دانلود کتاب بازدارندگی در شرایط عدم قطعیت – هوش مصنوعی و جنگ هسته‌ای

Read what has been published on the subject in the past few years, and you might get the impression that the problem of locating and targeting missilecarrying submarines and mobile missile launchers has already been solved. Paul Bracken suggests that Cyber technology is making the hunt for mobile missiles faster, cheaper, and better. This upsets nuclear stability because it opens the door to accurate strikes with conventional or nuclear weapons on the backbone deterrent systems of the second nuclear age, namely, mobile missiles.¹ Keir Lieber and Daryl Press agree that “In the ongoing competition waged by ‘hiders’ and ‘seekers’ waged by ballistic missile submarines, mobile landbased missiles, and the forces that seek to track them, the hider’s job is growing more difficult than ever before.”2 Brendan Rittenhouse Green and Austin Long concur: “The United States has invested massive resources into intelligence capabilities for a first strike, including successful innovation in tracking submarines and mobile missiles.”3 Predictions of this sort are nothing new. In fact, analogous prognostications have been made regularly for more than a century. Yet despite all the technological progress that has occurred over that time, dominant battlefield awareness somehow always seems to be just out of reach. Today, pundits suggest that artificial intelligence is the technology that will finally make this dream a reality. Is this time different? This book argues that the answer is “no” because two of the necessary enablers of situational awareness, information fusion and reasoning under uncertainty, are intrinsically hard. These tasks are not just computationally challenging, they are epistemologically and ontologically imposing as well. In plain language, this means that even if we knew the right question to ask the computer, it would take so long for the computer to find the right answer that we could be waiting essentially forever even with an arbitrarily powerful computer. Yet the bigger problem is that we cannot be sure we are posing the right question, so even if the computer finds a correct answer to the question we asked, it might not be what we needed. Adversaries can, and probably will, exploit this difficulty to thwart would-be “finders” using increasingly elaborate forms of military deception.