Agentic AI in Future Military Operations

Daniel Poor's article, "Boxes, Bubbles, and Blank Slates" highlights how AI systems, particularly Agentic AI, will reshape organizational decision-making. It argues that knowledge workers often waste time on non-value-added tasks, such as data entry and validation, that could be automated.

The article groups decisions into three categories: Box, Bubble, and Blank Slate decisions, each with varying degrees of ambiguity and complexity.

1. Box decisions are routine and rule-based, ideal for automation with simple coding.
2. Bubble decisions involve more context, allowing for greater discretion but still manageable by AI systems.
3. Blank Slate decisions require human creativity due to high uncertainty and ambiguity.

Daniel emphasizes that current systems are limited in automating many decisions, leaving human workers to perform these tasks. The cost of human decision-making includes labor costs, error rates, and delays. Although systems could automate Box and some Bubble decisions, IT departments often prioritize more significant projects, leaving routine decision-making to employees.

Agentic AI, which combines RPA, specialized AI, and generative AI like ChatGPT, can now fill these gaps. These systems dynamically gather contextual information, apply decision frameworks, and orchestrate execution, allowing more decisions to be automated. The result is more consistent and cost-effective decision-making.

Daniel's framework on Agentic AI, when applied to military operations and decision-making in the current and near-term Operating Environment of omnipresent unmanned robotic systems and cyber / electronic warfare, suggests a fundamental shift in how decisions are made on the battlefield. This transformation will be especially pronounced during great power competition and large-scale combat operations (LSCO), where the pace, complexity, and scale of operations far exceed what traditional human decision-making can manage.

Humans cannot compete with Robots in accordance with John Boyd's framework for decision-making: Observe, Orient, Decide, and Act (referred to as the OODA loop). Military Commanders must leverage Automation, Artificial Intelligence, and Agentic AI to compete in future wars.

1. Automating Routine Tactical Decisions (Box Decisions)

In military operations, Box decisions are similar to routine, rule-based tactical decisions with low ambiguity. These are decisions where clear “if-then-else” logic applies, such as:

• Autonomous defense systems engaging an incoming missile if it meets specific threat parameters.
• Robotic supply convoys rerouting based on predetermined criteria such as road conditions or enemy activity.
• Automatic response systems to cyber threats, where pre-programmed responses can isolate or neutralize a threat based on its signature.

We are already seeing the early version of this, as Dr. Gordon Cooke predicted, in Ukraine. In 2019, he stated "Networked smart weapons will also allow logistics systems to automatically initiate resupply actions as soon as combat begins, providing just-in-time logistics all the way to the forward edge. Supply and transportation assets will be able to begin rerouting truckloads of supplies across the battlespace to the point of need. At the tactical level, small robots will be able to bring loaded magazines to individual Soldiers as their weapons reach the end of their basic combat load." -Magic Bullets

In the near future, Agentic AI could automate these tactical-level decisions at scale, allowing systems to operate at machine speed without waiting for human intervention. For instance:

• Autonomous drones and robotic systems could rapidly assess and engage targets within predefined rules of engagement, reducing the time needed for human operators to make decisions.
• Electronic warfare systems could autonomously decide how to disrupt or deceive enemy communications based on real-time analysis of the electromagnetic spectrum.

Implications: The automation of these routine decisions would enable military forces to outpace adversaries in terms of reaction time and efficiency, allowing them to seize the initiative and minimize human error in high-tempo operations.

2. AI-Augmented Operational Decisions (Bubble Decisions)

Bubble decisions in military contexts involve higher ambiguity and require more nuanced judgment, such as:

• Deciding on the allocation of limited resources across multiple fronts.
• Determining the optimal mix of robotic, human, and cyber assets in a given battle space.
• Adjusting the operational tempo in response to evolving enemy tactics.

Agentic AI can assist in augmenting human decision-makers in these situations by gathering contextual data from diverse sources (ISR platforms, cyber sensors, and human intelligence) and proposing courses of action (COAs) based on probabilistic models. For example:

• AI could synthesize data from multiple domains (air, land, sea, cyber, space) to suggest the most effective employment of forces.
• During electronic warfare or cyber campaigns, AI could assess shifting enemy strategies and suggest countermeasures in real-time.

While humans would still make the final decision, the AI could offer preemptive options and help commanders prioritize and manage the growing complexity of the modern battlefield.

Implications: The increased use of AI in Bubble decisions would accelerate decision-making cycles (OODA loops), allowing military commanders to adjust to battlefield conditions more swiftly and with greater precision. The AI’s ability to handle vast data sets would also help mitigate information overload, a growing challenge in modern warfare.

3. AI and Command-Level Strategy (Blank Slate Decisions)

At the strategic level, Blank Slate decisions are characterized by high levels of uncertainty and ambiguity, where historical examples or precedents may not exist. In military contexts, these are:

• Deciding on grand strategy in the face of emerging technologies, new geopolitical threats, or unprecedented warfare conditions.
• Determining long-term military posture during great power competition, where conventional and hybrid warfare tactics constantly evolve.

Agentic AI, particularly Generative AI systems, could assist senior military leadership by providing strategic decision-support in these highly complex and novel situations. The AI would aggregate massive data sets from global intelligence, economic models, political trends, and military readiness, then offer scenarios that account for multiple contingencies.

For example:

• In a great power conflict, Agentic AI could assess the consequences of different operational strategies (e.g., forward deployment of autonomous assets vs. a defensive posture) and their implications on global alliances, economic stability, and political risks.
• During hybrid warfare, AI systems could help military leaders understand the interrelationship between cyberattacks, disinformation campaigns, and conventional force deployments, proposing comprehensive counter-strategies that align with broader political and strategic goals.

Implications: The introduction of Agentic AI into strategic decision-making would increase adaptability in highly fluid, uncertain environments. AI could explore a wider range of options than human decision-makers alone could consider, providing senior commanders with creative solutions to complex problems that blend both kinetic and non-kinetic forms of warfare.

4. Impact of Agentic AI on Military Systems and Cyberspace

Agentic AI would also have profound implications on military systems and their integration into cyber and electronic warfare.

• Cyber Warfare: AI agents could autonomously detect, analyze, and respond to cyber threats on the battlefield in real time, drastically reducing the window of vulnerability from an attack. The AI could adapt to new threats faster than humans and offer predictive insights into future attack vectors based on data patterns.
• Robotic and Autonomous Systems: Future robotic systems in LSCO environments could leverage Agentic AI for swarming tactics, self-repair, and decision-making in environments too dangerous for humans. These robots could decide how best to achieve mission objectives while autonomously navigating hostile terrain, enemy defenses, and operational constraints.

5. Risks and Challenges

While Agentic AI offers immense potential, its implementation in military operations poses significant challenges:

• Ethical Concerns: The delegation of lethal decisions to AI systems raises concerns about accountability, especially in scenarios where the AI makes a mistake or where the context may require more human empathy and moral judgment.
• Adversarial AI and Cyber Vulnerabilities: Widespread use of AI in military systems could also introduce new vulnerabilities, particularly in the form of adversarial AI attacks or the hacking of autonomous systems.
• Coordination of AI and Human Forces: The integration of AI and human decision-makers requires seamless coordination to ensure that decisions made by AI agents align with broader operational and strategic objectives.

Conclusion

In the future of military operations, robotic autonomous systems and Agentic AI will play a pivotal role in transforming decision-making processes. By automating routine decisions, augmenting complex operational choices, and supporting strategic planning, AI will enable military forces to operate at machine speed, enhance precision, and reduce errors. However, as AI becomes more integrated into military structures, it will require careful oversight to ensure ethical, secure, and effective use during great power competition and large-scale combat operations.

Disclaimer: Expressed opinions are my own and not that of my civilian or government employer. This article was co-written with ChatGPT 4o.