Army of None: Autonomous Weapons and the Future of War

Este libro es muy valioso porque alerta sobre los peligros de las armas autónomas , pero también es realista sobre la inevitabilidad de esa tecnología y su uso por parte de naciones y grupos armados en todo el mundo. Que hacer? Cómo regularlas? Es un valioso documento de Paul Scharre para discutir entre políticos, comandantes, defensores de derechos humanos y la sociedad en general.

As a former U.S. Army ranger and a current Pentagon defense expert, Paul Scharre clearly explores the strategic, legal, and ethical pros and cons of the three degrees of autonomy, i.e. sophistication and intelligence, in weapons systems.Autonomous weapon systems can be subdivided into:1) semi-autonomous weapon systems (human in the loop), e.g. the homing munitions;2) supervised autonomous weapon systems (human on the loop), e.g. the U.S. ship-based Aegis combat system;3) fully autonomous operating weapon systems (human out of the loop), e.g., the Israeli Harpy.Mr. Scharre convincingly demonstrates to his audience that artificial intelligence (AI) systems can outsmart humans in narrow tasks but fall short of humans in general intelligence. For example, the U.S.-led coalition lost several aircrafts due to U.S. Patriot fratricides during the second Gulf War. In reality, the blame for these fratricides was to be assigned to the weapon system …. and some of its human operators who lack the required critical mindset to make the right calls.Furthermore, the degree of autonomy granted to weapon systems does not exist in a vacuum. It also depends on what the (potential) adversaries of the U.S. are able and willing to do for gaining an edge on the U.S. This arm race could ultimately undermine the idealized centaur model of human-machine teaming in war.Unsurprisingly, some non-profit organizations and smaller states strive for deliberately banning fully autonomous weapons due to what they consider the accountability gap in their performance. Think for example about crisis stability, escalation control, and war termination. These non-profit organizations and smaller states consider the international humanitarian law (IHL) principles of distinction, proportionality, and other rules insufficient in the presence of fully autonomous weapons. In a nutshell, their template is the generally successful bans on land mines and cluster munitions. Alternatively, some form of regulations on the use of fully autonomous weapons could come from the countries with the most to lose from a total ban, i.e. those with the most advanced military.In the meantime, technology in autonomous weapon systems presses forward at full speed.

If the effect of artificial intelligence on warfare interests you, Mr. Scharre's book provides a comprehensive review of the subject from a practical, technological and ethical perspective. The Army Futures Command has just opened its headquarters in Austin, Texas, with a plan to ensure that the Army is prepared to fight the wars of the future. If this book is not on their reading list, it should be.Here are a few of the topics that the book covers. First, U.S. Army Ranger Scharre addresses the importance of "context" in combat by drawing on some of his experiences in combat in Afghanistan. The 7 year old girl sent by possibly hostile Afghans to recon his unit's position during an operation in Afghanistan. Under the definitions of war Ranger Scharre and his unit could have shot her, but they didn't because she looked to be about 7 years old. Would an autonomous weapon with the power to kill have made the same decision? Probably not.The book examines the question of "what is autonomy?" Not all nations define it the same way. What weapons have been built and used that could be considered autonomous. Some exist. Robots and drones - how are these being developed by the Pentagon and by private parties using off-the-shelf parts and software.Can autonomous weapons be used safely and, if they are used, what are the risks? Speed may be critical. If your opponent uses weapons that simply act faster than any human could, the likelihood is that other combatants will use these weapons as well.Does mankind "summon the demon" by building and deploying these machines? The book has a lengthy review of the parties and persons fighting to ban these weapons. If also looks at the effectiveness of various weapons bans both in the last century and over hundreds of years. Arms control has a very mixed history.One possible development may be the centaur warfighter - man plus machine. How do we control the warfighting capacities of machines with ever more capable artificial intelligence? My sense is that the range of outcomes will be tilted from "only with great difficulty" to "we can't."I was surprised by some of the information in the book, but I hadn't looked at the subject before. I found the author to be very candid, well read, and open in his presentation of the present condition of autonomous weapons. If you have an interest in the topic, I believe that the book will reward your time. Five stars.

I read widely on AI and selected this book due to its ratings and a sample. My background includes ten years as a designer of control and safety systems for complex processes. This book definitely encompasses the whole spectrum of issues about autonomous weapons, including picking up bits from many high risk industries. These end up scattered about like little nuggets. There is significant discussion of the ethics and morals of accidents, and some of the politics of controlling technology.I found this a difficult read. Though the sections appeared to break this down into specific areas, the content within each section was often a mix of material that logically could have been elsewhere. There didn't appear to be an obvious reason for the ordering of the sections. Finally, the writing style was workmanlike, which is acceptable but not helpful when such a complex issue is the subject.I gave the book three stars because it was content rich but needed better execution to achieve its mission - whatever that may have been.