New Frontiers of AI in War Planning: How AI Agents Transform Military Decision-Making

Throughout history, the rapid changes in geopolitical and military environments have affected decision-makers’ ability to achieve strategic or operational objectives. In a dynamic environment, slow adaptation to changing conditions can lead to catastrophic consequences. Historically, there are numerous examples where armies have suffered losses of life, battles, and even wars due to their inability to adapt to the environment. National security depends on planners’ ability to consider these dynamic changes, quickly identify gaps, seize opportunities, and keep pace to remain competitive in modern warfare.

For two important reasons, the defense sector should actively begin experimenting with Agentic AI tools in the Joint Operational Planning Process (JOPP)—a type of AI capable of independently completing a series of tasks to achieve specified complex objectives. First, compared to humans, agents have the potential to synthesize a wide range of traditional and non-traditional planning factors more quickly and comprehensively, helping to develop more thorough and objective Courses of Action (COA). Second, once a COA is selected, agents can also assist in rapidly issuing downstream instructions and commands, facilitating flat communication, and saving hundreds of man-hours in each planning cycle.

Agents are a capability that can quickly consider these changing operational conditions and help independently solve large-scale complex problems. This is different from the currently popular large language models, which rely on single prompts to perform simple, specific tasks. To create multiple dilemmas for close adversaries, it is necessary to continuously integrate capabilities across all force tools and domains, including the electromagnetic spectrum and information environment. In the Fourth Industrial Revolution, Agentic AI is a method of deploying various autonomous technologies to work collaboratively, which can perceive the environment and autonomously determine action plans to achieve established goals. Using this technology alongside human planners can produce an accelerated multidisciplinary thinking machine.

Imagine a planning unit with a multifaceted “agent” that understands geopolitical trends, global dynamics, and country policies related to conflict. It can also consider the limitations and constraints of the military across all operational domains by investigating multiple datasets. This “think tank” can also minimize the impacts of groupthink, self-interest, and counterproductive biases, bringing new opportunities and pathways to decision-makers. Kathleen Hicks, the Deputy Secretary of Defense, confirmed this viewpoint at the Pentagon’s 2023 Data, Analytics, and AI Strategy launch, stating: “From the perspective of deterrence and resisting aggression, AI systems can help accelerate commanders’ decision-making speed and improve the quality and accuracy of decisions. It is proposed that AI agents are a new frontier that defense should accelerate adopting to achieve these goals, as ‘AI enablers.'”

Alternatively, envision the U.S.—slow to adapt and constrained by traditional planning processes—competing against an adversary equipped with this “spear of thought” at strategic, operational, and tactical levels. When an adversary has a better and faster flow of information, no amount of high-tech equipment in the hands of operators can defeat a superior opponent. The implications of competing against an adversary with such intelligence and decision-making space are worth seriously considering the use of AI agents in parallel planning structures.

The value of AI in modern warfare and its impact on military operations and tactics has been evidenced by the Russia-Ukraine war. Earlier this year, Time magazine reported that Palantir Technologies’ AI software system was responsible for most targeting operations within Ukraine. Additionally, Palantir provided a software engineer to each battalion, accelerating what General Mark Milley, former Chairman of the Joint Chiefs of Staff, described as “the most fundamental change in the nature of warfare ever.” In fact, Defense One reported that the Pentagon has also been integrating “AI and machine learning into its intelligence, surveillance, and reconnaissance operations to help the Ukrainian military thwart some of Russia’s attacks.”

Agents in the Joint Operational Planning Process can provide information advantages at relevant speeds. Below are several ways agents can serve as effective means to achieve objectives:

1. Agents possess exceptional multi-domain situational awareness capabilities, enabling them to suggest troop posture to planners and create multiple dilemmas within the Multi-Domain Operations (MDO) framework by continuously organizing information about joint and allied forces and adversary movements.

2. Agents can help prioritize the Joint Integrated Priority Target List (JIPTL) based on real-time conditions in the operational space, including adversary capabilities, approach routes, risks, and opportunities.

3. Agents can track and assess potential logistical shortages (such as fuel, supplies, and ammunition) before they occur, ensuring sustainable support for dispersed forces across vast theaters of operation.

4. Agents can center “knowing oneself and knowing the enemy” in formulating operational actions. When the pressures of conflict naturally push people back to comfortable known states—our own ways of fighting—without needing the enemy’s vote, the red team is an element that planners may soon overlook.

5. Agents can instantly synchronize guidance and intent across the entire operational space, reducing the likelihood of friendly fire and enhancing tactical flexibility and lethality.

6. Finally, and most importantly, planners can leverage AI to generate and issue all downstream orders produced during the cyclical planning process, saving hundreds of man-hours in each cycle for tedious, repetitive administrative inputs and allowing more operators to be redirected to combat.

There is still much to learn about the risks of Agentic AI and its adaptability in controversial communication environments. Theoretical discussions on ethics, safety, and best practices should continue. However, many countries are fiercely competing in the AI race, clearly seeking technological advantages. Future wars will almost certainly be won first in the information domain.

Military leaders should accelerate experimentation and adopt AI agent tools in the Joint Operational Planning Process. It is crucial that they approach this with an iterative mindset, striving to reduce risks (machine learning will assist in this) rather than waiting for a perfect product to implement. When standing on the edge of a technological revolution, the risks of significant leaps must be accepted.

Reference Source: HKS

Reprinted from: ZHUAN ZHI Intelligent Defense