Rapid-fire AI Capabilities and Boundaries: Exploring the Potential and Limitations of Artificial Intelligence
The United States Army is enhancing its Decision, Detect, Deliver, Assess (D3A) targeting methodology by integrating artificial intelligence (AI), aiming to accelerate and improve the entire targeting cycle while maintaining human accountability.
Key aspects of this integration include faster, data-driven decision-making, enhanced detection and target identification, automation in target delivery, and improved assessment. AI systems provide real-time data analytics, creating dynamic battlefield scenarios and adaptive feedback, enabling Soldiers and commanders to make quicker and better-informed decisions during live engagements and training exercises.
In the detection phase, AI systems accurately distinguish threats from benign activities with high accuracy, improving the detection phase of D3A by reducing false alarms and augmenting human operators' situational awareness. AI also supports proper combat techniques, assisting in target engagement through precision aiming and control of weapon systems while providing continuous real-time evaluation of effectiveness and adjustments as necessary.
AI facilitates assessment by generating detailed reports and feedback rapidly, allowing faster adjustments to tactics and ensuring effective outcome measurement within operations. Despite automation, humans remain accountable, guiding AI outputs, validating target information, and making final decisions before lethal force application, ensuring compliance with laws and ethical standards.
AI/ML create nuanced training environments that simulate realistic combat conditions, allowing soldiers to develop decision-making under AI-augmented scenarios. This supports continuous refinement of skills essential for effective targeting while familiarizing personnel with AI-enabled systems.
Jesse R. Crifasi, a retired US Army chief warrant officer 4, a senior advisor in the defense industry specializing in joint fires and targeting, and a PhD student in public policy and national security at Liberty University, has emphasised the importance of AI integration in the Army's targeting operations. He has authored multiple doctrinal and technical assessments on digital fires and artificial intelligence integration in targeting operations.
The integration of AI into D3A is about optimising the targeting workflow, particularly in accelerating sensor-to-shooter kill chains, reducing cognitive burden, and improving commanders' decision-making in contested environments. As the Army advances under multidomain operations, strategic competition, and transformation initiatives, the necessity of integrating artificial intelligence into targeting workflows becomes paramount.
AI should augment, not replace, critical targeting functions like rules of engagement validation, proportionality assessments, and determinations of military necessity. AI offers scaling advantages, particularly in data processing and decision acceleration. The goal is to embed AI where it adds the most value while ensuring humans remain central at key decision points.
Large language models (LLMs) like Meta's LLaMA do not inherently understand doctrinal terminology or contextual nuance. Prompting a commercial, off-the-shelf LLM on how to destroy a target would not result in a comprehensive understanding of the concept. Achieving the effect "destroy" beyond physical damage requires structured training on doctrinal lexicons, rules-based decision trees, and munitions modeling.
AI in targeting presents a moral dilemma, it must be employed as a tool, not as a substitute for the warfighter's judgment. Time is the most compelling performance metric for evaluating AI effectiveness in the targeting process. D3A remains the United States Army's targeting methodology, and the views expressed are Crifasi's own and do not reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.
- The integration of artificial intelligence (AI) into the United States Army's Decision, Detect, Deliver, Assess (D3A) targeting methodology is paramount as it aims to automate and enhance the entire targeting cycle while maintaining human accountability.
- In the detection phase, AI systems help reduce false alarms, improve situational awareness, and accurately distinguish threats from benign activities.
- AI/ML create nuanced training environments, allowing soldiers to develop decision-making under AI-augmented scenarios, essential for effective targeting and familiarizing personnel with AI-enabled systems.
- AI should be employed as a tool to assist in critical targeting functions like rules of engagement validation, proportionality assessments, and determinations of military necessity, ensuring humans remain central at key decision points.
