Once opposing forces engage in conflict, strategy gives way to the unexpected, the unplanned, the confusion that is the fog of war. Today’s modern forces that command the most complete situational awareness of the battlefield at the tactical edge almost always win the day.
Tremendous tactical advantages can be gained from a form of edge computing that relies on artificial intelligence (AI) software applications running on devices that are lightweight, small in size, and consume as little power as possible. Such devices can be carried easily by an individual in battle, and even discarded or destroyed if circumstances demand.
AI as Game Changer
You have probably heard the phrase, “AI is changing everything.” That’s proving increasingly true when it comes to military applications. In fact, AI and machine learning (ML) technologies have been used by the U.S. Department of Defense (DoD) for decades. For example, ML algorithms are essential to the rapid data acquisition and targeting command and control capabilities of the Aegis Combat System. But what’s different now is a growing consensus that AI can be a foundational game changer like no other in modern military history. The question now is not whether AI, it’s what kind of AI?
Private AI, an architectural approach that brings AI models to data, offers a compelling case for the future for modern enterprises, including the DoD, because it meets the enhanced data security and privacy requirements that will be needed in the AI frameworks of the future. These essential elements to building a modern private cloud, such as enhanced encryption, are not part of any battlespace AI framework today.
Decision Making in DDIL Environments
Building and deploying a highly flexible, adaptive AI framework that leans in on features such as enhanced security, privacy, and data modeling is tailor-made for military applications. Such applications must also be highly collaborative. Joint-force advantage needs situational awareness across a battlespace that is nimble — not just for mission flexibility but for the battlespace coalition — and designed to integrate partnerships with other agencies and allied forces.
In conflict scenarios, the data must be collected and analyzed as close to local conditions as possible. It must enhance the overall situational awareness that the battlespace commander needs to make real-time decisions.
But make no mistake: this is a complex problem with its own set of challenges. For decades, U.S. military doctrine had essentially taken for granted that our military superiority would always allow networked weapons systems and troops to transmit collected data up the Chain of Command through battlefield and satellite communications systems. The experience of drone communications in Ukraine demonstrates the potential fallacies in that assumption. It is clear now that our adversaries have the technology to disrupt these communications.
An edge-based approach to data collection, communications, and analytics is required in denied, disrupted, intermittent, and limited (DDIL) environments where cloud and other communications channels could be compromised or eliminated. Military organizations, including our own, no longer have the luxury of “phoning home” for instructions. These systems must now think for themselves in contested battlespace environments with the compute capacity that can be supplied at the edge.
National Security Demands Private AI Solutions
A software-defined approach to private AI military applications brings the most advanced AI innovation and benefits of the commercial sector to our national defense. Chief among these benefits is that a software-defined approach embraces the open ecosystem of software and hardware vendors that are leading AI innovation today.
In a BRAVO Hackathon conducted in 2023 among joint all-domain forces and hosted by U.S. Air Force Special Operations Command, one of the teams developed a private AI application that allows military forces to identify potential enemy objects in adverse cross-domain environments — under the sea or on land — while cut-off from any cloud-based communications and services.
The private AI model developed for this application demonstrates many of the benefits of private AI using a software-defined approach. It did not require a football field-sized server farm for its processing power. Its AI model was trained on the specific capabilities required for its mission. It integrated various off-the-shelf electronic components and commoditized the hardware needed. It brings with it minimal costs and enormous bang for the buck. And its dramatic tactical benefits demonstrate why the Special Operations community is a major driver inside the U.S. DoD for private AI and edge computing innovation.
A software-defined approach to AI lessens the need to update expensive hardware when upgrading the AI software. That’s a primary reason why a software-defined approach to private AI is the future of the enterprise, including the largest and most complex enterprise in the world, the U.S. DOD.
Our national security demands private AI solutions. A smart marriage between software-defined, managed private AI systems running on top of secure hardware creates weapons systems that are more nimble, agile, and capable. It is at the tactical edge where such advantage-granting capabilities are most essential to our war fighters.