SUPER-INTELLIGENT MACHINES SUPER·I NTELLIGENT MACHINES Bill Hibbard University of Wisconsin at Madison Madison, Wisconsin Springer Science+Business Media, LLC Library of Congress Cataloging-in-Publication Data Hi bbard, Bi ll. Super-intelligent machineslBill Hibbard. p. cm. lncludes bibliographical references and index. ISBN 978-1-4613-5227-3 ISBN 978-1-4615-0759-8 (eBook) DOI 10.1007/978-1-4615-0759-8 1. Artificial intelligence. 2. Computers. 3. Machine learning. I. Title. Q335 .H53 2001 006.3-dc21 2002073080 ISBN 978-1-4613-5227-3 ©2002 Springer Science+Business Media New York Origina11y published by Kluwer AcademiclPlenum Publishers in 2002 Softcover reprint of the hardcover 1s t edition 2002 10987654321 A C.I.P. record for this book is available from the Library of Congress Ali rights reserved No part of this book 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 PREFACE We all see the acceleration of technological change, and many of us feel it in our lives. Will it eventually slow down, or will it continue until it drives us all mad? This book is my prediction of how technological acceleration will eventually resolve itself in a fundamental change in the nature of human life. We are only a few generations away from an event with impact equal to the first appearance of humans on earth. The idea of the event will seem distasteful to many and the actual event certainly poses real dangers. But it will also be a great opportunity. If done right, it will be a real blessing to those who experience it. I wish I could be one of them, and am writing this book as an indirect way to participate in the event. Richard Wagner was an anti-Semite admired by the Nazis. But many decent people admire his operas for their musical beauty and drama, and I have chosen to illustrate my ideas by their symmetry with his operas. If this offends you, please forgive me. Many of the notes in this book include references to World Wide Web pages. It is possible that some of these pages may no longer be available when you want to view them. However, you should be able to find all of them in the WayBackMachine web archive at http://www.archive.orgl. Thank you to A. 1. Hibbard, Hal Snyder, Josh Snyder, Michael Bottinger, Ugo Taddei, Grant Petty, Bob Norton, Sandy Schink, John v vi Super-Intelligent Machines Moore, Jim Green, Dave Allen, John Benson, Bob Krauss, Brian Osborne and the editors and reviewers at Kluwer AcademiclPlenum Publishers for reviewing drafts and other help with this book. CONTENTS 1. Gotterdiimmerung ............................................................................ 1 Part I Humans Will Create Super-Intelligent Machines 2. The Basics of Machine Intelligence ................................................... 7 3. Computers as Tools ........................................................................ 15 4. Arguments Against the Possibility of Machine Intelligence ............. 27 5. The Current State of the Art in Machine Intelligence ...................... .41 6. Neuroscience .................................................................................. 79 7. Dawn of the Gods ........................................................................... 99 Part II Super-Intelligent Machines Must Love All Humans 8 Good God, Bad God ..................................................................... 113 9. Brain Engineering ......................................................................... 133 10. Current Public Policy for Information Technology ........................ 145 11. Public Education and ControL ..................................................... 155 12. Visions of Machine Intelligence .................................................... 165 vii viii Super-Intelligent Machines 13. Endings ........................................................................................ 175 Part III Should Humans Become Super-Intelligent Machines? 14. Current Connections Between Brains and Machines ...................... 185 15. Humans Minds in Machine Brains ................................................ 189 16. Humans Will Want to Become Super-Intelligent Machines ........... 193 17. Super-Intelligent Humans Must Love All Humans ........................ 199 Part IV Conclusion 18. The Ultimate Engineering Challenge ............................................. 205 19. Inventing God ............................................................................... 209 20. Messages to the Future .................................................................. 213 Bibliography ....................................................................................... 217 Index ................................................................................................... 223 Part I HUMANS WILL CREATE SUPER-INTELLIGENT MACHINES Chapter 1 GOTTERDAMMERUNG "If God did not exist, it would be necessary to invent him." -Voltaire Richard Wagner's four-opera cycle, Der Ring des Nibelungen, is the story of how the Norse god Wotan tried to save his world by creating a human hero with free will who could do things that Wotan could not.l But by creating free humans Wotan ensured that, in the final scene of Gotterdammerung, the Norse gods were destroyed by fIre in Valhalla and the world passed to human control. In reality the Norse gods were destroyed by knowledge. Or more precisely, when people began to know how nature works they ceased to believe in the Norse gods and other ancient explanations that were incompatible with that knowledge. And they had confidence in their knowledge because they could use it to control the world. There is still a place for religious belief outside of scientific knowledge. For example, science does not offer an explanation of why anything exists at all. The arbitrariness of existence is profoundly disturbing 1 B. Hibbard, Super-Intelligent Machines © Kluwer Academic/Plenum Publishers 2002 2 Super-Intelligent Machines to many people, including myself. This existential void is a motive for modern religious belief. Similarly, while science can explain how one species evolves into another, it does not yet offer a good explanation of how life first evolved from inanimate molecules. This miracle of life is also a motive for modern religious belief. God is offered as the creator of existence and of life. There is a knowledge gap between our subjective experience of our minds, and our scientific understanding of our brains. As science unravels mental behavior in terms of physical brain processes, it is difficult for some people to accept that their experience of consciousness can have a physical explanation, no matter how complex. These people fill the gap between their minds and brains by religious belief that consciousness resides in a soul outside the physical world. . But we do not require religion to accept each other's consciousness. We know that other people are conscious based on our emotional connections with them. We could say this gap in scientific knowledge is filled by emotional knowledge. Does anyone really doubt that their spouse, their children or their parents have minds similar to their own? Science gradually answers more and more of our questions. There are serious efforts to explain how life evolved from inanimate molecules, and to explain the physical basis for consciousness. Some physicists even toy with explanations for existence itself. The sense of the inevitable success of science leads many people to abandon religion altogether, and instead place their faith in science. However, a critical event in the progress of scientific knowledge is imminent. Science will provide a physical explanation and demonstration of consciousness by building a conscious machine. We will accept it as conscious because of our emotional connection with it. The fundamental instrument of creating knowledge, the human mind, will be known and subject to human control. I think this will happen within about 100 years, and some computer scientists think it will happen sooner. Shortly after this event it will be possible to construct machines with much greater than human intelligence. Our ability to understand the mind of a super-intelligent machine will be similar to our pets' ability to understand our own minds. And the knowledge gap that has been so steadily shrinking over the centuries will start to grow. Not in the sense that scientific knowledge will shrink, but in the sense that people will have less understanding of their world because of their intimate relationship with a mind beyond their comprehension. Our