The Air Force is increasingly experimenting with artificial intelligence to improve battle management. This is both necessary and overdue. The volume of information available to tactical command and control crews already challenges what humans can reasonably process, and future conflict will add more sensors, shooters, data links, autonomous systems, and increasingly compressed decision timelines. Artificial intelligence offers an opportunity to automate many of the processes that consume an Air Battle Manager’s attention without requiring the judgment that defines air battle management.
Recent experimentation suggests the gains could be significant. During the Multi-Domain Decision Advantage Sprint for Human-Machine Teaming, or MASH, an Air Battle Manager from the 552nd Operations Support Squadron reported that his team could complete five or six taskings with AI assistance in roughly the time previously required to complete one. That kind of capacity increase matters in an environment where the number of decisions required of a C2 crew may increase faster than the number of operators available to make them.
Applied deliberately across integrated surveillance and identification, force management, information management, and continuum of control, AI can improve the speed and quality of tactical decision-making while allowing Air Battle Managers to focus on translating commander’s intent and delegated C2 authorities into action. The challenge is therefore not determining whether AI should perform air battle management functions, but determining where automation ends and the exercise of authority begins.
Applying AI to Air Battle Management
C2 Coord has previously addressed the promise and limitations of AI in command and control and air battle management. The basic argument is established. AI can help operators manage increasing volumes of information, recognize patterns, develop courses of action, and reduce cognitive workload. It cannot substitute for experienced operators exercising judgment in contested and ambiguous environments.
The next step is to move beyond discussing AI as a generic C2 capability and determine where it provides value within air battle management itself. As discussed during the Mitchell Institute’s Tactical C2 Imperative forum, air battle management entails the real-time direction of air operations within commander’s intent through four interrelated functions: integrated surveillance and identification, force management, information management, and continuum of control. Each contains processes well suited for machine assistance, but each also contributes to decisions that may require context, judgment, and delegated authority.
Integrated surveillance and identification provides perhaps the clearest opportunity. Air Battle Managers continuously evaluate information from multiple sensors and sources to understand what is operating in the battlespace and determine what those entities may be doing. Much of that work relies on correlation and pattern recognition. An AI-enabled system could continuously evaluate modes and codes, point of origin, track behavior, flight profiles, identification criteria, electronic signatures when available, and other relevant information across hundreds or thousands of tracks.
A track squawking an expected mode and code but originating from an unexpected location, deviating from an expected route, or behaving inconsistently with its reported identity should attract attention. AI can identify those inconsistencies faster than an operator manually searching multiple sources. More importantly, the system should provide the evidence supporting its assessment rather than simply assigning an identification. The Air Battle Manager must still understand why a recommendation was made and apply that information against identification criteria, rules of engagement, commander’s intent, and the broader operational situation.
Force management presents a different opportunity. Consider management of the theater fuel plan. An Air Battle Manager may simultaneously consider Link-reported fuel states and tanker offloads, receiver requirements, aircraft types, recovery locations, crew reports, and the original plan. None of these inputs is especially difficult to understand on its own. Continuously evaluating all of them while managing the rest of the air fight is another matter.
AI could compare actual tanker offloads against the plan, evaluate receiver fuel states, account for aircraft type and expected consumption, incorporate crew-reported changes, and project whether the fuel plan remains viable. If one tanker delivers more gas than planned, another is delayed, and a fighter package consumes fuel faster than expected, the system should identify the developing shortfall before it becomes a crisis and recommend options to mitigate it. The same principle applies to weapons availability, fighter allocation, sensor coverage, recovery options, and other finite resources. AI can continuously run the calculations and flag emerging problems, allowing Air Battle Managers to focus on how to prioritize combat power against the commander’s objectives.
Information management may offer the quickest return. Modern C2 crews do not lack information; the challenge is determining which information matters. Relevant data arrive through tactical data links, voice communications, intelligence products, mission systems, collaborative applications, and multiple chat rooms. Requiring operators to continuously monitor every source while maintaining situational awareness and directing forces imposes cognitive workload without necessarily improving decision quality.
AI could monitor approved information sources, identify information relevant to the crew’s mission, correlate it with the tactical picture, and alert the appropriate crew position. A threat report from another agency, a change to airspace, tanker status, weather, or mission priorities should not depend on an operator monitoring the correct chat room at the right moment. The objective is not another summary screen. It is getting relevant information to the operator who needs it while there is still time to act.
Continuum of control presents perhaps the greatest opportunity and the clearest limit. AI could continuously evaluate fighter position and status, target position and aspect, identification status, threat laydown, weapons employment zones, fuel, supporting assets, and established commit criteria. It could recognize when those criteria are satisfied, identify the forces best positioned to respond, and recommend a tactical action faster than an operator could manually assemble the same information.
This capability could substantially improve tactical execution, but satisfying predetermined criteria is not always the same as deciding to commit forces. The commander may have accepted risk in one area to preserve combat power for another. Priorities may have changed, authorities may be conditional, or the Air Battle Manager may possess information that has not entered the machine-readable battlespace. AI can clarify the tactical situation and make available options easier to understand. It should not make responsibility for the resulting decision ambiguous.
Automation and Authority
These applications suggest a practical way to approach AI integration. Tasks dominated by correlation, calculation, monitoring, and recognition are strong candidates for automation. Bounded tactical problems in which multiple variables can be evaluated against established criteria are appropriate for AI-generated recommendations. Decisions requiring interpretation of commander’s intent, acceptance of operational risk, or exercise of delegated C2 authority require greater human involvement.
Those boundaries will change as AI technology matures. Tasks that initially require a machine recommendation and human approval may eventually become candidates for automated execution. Commanders may also establish conditions under which automated systems can act within predetermined parameters, particularly when engagement timelines make additional human intervention impractical. As C2 Coord has previously explored in discussing autonomy, some of the Air Battle Manager’s work may consequently move upstream toward establishing the conditions under which machines execute. That evolution should be deliberate and based on authorities and acceptable risk, not simply on what the technology can do.
Effective air battle management has always depended on delegation. A commander cannot personally direct every fighter, tanker, sensor, weapon, or tactical decision across a theater. Authorities are delegated so subordinate commanders and tactical C2 elements can continue operating at the speed combat requires. Air Battle Managers execute within those authorities to manage forces, respond to changing conditions, and maintain alignment with commander’s intent when circumstances depart from the plan.
AI does not remove this requirement. Faster decision cycles may make the boundaries of delegated authority more important. A future system may identify a threat, determine available effects, rank potential effectors, account for supporting requirements, and develop a recommended course of action in seconds. That capability is valuable, but it does not answer what happens when the recommended action conflicts with commander’s intent, assumes risk the commander has not accepted, or encounters circumstances outside the parameters on which the system was designed. Those are command-and-control problems, not computing problems.
That is why describing human-machine teaming as simply keeping a human “in the loop” is insufficient. Air Battle Managers do more than approve technically feasible options generated by a machine. They interpret commander’s intent, understand relationships between tactical actions, manage competing priorities, recognize when assumptions have changed, and make decisions within delegated authorities amid the uncertainty inherent in combat. AI should reduce attention devoted to processes that do not require those abilities so operators can devote more attention to the decisions that do.
Implications for the Air Battle Manager
Increasing automation does not necessarily reduce the importance of the Air Battle Manager. It changes where the Air Battle Manager provides value. As machines assume more routine cognitive work, the remaining human decisions may become fewer but more consequential.
An operator who previously devoted significant attention to maintaining situational awareness may increasingly receive an automatically correlated picture. An Air Battle Manager who manually calculated whether a force-management problem was developing may instead receive an alert accompanied by several possible solutions. A controller who mentally evaluated each element of commit criteria may receive an immediate indication that those criteria have been satisfied. None of this eliminates judgment. It shifts human attention toward the areas where judgment matters most.
Training must evolve accordingly. Future Air Battle Managers will need to understand the assumptions behind machine-generated recommendations, recognize when those assumptions no longer match operational reality, determine when an optimized solution conflicts with commander’s intent, and know precisely what authorities permit them to act. They must also retain sufficient tactical expertise to recognize when the machine is wrong. Overreliance on automation without the expertise to challenge its output would simply replace one form of cognitive limitation with another.
This also has implications for how AI-enabled C2 systems are designed. Explainability cannot be an afterthought. An operator receiving a recommendation to identify a track, reposition a tanker, reprioritize forces, or commit fighters should be able to understand the information and assumptions that produced it. A recommendation delivered faster is of limited value if the Air Battle Manager cannot determine whether it remains valid when conditions change.
Preserving the Decision Advantage
The future battlespace will increasingly exceed what unaided humans can effectively manage. The Air Force should therefore identify tasks within each air battle management function that can be automated now, experiment with where machine recommendations improve tactical decision-making, and deliberately establish the authorities and risk thresholds governing automated execution.
The objective is not to preserve manual processes simply because Air Battle Managers have historically performed them. AI should correlate information, monitor the fuel plan, find the critical message, recognize when tactical criteria are satisfied, and present viable options faster than humans can do alone. This creates greater capacity for Air Battle Managers to focus on context, judgment, risk, and executing delegated authority.
The value of AI-enabled air battle management should ultimately be measured by whether it improves the joint force’s ability to translate commander’s intent into timely tactical action. Achieving that advantage will require both capable machines and Air Battle Managers with the expertise to understand when to trust them, when to question them, and when circumstances demand something the algorithm did not anticipate.
AI can accelerate air battle management. It cannot resolve the fundamental question of command. Technology may change how quickly a decision can be developed, but speed does not confer authority. The machine can recommend; the commander still commands.
Col Grant “SWAT” Georgulis, USAF, is a Master Air Battle Manager and currently assigned as the Deputy Chief of C2 Inspections as part of the Headquarters NORAD and NORTHCOM Inspector General team. He most recently finished a year-long Air Force National Defense Fellowship at The Mitchell Institute for Aerospace Studies for the academic year 2024-2025. He entered the Air Force in 2007 through the ROTC program at Texas State University–San Marcos. SWAT has served on a combatant command component staff, was an Air Force Weapons School instructor, and graduated from the Naval War College’s College of Naval Command and Staff and Air University’s School of Advanced Air and Space Studies. He previously commanded an E-3G Squadron, the 965 Airborne Air Control Squadron, at Tinker Air Force Base, Oklahoma.
The views expressed are those of the author and do not reflect the official policy or position of the U.S. Air Force, Department of Defense, or U.S. government.
Photo by Kindel Media on Pexels
Leave a Reply