logo资料库

人工智能的军事应用 Military Applications of Artificial Intelligence.pdf

第1页 / 共224页
第2页 / 共224页
第3页 / 共224页
第4页 / 共224页
第5页 / 共224页
第6页 / 共224页
第7页 / 共224页
第8页 / 共224页
资料共224页,剩余部分请下载后查看
C O R P O R A T I O N FORREST E. MORGAN, BENJAMIN BOUDREAUX, ANDREW J. LOHN, MARK ASHBY, CHRISTIAN CURRIDEN, KELLY KLIMA, DEREK GROSSMAN Military Applications of Artificial Intelligence Ethical Concerns in an Uncertain World
For more information on this publication, visit www.rand.org/t/RR3139-1 Library of Congress Cataloging-in-Publication Data is available for this publication. ISBN: 978-1-9774-0492-3 Published by the RAND Corporation, Santa Monica, Calif. © Copyright 2020 RAND Corporation R® is a registered trademark. Cover: Drones: boscorelli/stock.adobe.com Data: Anatoly Stojko/stock.adobe.com Cover design: Rick Penn-Kraus Limited Print and Electronic Distribution Rights This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited. Permission is given to duplicate this document for personal use only, as long as it is unaltered and complete. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial use. For information on reprint and linking permissions, please visit www.rand.org/pubs/permissions. The RAND Corporation is a research organization that develops solutions to public policy challenges to help make communities throughout the world safer and more secure, healthier and more prosperous. RAND is nonprofit, nonpartisan, and committed to the public interest. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors. Support RAND Make a tax-deductible charitable contribution at www.rand.org/giving/contribute www.rand.org
Preface The research in this report was conducted over the course of one year, from October 2017 to September 2018. The completed report was originally delivered to the sponsor in October 2018. It was approved for public distribution in March 2020. Since the research was completed and delivered, new organizations have been created and important steps have been taken to address many of the topics the report describes. As a result, this report does not capture the current state of the topic at the time of publication. Although expert and public opinions may have shifted, we believe the report documents a useful view of perspectives. The field of artificial intelligence (AI) has advanced at an ever-increasing pace over the last two decades. Systems incorporating intelligent technologies have touched many aspects of the lives of citizens in the United States and other developed countries. It should be no wonder then that AI also offers great promise for national defense. A growing number of robotic vehicles and autonomous weapons can operate in areas too hazardous for human combatants. Intelligent defensive systems are increasingly able to detect, analyze, and respond to attacks faster and more effectively than human operators can. And big data analysis and decision support systems offer the promise of digesting volumes of information that no group of human analysts, however large, could consume and helping military decisionmakers choose better courses of action more quickly. But thoughtful people have expressed serious reservations about the legal and ethical implications of using AI in war or even to enhance security in peacetime. Anxieties about the prospects of “killer robots” run amok and facial recognition systems mistakenly labeling innocent citizens as criminals or terrorists are but a few of the concerns that are fueling national and international debate about these systems. These issues raise serious questions about the ethical implications of military applications of AI and the extent to which U.S. leaders should regulate their development or restrain their employment. But equally serious questions revolve around whether potential adversaries would be willing to impose comparable guidelines and restraints and, if not, whether the United States’ self-restraint might put it at a disadvantage in future conflicts. With these concerns in mind, the Director of Intelligence, Surveillance, and Reconnaissance Resources, Headquarters, United States Air Force (USAF), commissioned a fiscal year 2017 Project AIR FORCE study to help the Air Force understand the ethical implications of military applications of AI and how those capabilities might change the character of war. This report, which is one of the products of that study, seeks to answer the following questions: (1) What significant military applications of AI are currently available or expected to emerge in the next 10–15 years? (2) What legal, moral, or ethical issues would developing or employing such systems raise? (3) What significant military applications of AI are China and Russia currently iii
pursuing? (4) Does China, Russia, or the United States have exploitable vulnerabilities due to ethical or cultural limits on the development or employment of military applications of AI? (5) How can USAF maximize the benefits potentially available from military applications of AI while mitigating the risks they entail? The research described in this report was conducted within the Strategy and Doctrine Program of RAND Project AIR FORCE. RAND Project AIR FORCE RAND Project AIR FORCE (PAF), a division of the RAND Corporation, is USAF’s federally funded research and development (R&D) center for studies and analyses. PAF provides the Air Force with independent analyses of policy alternatives affecting the development, employment, combat readiness, and support of current and future air, space, and cyber forces. Research is conducted in four programs: Strategy and Doctrine; Force Modernization and Employment; Manpower, Personnel, and Training; and Resource Management. The research reported here was prepared under contract FA7014-16-D-1000. Additional information about PAF is available on our website: www.rand.org/paf This report documents work originally shared with USAF on September 27, 2018. The draft report, issued on October 10, 2018, was reviewed by formal peer reviewers and USAF subject matter experts (SMEs). iv
Table of Contents Preface ............................................................................................................................................ iii Figures........................................................................................................................................... vii Tables .............................................................................................................................................. x Summary ........................................................................................................................................ xi Acknowledgments ........................................................................................................................ xix Abbreviations ................................................................................................................................ xx 1. Introduction ................................................................................................................................. 1 Background ................................................................................................................................. 2 Purpose and Scope ...................................................................................................................... 5 Methodology ............................................................................................................................... 6 Organization ................................................................................................................................ 7 2. The Military Applications of Artificial Intelligence ................................................................... 8 What Is Artificial Intelligence? ................................................................................................... 8 Autonomy and Automation ........................................................................................................ 9 Approaches to Control and Oversight ...................................................................................... 11 Recent Progress in Artificial Intelligence That Is Driving Military Applications .................... 13 Benefits of Artificial Intelligence in Warfare ........................................................................... 15 Risks of Artificial Intelligence in Warfare ............................................................................... 21 The Need for a Closer Examination of the Risks of Military Artificial Intelligence ............... 22 3. Risks of Military Artificial Intelligence: Ethical, Operational, and Strategic .......................... 24 Stakeholder Concerns About Military Artificial Intelligence ................................................... 24 Taxonomy of Artificial Intelligence Risks ............................................................................... 29 Perspectives on Mitigating Risks of Military Artificial Intelligence ........................................ 40 Conclusion ................................................................................................................................ 48 4. Military Artificial Intelligence in the United States ................................................................. 49 Brief History of Military Artificial Intelligence Development in the United States ................ 49 Mistakes and Near Misses Lead to Caution .............................................................................. 52 Summary of Current Capabilities and Future Projections ........................................................ 53 U.S. Policies to Mitigate Risks ................................................................................................. 57 Conclusion ................................................................................................................................ 59 5. Military Artificial Intelligence in China ................................................................................... 60 Current Artificial Intelligence Systems and Plans for the Future ............................................. 60 Intelligence Development ................................................................................................... 70 Chinese Ethics and Artificial Intelligence ............................................................................... 72 Conclusion ................................................................................................................................ 81 The People’s Liberation Army’s Unique Advantages and Disadvantages in Artificial v
6. Military Artificial Intelligence in Russia .................................................................................. 83 Current Capabilities and Future Projections ............................................................................. 83 Translating the Vision to Reality .............................................................................................. 90 Ethical Considerations .............................................................................................................. 94 Conclusion ................................................................................................................................ 98 7. Assessment of U.S. Public Attitudes Regarding Military Artificial Intelligence ................... 100 Importance of Public Perception ............................................................................................. 100 Prior Surveys ........................................................................................................................... 101 Results of the Survey of U.S. Public Attitudes ....................................................................... 102 Conclusion .............................................................................................................................. 116 8. Findings and Recommendations ............................................................................................. 118 Findings .................................................................................................................................. 118 Recommendations ................................................................................................................... 125 Final Thoughts ........................................................................................................................ 127 Appendix A. Expert Interviews: Methods, Data, and Analysis .................................................. 129 Appendix B. Public Attitudes Survey: Methods, Data, and Analysis ........................................ 142 References ................................................................................................................................... 169 vi
Figure 2.3. Risks of Military Applications of Artificial Intelligence Identified in Figure 3.2. United Kingdom Framework for Considering Human Control Throughout the Figure 2.2. Potential Benefits of Military Applications of Artificial Intelligence Identified in Figure S.1. Taxonomy of Artificial Intelligence Risk ................................................................. xiii Figure 2.1. Taxonomy of Artificial Intelligence Technologies .................................................... 11 Structured Interviews ............................................................................................................ 16 Structured Interviews ............................................................................................................ 21 Figure 3.1. Taxonomy of Artificial Intelligence Risk ................................................................... 30 Life Cycle of a Weapon System ........................................................................................... 44 Figure 7.1. Public Concerns About the Lack of Accountability of Autonomous Weapons ....... 103 Figure 7.2. Public Sentiment Regarding Autonomous Weapons and Human Dignity ............... 104 Figure 7.3. Public Sentiment on Human Emotion and War ....................................................... 104 Figure 7.4. Public Opinions About Autonomous Weapons and the Likelihood of War ............ 105 Figure 7.5. Public Opinion on the Need for an International Ban on Autonomous Weapons .... 105 Figure 7.6. Public Belief That Autonomous Weapons Will Be More Precise Than Humans .... 106 Figure 7.7. Public Support for Continued U.S. Investment in Military Artificial Intelligence .... 107 Figure 7.8. Autonomous Missiles in Offensive Operations Without Human Authorization ...... 108 Figure 7.9. Autonomous Missiles in Offensive Operations With Human Authorization ........... 108 Figure 7.10. Defensive Use of Autonomous Drones Against Enemy Drones ............................ 109 Figure 7.11. Preemptive Use of Autonomous Drones Against Enemy Drones .......................... 109 Figure 7.12. Use of Autonomous Drones to Attack Enemy Combatants ................................... 110 Autonomous Weapons ........................................................................................................ 111 Autonomous Weapons ........................................................................................................ 112 Figure 7.15. Use of Military Artificial Intelligence to Identify Enemy Targets ......................... 113 Attack Enemy Targets ......................................................................................................... 113 Enemy Combatants ............................................................................................................. 114 Enemy Combatants ............................................................................................................. 115 Figure 7.19. Concerns About Exploiting Vulnerabilities in Commercial Software ................... 115 Leaders in Compromising Situations .................................................................................. 116 Figure 7.13. Use of Autonomous Weapons to Avoid Defeat When the Enemy Is Not Using Figure 7.20. Use of Artificial Intelligence to Generate Fake Videos to Show Foreign Figures Figure 7.14. Use of Autonomous Weapons to Avoid Defeat When the Enemy Is Using Figure 7.16. Use of Military Artificial Intelligence to Advise Commanders on How to Figure 7.17. Use of Biometric Analysis at Military Checkpoints to Identify Figure 7.18. Use of Biometric Analysis and Robotics to Identify and Subdue vii
Figure B.7. The United States Should Work With Other Countries to Ban Figure B.8. The Development of Autonomous Weapons Will Make the Occurrence of Wars Figure B.4. Autonomous Weapons Are Ethically Prohibited Because They Violate the Figure B.5. Autonomous Weapons Are Ethically Prohibited Because They Cannot Be Held Figure B.6. The U.S. Military’s Testing Process Will Ensure That Autonomous Weapons Are Figure A.1. Time Ranges Considered by Experts Interviewed .................................................. 137 Figure A.2. Number of Times a Benefit Is Mentioned (N = 24) ................................................ 138 Figure A.3. Number of Times a Risk Is Mentioned (N = 24) ..................................................... 140 Figure B.1. War Is Always Wrong ............................................................................................. 143 Figure B.2. Autonomous Weapons Are More Likely to Make Mistakes Than Humans ............ 143 Figure B.3. Removing Human Emotions From Decisions in War Is Beneficial ........................ 144 Dignity of Human Life ........................................................................................................ 144 Accountable or Punished for Wrongful Actions ................................................................. 145 Safe to Use .......................................................................................................................... 145 Autonomous Weapons ........................................................................................................ 146 More Likely ........................................................................................................................ 146 Figure B.9. Autonomous Weapons Will Be More Accurate and Precise Than Humans ........... 147 Artificial Intelligence Technology for Military Use ........................................................... 147 Missiles Have Human Authorization .................................................................................. 148 Human Authorization .......................................................................................................... 148 Without Human Authorization ........................................................................................... 149 Report Enemy Combatants ................................................................................................. 149 and Subdue Enemy Combatants ......................................................................................... 150 Swarm That Is Attacking .................................................................................................... 150 Autonomous Drones to Preemptively Destroy an Enemy Autonomous Drone Swarm ..... 151 Figure B.15. It Is Ethically Permissible for the U.S. Military to Use a Robot With Facial Recognition or Other Biometric Analysis at a U.S. Military Checkpoint to Identify Figure B.14. It Is Ethically Permissible for the U.S. Military to Use a Robot With Facial Recognition or Other Biometric Analysis at a Military Checkpoint to Identify and Figure B.16. It Is Ethically Permissible for the U.S. Military to Use a Swarm of Armed Autonomous Drones to Protect U.S. Soldiers from an Enemy Autonomous Drone Figure B.13. It Is Ethically Permissible for the U.S. Military to Use Missiles That Autonomously Search for and Destroy Enemy Targets in Close Proximity to Civilians Figure B.10. It Is Ethically Permissible for the U.S. Military to Continue to Invest in Figure B.11. It Is Ethically Permissible for the U.S. Military to Use Missiles That Autonomously Search for and Destroy Enemy Targets in War Zones Only If the Figure B.12. It Is Ethically Permissible for the U.S. Military to Use Missiles That Autonomously Search for and Destroy Enemy Targets in War Zones Without Figure B.17. It Is Ethically Permissible for the U.S. Military to Use a Swarm of Armed viii
分享到:
收藏