the Turing Test, and the
(Better) Lovelace Test, in: Minds and Machines 11(1): 3-27; Feb 2001
B. Jack Copeland: The Turing Test, in: Minds and Machines 10(4): 519-539; Nov
2000
Bruce Edmonds: The Constructibility of Artificial Intelligence (as Defined by the
Turing Test), in: Journal of Logic, Language and Information 9(4): 419-424; Oct
2000
Gerald J. Erion: The Cartesian Test for Automatism, in: Minds and Machines 11(1):
29-39; Feb 2001
S. Hamad: Minds, Machines and Turing, in: Journal of Logic, Language and
Information 9(4): 425-445; Oct 2000
Larry Hauser: Look Who's Moving the Goal Posts Now, in: Minds and Machines
11(1): 41-51; Feb 2001
James H. Moor: The Status and Future of the Turing Test, in: Minds and Machines
11(1): 77-93; Feb 2001
Gualtiero Piccinini: Turing's Rules for the Imitation Game, in: Minds and Machines
10(4): 573-582; Nov 2000
William J. Rapaport: How to Pass a Turing Test, in: Journal ofL ogic, Language and
Information 9(4): 467-490; Oct 2000
Edmund M.A. Ronald: Intelligence is not Enough: On the Socialization of Talking
Machines, in: Minds and Machines 11(4): 567-576; Nov 2001
Ayse Pinar Saygin, Ilyas Cicekli, Varol Akman: Turing Test: 50 Years Later, in:
Minds and Machines 10(4): 463-518; Nov 2000
Susan G. Sterrett: Turing's Two Tests for Intelligence, in: Minds and Machines
10(4): 541-559; Nov 2000
Saul Traiger: Making the Right Identification in the Turing Test, in: Minds and
Machines 10(4): 561-572; Nov 2000
Sean Zdenek: Passing Loebner's Turing Test: A Case of Conflicting Discourse
Functions, in: Minds and Machines 11(1): 53-76; Feb 2001

PREFACE
In 1950 Alan Turing (1912-1954) published his famous article,
"Computing Machinery and Intelligence" in the journal Mind. This article is
arguably the most influential and widely read article in the philosophy of
artificial intelligence. Indeed, most of the debate in the philosophy of
artificial intelligence over the last fifty years concerns issues that were raised
and discussed by Turing. Turing's genius was not only in developing the
theory of computability but also in understanding the impact, both practical
and philosophical, that computing machinery would have. Turing believed
that computers, if properly designed and educated, could exhibit intelligent
behavior, even behavior that would be indistinguishable from human
intelligent behavior. His vision of the possibility of machine intelligence has
been highly inspiring and extremely controversial.
In this classic article Turing presented his well known imitation
game and predicted that about the year 2000 "an average interrogator will
not have more than 70 per cent chance of making the right identification
after five minutes of questioning" in the imitation game. Based on the
results of the Loebner 2000 contest and the accomplishments in the field of
AI, as impressive as they are, Turing's prediction remains unfulfilled.
Therefore, this is an appropriate time to reassess the Turing test. How should
the Turing test be understood in light of recently published materials by
Turing? What is the status of traditional criticisms of the test? Can the
Turing test be defended against such criticisms? What are new criticisms of
the test? Are there superior tests that might replace the Turing test? What is
the significance of the Loebner contests? Does the Turing test have a future
in AI or has it outworn its usefulness?
For fifty years the Turing test has been the elusive standard in
artificial intelligence. Should it be a standard in artificial intelligence at all?
Why hasn't it been passed? And what conclusions should we draw if it were
passed? This book contains insightful papers that address the basic issues
about the nature and viability of the Turing test. The book has the most
recent scholarship on the subject and yet provides an overview of the last
half century debate about the merits of test. The book should serve as an aid
to scholars and a guide to students. Of course, it is also intended as a tribute
to Alan Turing, a mathematician and philosopher much ahead of his time.
James H. Moor
The Turing Test*
B. JACK COPELAND
The Turing Archive for the History of Computing, University of Canterbury, Private Bag 4800,
Christchurch, New Zealand and Dibner Institute for the History of Science and Technology,
Massachusetts Institute of Technology, Cambridge MA 02139, USA;
E-mail: [email protected]
Abstract. Turing's test has been much misunderstood. Recently unpublished material by Turing
casts fresh light on his thinking and dispels a number of philosophical myths concerning the Turing
test. Properly understood, the Turing test withstands objections that are popularly believed to be fatal.
1. Introduction: The State of Machine Intelligence Circa 1950
The birth of Artificial Intelligence is usually placed at approximately 1956, the year
in which a program written by Newell, Simon, and Shaw -later named the Logic
Theorist - successfully proved theorems from Whitehead and Russell's Principia
Mathematica, and also the year of John McCarthy's Dartmouth Summer Research
Project on Artificial Intelligence, the conference which gave the emerging field
its name. However, this received view of the matter is not historically accurate. By
1956, computer intelligence had been actively pursued for some 10 years in Britain
- under the name machine intelligence - and the earliest AI programs to run were
written there in 1951-52. That the earliest work in the field was done in Britain is in
part a reflection of the fact that the first electronic stored-program digital computers
to function were built in that country (at Manchester University (the MUC, 1948)
and Cambridge University (the EDSAC, 1949». Another significant factor was the
influence of Turing on the first generation of computer programmers.
Turing was thinking about machine intelligence at least as early as 1941 (D.
Michie, personal communication, 1998). He is known to have circulated a typewrit
ten paper on machine intelligence among his wartime colleagues at the Govemment
Code and Cypher School, Bletchley Park. Now lost, this was undoubtedly the
earliest paper in the field. It probably concerned machine learning and heuristic
problem-solving. Both were topics that Turing discussed extensively during the
war years at GC & CS, as was mechanical chess. In 1945, Turing expressed the
view that a computer 'could probably be made to play very good chess' (1945:
41).
Turing's 'Proposal for Development in the Mathematics Division of an Auto
matic Computing Engine (ACE), (Turing, 1945), which was written at the National
Physical Laboratory, London, between October and December 1945, was the first
relatively complete specification of an electronic stored-program general-purpose
digital computer. The slightly earlier - and better known - 'First Draft of a Re
port on the EDVAC' contained little engineering detail, in particular concerning
.... Minds and Machines 10: 519-539,2000. 1
.,,, © 2001 Kluwer Academic Publishers.