AI, quantum could reshape command decisions, but human judgment remains critical, expert tells maneuver leaders

FORT BENNING, Ga. — A commander on a future battlefield may have access to more information than any commander in history — feeds from unmanned systems and sensors, intelligence reports, friendly-force locations, targeting data, equipment readiness, ammunition and fuel levels, weather, terrain, airspace and activity across the electromagnetic spectrum.

Where past commanders were challenged simply by obtaining information, future commanders will struggle to determine, in seconds, which information matters, which can be trusted, and how to act on it.

Dr. Kristin Milchanowski told Army leaders at the Maneuver Warfighter Conference Sept. 16 that artificial intelligence has the potential to dramatically expand the information and options available to commanders, but that capability does not diminish the importance of human judgment.

Rather, it could make that judgment more consequential.

“Decision advantage is not about deciding things faster,” Milchanowski said. “It’s about arriving at the right decision earlier with fewer unknowns and then acting on it.”

Milchanowski is chief artificial intelligence and quantum officer at BMO Financial Group, one of North America's largest banks, and founding director of the BMO Institute for Applied Artificial Intelligence & Quantum. She holds a doctorate in decision sciences and completed postdoctoral studies in artificial intelligence at the University of Oxford. In her current role, she leads efforts to scale AI across a large, highly regulated organization while managing the governance and risk that goes along with its use.

Her message to Infantry and Armor leaders was not simply about adopting new technology. It was about how leaders must adapt as machines become increasingly capable of processing information, identifying patterns, and generating options at speeds humans cannot match.

“The AI wave that you are all living in and experiencing is going to change the questions you have to ask,” Milchanowski said. “The quantum wave that is coming is going to change what you can know.”

More information, different problems

The Army is already experimenting with ways AI could reduce the cognitive burden on commanders and staff. Researchers at the U.S. Army Combat Capabilities Development Command Army Research Laboratory, for example, are testing COA-GPT, an experimental AI-supported platform intended to help Soldiers rapidly develop and evaluate courses of action during battlefield planning.

But Milchanowski cautioned that more data does not automatically produce better decisions.

Instead, “it produces a different and harder set of leader problems, and the moment information becomes abundant, the leader’s job changes,” she said.

Milchanowski drew on her experience integrating AI into large organizations, telling the audience she initially introduced the technology as a way “to protect my people’s time, and to protect the judgment they bring to the table.”

As AI systems assume more of the work of assembling, correlating, and analyzing information, she said, leaders must become more deliberate about interrogating the products those systems provide.

She suggested commanders approach AI-generated information with three questions: What information should I disregard? What assumptions are the system making? And what would have to be true for its assessment to be wrong?

“The abundance of information didn’t lighten the burden of command,” Milchanowski said. “It just relocated it.”

Trust — and doubt — as command variables

That relocation becomes particularly important as an AI system repeatedly proves itself reliable.

Milchanowski argued that one of the most dangerous moments in using an AI-enabled system may be when users have become accustomed to it being right.

“Your instinct is very natural,” she said. “The system performs, the picture sharpens, and you extend it more trust.”

But as confidence increases, she said, scrutiny can decrease, making subtle errors or changes in the system more difficult to detect.

“The risk curve runs opposite of the confidence curve,” Milchanowski said, “and that means that confidence is now a command variable.”

Her recommendation to leaders was counterintuitive: schedule doubt.

Even when systems appear to be functioning correctly, staff should deliberately verify outputs manually and maintain ways to challenge the common picture, she said.

Integration itself can introduce another vulnerability.

Historically, commanders could gain confidence when independent sources arrived at the same conclusion — or recognize uncertainty when they disagreed. But when numerous sources are integrated into a single system, Milchanowski said, the resulting picture can appear internally consistent even when its underlying assumptions are wrong.

“It manufactures a shared blind spot,” she said, describing the potential result as “a single point of failure.” Her proposed solution is to deliberately engineer disagreement into the process — retaining an independent feed, red-team function or other mechanism charged with asking what the operational picture would look like if it were wrong.

That skepticism, she said, should not be confused with resistance to technology.

“I like the skeptics …” Milchanowski said. “… They’re the ones who are going to ask why and inject why into the system. That is so important because machine-augmented command becomes an advantage, but only when we’re asking that question.”

Human judgment in a faster loop

AI systems can also embed human assumptions into decisions at machine speed, Milchanowski cautioned. Commanders and staff establish priorities, thresholds, and decision criteria that can effectively encode command intent into systems. If those assumptions contain bias or error, faster processing can amplify rather than eliminate the problem.

“The faster the loop, the more essential the human checkpoints are,” she said.

A lesson from her own experience deploying AI at scale, she said, was equally direct.

“If I couldn’t say which decisions my systems were making without me, then I delegated more than I intended.” The issue is particularly relevant as the Army experiments with distributed command and control and disperses staff to improve survivability on a battlefield where concentrated command posts may be increasingly vulnerable.

Milchanowski pointed to recent Army Project Convergence experimentation as an example of an environment where AI could help dispersed staff continue producing planning products and analyzing large volumes of information.

“I actually see AI as being how our staff survives the modern battlefield,” she said.

Her emphasis on retaining human judgment echoed a theme heard earlier during the conference.

“We must always think about how technology enables the Soldier, the formation,” said Maj. Gen. Colin P. Tuley, Maneuver Center of Excellence and Fort Benning commanding general. “Technology won’t replace maneuver … used effectively, it can very much enable maneuver.”

Combined Arms Command Command Sgt. Maj. Shaun D. Curry offered an even more grounded reminder.

“Our profession is in the dirt, and if you are going to rely on AI forever, you’re not going to succeed,” Curry said.

A second wave

If AI changes how commanders use information, Milchanowski argued that another emerging technology — quantum — could eventually change what information can be known, versus only estimated.

She described AI and quantum as two distinct technological waves.

AI can help generate and evaluate more options and push a commander's “decision space” farther forward, she said, but its outputs remain estimates. Milchanowski described future quantum computing capabilities as potentially reducing some of the uncertainty inherent in computationally difficult problems, with possible applications ranging from logistics and resource allocation to materials science.

She connected that possibility to a problem from her own earlier work involving the Bradley Fighting Vehicle. Following the Sept. 11 attacks, Milchanowski said she worked on a requirement to increase protection on the underside of Bradleys without increasing vehicle weight.

“I was asked to make the floor stronger, and I couldn’t make it any heavier,” she recalled. “I had a week to come up with the math to facilitate all of that and get it into theater yesterday.”

Asked recently whether today's emerging technologies could approach the same problem differently — potentially designing materials at the molecular level to improve survivability — her answer was, “Quite possibly, yes. Our capabilities in technology have gotten this advanced.”

Milchanowski told the audience that AI could eventually help search enormous numbers of potential designs while quantum technologies could expand what is computationally possible in areas such as chemistry and materials.

“The future war is going to be battled with both of these waves,” she said. “And frankly, I don’t think we’re doing a good enough job talking about both waves.”

Learning the next tools of warfare

For Milchanowski, however, the central challenge presented by either technology is ultimately a leadership problem.

“Neither wave is really actually about technology,” she said. “Both are about whether or not leadership is going to adapt at the pace of capability across both of those waves.”

She compared the moment to previous technological changes in warfare. Leaders confronted with the machine gun or tank could not afford simply to dismiss a new capability because it was unfamiliar, she said. They had to study it, understand it, and determine how to employ it.

“Being dismissive of AI and quantum is not caution,” Milchanowski said. “It’s resigning the future of warfare to our enemies.”

Lt. Gen. James P. Isenhower III, Combined Arms Command commanding general, similarly emphasized the need for formations capable of evaluating rapidly emerging technologies rather than assuming every new capability will prove useful.

“Given how fast things are moving and the character of war, it’s important that we have formations that can take technology immediately and tell us whether or not we think it has promise,” Isenhower said.

For Milchanowski, learning to employ those technologies does not mean transferring the commander's responsibility to a machine.

“We have to keep human judgment in the loop,” she said, “and we have to let machines extend rather than replace it.”

Editor’s note: This article was drafted with the assistance of generative artificial intelligence. The author reviewed, edited, and fact-checked the final article for accuracy and compliance with Department of War policies and guidance.

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.