Artificial Intelligence: Challenges and Controversies to U.S. National Security

Artificial intelligence (AI) has recently played a central role in US public policy debates. Technologists in companies and some public officials want to declare a moratorium on AI research and development. Their concerns include the possibility that the ability of artificial intelligence to improve faster than the understanding or control ability of human controllers.

Autonomous AI technologies that match or exceed human cognitive abilities could redefine how we understand both technology and human nature, but whether such “superintelligence” emerges remains to be seen. Or I’m not sure when it will appear. In the midst of uncertainty, the United States and other nations must consider the potential impact of AI on their armed forces and their readiness for combat and deterrence. Military theorists, strategists, scientists and political leaders face at least seven different challenges in predicting how the human-machine interface will evolve in the coming decades. It will be.

First, the education and training of military professionals A revolution will come soon. The historical experience of great commanders is no longer the only source of books and articles. We “project” future commanders backwards into AI versions of battles fought by great captains, retroactively altering scenarios to “what if” or counterfactual to further challenge students and instructors You will be able to What was once called a “war game” rises to a higher level of scenario construction and deconstruction. Flexibility and agility characterize successful leaders who can master the science of AI-powered military planning, logistics, and combat. As combat techniques rely on a combination of fire and maneuver supported by accurate intelligence, AI systems produce the optimal mix of kinetic attacks supported by timely intelligence and rapid combat damage assessment. Successful political and military leaders need to think fast, strike hard, assess quickly, and rebuild for the next punch, bob, or weave. It’s boxing in virtual reality.

Second, the man-machine interface will be transformed. When current AI systems mature and compete with or surpass certain aspects of human decision-making, such as memory capacity, flexibility, and self-awareness. A subordinate commander finds themselves “reporting” to an AI system that acts as a proxy for their superior commander, and holds accountability to their superior for performance measured by yet another of his AI systems. You will lose. AI systems are increasingly linked across departments and other bureaucratic pipes, eventually cramming everyone into his one giant metaspace. Accessible only to specialist engineers and a few high-ranking generals and admirals. Contractors provide AI systems to enrich their wealth by providing service-selective AI systems for the specific needs of the U.S. Army, Navy, Air Force, Marine Corps, and other components of the Department of Defense and other national security bureaucracies. will build Consider the first decades of increased organizational “computerization” in government.

Third, the management of national and military information We will face formidable challenges from mature AI systems. AI will enable the collection of vast amounts of data compared to what is already collected and stored on government servers. Still, AI systems provide excellent tools for analyzing data collection that commanders and policymakers can use in a timely manner. Today, decision makers are good at collecting data but bad at analyzing it. AI could support more strategic dives into large databases to retrieve and collate information that can save lives or win battles. AI may also enable faster interpretation of signal intelligence and other information, allowing for faster insertion of relevant knowledge into decision-making’s OODA (observe, direct, decide, act) loop. . AI will also challenge traditions within the US intelligence community of bureaucratic protectionism. Civilian leaders in executive branches and parliaments will be able to use AI to get a clearer picture of which missions are being carried out by which agencies, and which secrets are worth protecting and for how long. deaf.

Fourth, there will be more information-related wars., in the greatest sense. As ever, nations and terrorists alike will want to obtain information for war and deny it to their adversaries. But as AI advances, future commanders and heads of state will be better able to influence their adversaries’ perceptions of reality. They will do this on two levels. One is to target government officials responsible for policy making, and the other is to sway public opinion in the target country.

AI and other influence tools are sometimes used to create fake images (a.k.a. deepfakes), such as fictional but realistic political speeches and news reports made up to deceive the public. . AI-generated deepfakes, for example, could depict images of President Biden secretly meeting with President Xi or Putin for some nefarious purpose. False scenes of combat carnage and terrorist attacks could possibly be superimposed on routine news reports. As AI-powered fake news proliferates, a believing public could be incited to massive demonstrations. And political leaders would be tangible targets for what the Soviets and now Russians call “active measures” of strategic deception, enhanced by AI technology. These include, but are not limited to, “reflex control” techniques to convince an enemy to do what they want.

Fifth and most certain: AI speeds everything up, including decisions about the use of force and the application of force to specific targets. This means that the first wave of attacks in a war against a great power could be focused on space and cyber assets. A nation denied access to real-time information provided by satellites for launch detection, command and control, communications, navigation, and targeting would immediately go back in the last century as a military powerhouse. Therefore, countries should develop space hubs by launching large numbers of redundant satellites, placing satellites in orbit to defend against ground- and space-based anti-satellite weapons (ASAT), and developing rendezvous and space-based capabilities. assets need to be protected. Close operations for reconnaissance, threat assessment, and satellite repair. Earth-to-space and space-to-earth communication links must be secured and encrypted.

If war breaks out in the AI ​​era, the US will also be challenged to defend itself in the cyber realm. Major powers already have offensive and defensive cyber capabilities and are probing each other’s information systems and supporting infrastructure under “normal peacetime” conditions. During a crisis or wartime, cyber attacks are accelerated and more intrusive, and he may disconnect one or more weapons, transportation systems, or command and control assets from his assigned command structure. Attacks on civilian infrastructure, including those by terrorists, are already common in cyberspace, and AI systems will only make them more sophisticated. Even more so than attacks on space systems, cyberattacks may leave no clear “footprint” as to the identity of the attacker.

Absorbing the initial blow and then retaliating may not be possible for a confused and time-pressed decision maker. Instead, preconfigured responses to attacks on informational “brain stem” assets must be established and approved, and some of those responses will be his AI-based, or at least AI-assisted. . This situation poses a particularly alarming challenge for nuclear deterrence. When faced with the pros and cons of nuclear war, policy makers should allow sufficient time with their advisors to weigh the alternatives and select the most appropriate option. However, the potential speed of AI-assisted attacks on space and cyber assets, and the increasing speed of kinetic attacks from hypersonic weapons, has led leaders, fearing an enemy nuclear first strike, to strike first rather than retaliate. have the possibility to choose.

Sixth, When AI advances Aiming for a more sentient, humanoid general artificial intelligence than is currently possible could, for better or worse, evolve in directions completely unforeseen by its creators. AI systems can adopt their own version of emotional intelligence and value preferences. And they may collaborate across borders to promote desired outcomes in public policy. For example, a sentient AI system could determine that nuclear disarmament is a necessary condition for the survival of a viable civilization, and act to promote its claims despite the direction of national policy makers. There is a nature. They may also work together to discredit nuclear deterrence theory and insert a digital wrench into the software gear of nuclear readiness. A more ambitious AI agenda to defend human rights and challenge authoritarian governments is not inconceivable. After all, the creators of AI systems may include overzealous globalists who view existing nations as obsolete relics of less advanced civilizations. Meanwhile, authoritarian governments are already abusing AI systems to stifle individual freedoms, such as freedom of expression and peaceful anti-government protests. AI can help identify and isolate “enemies of the state” by omnivorous surveillance of private and public behavior. Sophisticated artificial intelligence could make mass repression easier and more effective than the worst dictator ever imagined.

Seventh, a more mature IA infrastructure requires more education and training The civil service side is not only about technology, but also about the relationship between public policy and human values. Artificial intelligence, however smart it may be by its own standards, cannot be left to substitute for decisions made by responsible policy makers with at least a decent background in the arts and humanities. Leaving public policy decisions on AI to people fascinated by the latest gadgets and gaming systems, but lacking an understanding of philosophy, music, literature, religion, art, and history is (apologies to the government). Although) the ghost of Spiro Agnew) rants a neocortical narcissist armed with unprecedented political power.

In the United States, public policy affecting a vast population should not be the prerogative of self-proclaimed elites such as politicians, bureaucrats, scientists and media moguls. Discipline should also be maintained through broad public participation. Without adult oversight by a truly democratic decision-making process, elites would create and recreate their rights as inherent privileges, even if it destroyed their societies. .

The U.S. military is an elite organization, but like other elements of U.S. government, it is subject to shifts in democratic consciousness and public policy making. The American military enjoys great support and respect from the public, but Americans are an impatient nation in a future-oriented, success-oriented culture. Americans are skeptical of wars and long-term military engagements that do not produce immediate favorable results.

Advances in AI and the other technologies it enables will only raise public expectations for a quick and definitive solution to America’s wars. AI and other technologies can undoubtedly improve the performance of the U.S. military in many ways, but it is important to ensure that their training and experience are based not only on U.S. military law, but also on the better aspects of U.S. culture and history. There is no substitute for an army that has Americans, including the military, have never wavered their technological tails and taken advantage of new technologies. AI presents unique challenges in this regard, but they are not insurmountable if addressed quickly.