The New Fire: War, Peace, and Democracy in the Age of AI

Download PDF

In November 2023, the Chief Digital and Artificial Intelligence (AI) Officer of the Department of Defense (DOD), Craig Martell, stated, “Technologies evolve. Things are going to change next week, next year, next decade. And what wins today might not win tomorrow.” This sense of rapid technological change and disruption echoes what Ben Buchanan and Andrew Imbrie have written in The New Fire. The authors are deeply steeped in the world of cyber and AI. Buchanan is the former Director of the CyberAI Project at Georgetown University’s Center for Security and Emerging Technology (CSET), currently Special Advisor for AI at the White House, and the author of The Hacker and the State: Cyber Attacks and the New Normal of Geopolitics (Harvard University Press, 2020). Imbrie is an Associate Professor of Practice in the Gracias Chair in Security and Emerging Technology at Georgetown University’s Walsh School of Foreign Service, a former fellow at CSET, and the author of Power on the Precipice: The Six Choices America Faces in a Turbulent World (Yale University Press, 2020). The authors bring their considerable experience to bear across 10 thematically arranged chapters in The New Fire as they examine the risks and opportunities of AI and the implications of abusing the AI “fire” in programs that claim to not only protect society but also empower autonomous systems to act against humans. It is a risk that could quickly grow out of control, especially in autocratic states.

Military and security professionals looking to explore the potential legal, ethical, and machine-learning aspects of AI will find plenty to learn in The New Fire, which is part history, part technology primer, and part policy treatise. The authors provide much to consider with their argument that AI’s critical components, data, algorithms, and computing power can be used for malign purposes—including disinformation, hacking, and weaponry—unless the AI is controlled and deliberately developed for positive goals. The book’s early chapters provide historical and technical background that helps to explain the gradual development of AI systems and how they became capable of using machine learning to accomplish increasingly complex tasks.

Chapter 5 is where The New Fire illuminates how the United States and China diverge in their AI programs and how governments choose to prioritize control, freedom, and ethical action. The future of AI, especially autonomous systems, is tied to how much control the state cedes to machines, sometimes at the expense of its citizens. For example, while the Pentagon’s Project Maven AI effort experienced opposition from civilian companies, China’s strategy of civil-military fusion compelled civilian tech companies to serve the state’s needs. Subsequent chapters focus on the directions that the United States and China take as geopolitical competitors and the style of AI that each political system follows, which influences how they will employ AI. Buchanan and Imbrie assume that autocratic regimes will have fewer qualms than democratic governments regarding restrictions on au- tonomous machines taking human life.

Buchanan and Imbrie’s exploration into hacking, machine learning, and deception may be the most useful chapters of The New Fire. Leaders interested in AI and future war may see the possibilities for autonomous deception not only for offensive operations but also for how to structure defenses by creating systems, recruiting personnel, and forming alliances that can create machine-learning tools able to detect deepfakes and defeat autonomous attacks. Future wars will certainly occur across multiple domains, and although the United States enjoys strategic advantages over its competitors, the Chinese and Russians have demonstrated their willingness to exploit areas of the U.S. democratic system in ways that can cor- rode faith in leadership, sow discord, and divert valuable resources.

Buchanan and Imbrie conclude their work by illustrating how AI programs and systems can be used for state monitoring of internal conditions and to silence dissent. China’s use of AI to strengthen its autocratic hold on internal security is one example of how AI can enable a state to gain complete information and data dominance over a population. In contrast, the United States also uses AI to defend from external threats. Other countries have the freedom to learn from the U.S. or Chinese AI applications, and many have not chosen the U.S. model.

One theme that the authors refer to repeatedly is that “[t]echnology is not deterministic; it shapes and is shaped by human choices and society,” and this perspective provides insight into the authors’ belief that core values should help shape how developers create AI algorithms. The potential application of AI to gain an advantage over competing states is where democracies and autocracies appear to have diverged, at least for the present, but the potential still exists for autonomous AI systems to be abused through a slippery slope of justifications that could start with a security crisis and end with unintended consequences (for example, riots leading to autonomous AI-enabled crowd-control systems making decisions in accordance with programming for a general situation, but lacking the nuance necessary for a graduated use of force). Democratic states are going through a period of debate on how to regulate and define AI systems within the context of personal freedoms and state power. However, autocracies appear to have invested heavily in AI with the belief that it will give them advantages against enemies, real or imagined. We can expect China, Iran, North Korea, Russia, and other competitors to use AI in gray zone, hybrid, or asymmetric warfare against the United States and its allies. How governments choose to employ AI and what parameters are in place to determine what is legal and ethical may be where autocracies and democracies create their advantages or reveal their weaknesses.

Buchanan and Imbrie have written an interesting exploration of AI and policy. Their examination of competing political systems and their choices of AI projects give readers examples of where they should consider the implications of AI policy and programs. The initial wave of excitement should give way to contemplation of how AI is pushing the boundaries of the possible and feasible for warfighting and state security. The ethics of employing lethal autonomous AI-enabled systems—land, sea, air drones, for example—should be considered within the context of Just War. The potential of AI to be used for malign purposes exists alongside the potential for unlawful use of force.

The future applications for AI, whether for autonomous weaponry, hacking, or even disinformation, are worth discussing and debating as policy issues. For the politicians, leaders, and corporate heads who will use AI to generate advantage over the competition, The New Fire provides a warning that, just like fire, AI is a tool that can burn destructively if not carefully employed. However, AI and machine learning are rapidly advancing, and as a result, AI is a tempting ally to sort through the data and the noise in an ever-tightening decision cycle. JFQ