The 101st Airborne Division’s 3rd Mobile Brigade Combat Team has concluded that large language models (LLMs) are ill-suited for tactical course-of-action development after a year-long integration of artificial intelligence into its planning processes, according to its commander, Col. Ryan Bell.
Speaking to reporters following an April rotation at the Joint Readiness Training Center at Fort Polk, La., Bell detailed the brigade’s pragmatic assessment of AI’s capabilities and limitations. The unit trained LLMs on joint, Army, and division doctrine, creating specialized bots for each staff section to accelerate understanding of the operating environment.
AI Proves Ineffective for Tactical Maneuver Planning
Despite industry hype positioning AI as a key tool for battlefield scenario generation, Bell explicitly ruled out using the technology for course-of-action development. “Large language models don’t really understand three-dimensional space,” he said. “That’s where you need the expertise of a skilled staff to understand the art of war fighting.”
The brigade found that while AI could simulate and test potential courses of action ahead of enemy contact, it could not replace human judgment in crafting maneuver plans. The technology’s inability to grasp spatial relationships in a contested environment remains a fundamental constraint.
Operational Gains in Tempo and Orders Processing
Where AI delivered measurable value was in accelerating the orders process and enabling deeper wargaming. Bell reported that using AI tools, the brigade could push a warning order to subordinate units in under 30 minutes—a process that previously required hours.
In a defensive scenario, battalions received their orders 72 hours ahead of the opposing force’s reconnaissance push. “It enabled the battalions to fully work through their planning cycle… put in 100 percent of the plan obstacles, refine, rehearse,” Bell said. This time advantage allowed units to build defenses that withstood multiple chemical attacks and robotic breach attempts.
Data Processing at Scale
Over ten days of operations, the brigade’s intelligence section processed more than 25,000 spot reports from drone sensors using AI. “It helped us make sense of the battlefield and respond faster than Geronimo,” Bell noted, referring to the opposing force.
The commander emphasized that tempo will be a decisive characteristic in future operating environments. While drones accelerate sensing and striking, AI proved its worth at the staff level by compressing planning cycles and sustaining operational momentum.
The 101st’s experience offers a clear signal for defense procurement: AI excels at accelerating administrative and analytical workflows but cannot substitute for human expertise in the art of war. As the Army pushes toward broader AI adoption, the distinction between tactical decision-making and operational support will define where the technology delivers—and where it falls short.
— Originally reported by Breaking Def.. Adapted and republished with editorial context for SpaceSecurityNews.
