By Archana Pyati
In November 2025, the Department of Defense announced that it would focus on six critical technology areas that will “define the future of American military superiority.” The six areas represent “the cutting edge of research and engineering” aimed at ensuring that the United States military “remains the most lethal fighting force in the world.” Underpinning these six technology areas are decades of basic science and fundamental research conducted at America’s research universities.
In a series of articles, AAU is highlighting the contributions of university research in developing these critical technology areas. This article – the first of six – discusses the university-based research that enables the artificial intelligence technology now being deployed in the battlefield to support our armed forces and secure and protect the nation.
On the battlefield, the ability to make decisions and act against adversaries with speed and precision is paramount. Slow or misinformed decisions can result in loss of life, mission failure, and defeat.
Warfighters and commanders must rely on seemingly infinite sources of intelligence and data to forecast and respond effectively to threats. Indeed, we now live in an era when these sources of information have proliferated beyond the ability of human beings to process and act upon all of them.
That is why artificial intelligence (AI) has become an integral part of U.S. military operations and is currently being deployed across all branches. AI has transformed how the military conducts operations through three distinct capabilities: analyzing huge amounts of data, accelerating decision-making, and enabling autonomous action.
AI allows the military to recognize targets faster; predict and neutralize threats with greater accuracy and efficiency; secure and optimize supply chains; protect soldiers using drones and other autonomous weapon systems; conduct simulations and trainings for troops; and generate real-time situational awareness through sensors, satellite imagery, robots, and other modes of surveillance.
Believe it or not, the Department of Defense has supported university-based AI research since the 1950s through an initiative undertaken by the Defense Advanced Research Projects Agency (DARPA) and other service branches. During the 1991 Persian Gulf War, the military used a DARPA-developed AI tool – the Dynamic Analysis and Replanning Tool (DART) – for logistical planning and supply transport. The Pentagon recently adopted the Maven Smart System, an AI-enabled platform for targeting and logistics that expands upon Project Maven, which launched in 2017.
Foundations in Computer and Cognitive Sciences
The foundational scientific research behind the AI currently modernizing U.S. military capabilities originated not in the Pentagon or government labs, but as theoretical models at universities. Early AI researchers could not have predicted the future applications of their work.
The power of AI is its ability to approximate aspects of human cognition. AI models can be trained to recognize patterns and learn from data; advanced models can even generate novel outputs not specified in advance by a human programmer. The AI used by the military and in countless other contexts integrates several techniques and technologies developed by computer and cognitive scientists: artificial neural networks, deep learning, backpropagation, parallel computing, and the hardware to power these processes.
University researchers pioneered many of the foundational innovations behind AI. The Dartmouth Summer Research Project on Artificial Intelligence conducted in 1956 is considered by many to have inaugurated the field by convening John McCarthy, Marvin Minksy, and other early AI researchers. McCarthy and Minsky formed one of the world’s first AI labs at the Massachusetts Institute of Technology. McCarthy, who joined the Stanford University faculty in 1962, also founded the Stanford Artificial Intelligence Lab (SAIL).
Although they were not the first to propose the idea of artificial neural networks, university researchers Geoffrey Hinton and John J. Hopfield did much to advance this technology by applying concepts from physics to develop neural networks, inspired by the brain’s complex circuitry. The networks could be trained for deep learning, a more advanced form of machine learning that can process and interpret complex forms of data with less human intervention. Hinton and Hopfield received the Nobel Prize in physics in 2024 for their achievements.
Hinton and other university-based cognitive scientists demonstrated their ideas through the Boltzmann machine, which can learn from examples rather than explicit instructions. Hinton also popularized the backpropagation algorithm (which trains neural networks to find “hidden layers”) along with two other AI pioneers, David E. Rumelhart and Ronald J. Williams , in a seminal 1986 research paper. The backpropagation algorithm eventually became the essential key to unlocking neural networks’ full deep learning power, overcoming limitations of earlier models.
At the same time, Hinton, Rumelhart, and James L. McClelland developed a new framework for understanding human cognition – parallel distributed processing (PDP) – outlined in two textbooks that redefined the field. The researchers formed the PDP Research Group, and their books spurred the cognitive science community to develop, explore, and test new computational models for human learning, language, and cognitive development.
Hinton, McClelland, and Rumelhart received the Golden Goose Award in 2024 in recognition of their work. Their discoveries would not have been possible without sustained support from the Department of Defense, the National Institutes of Health, and the National Science Foundation.
All these research university coauthors intersected early in their careers at the Institute for Cognitive Science at the University of California, San Diego. Each would go on to have distinguished careers at several leading research universities, including UCSD and Stanford University (Rumelhart and McClelland), Carnegie Mellon University and the University of Toronto (Hinton), and Northeastern University (Williams).
Yann LeCun, a professor of computer science and engineering at New York University and former chief AI scientist at Meta, has advanced the training of neural networks even further, particularly in the area of image recognition.
From Theory to the Battlefield
What unites these scientific advances is that they emerged from curiosity-driven research in universities, not from efforts to solve immediate battlefield problems conducted by federal or industry laboratories. Only decades later would these foundational ideas – developed without any clear military application in mind – become central to modern defense capabilities.
Today, the military’s integration of AI is thoroughgoing and is redefining the terms of war. In January, the DOD launched an AI acceleration strategy in warfighting, intelligence, and operations with the aim of transforming the U.S. military into an “AI-first” force.
In the moments that U.S. warfighters call upon AI to help them meet their objectives and return home safely, it can be easy to forget that this technology – like so many others – began not on the battlefields of today, but decades ago as cutting-edge ideas in the classrooms and laboratories of America’s leading research universities.
Archana Pyati is editorial and content officer at AAU.