Three Questions About the Future of AI in Defense, Thoughts from the AI lead at AV's Innovation Center
October 28, 2025
By Zarinah Casanova, AI Lead for AV’s Innovation Center
Artificial intelligence has entered our daily lives—but on the battlefield, its role carries life-and-death consequences. That shift from everyday convenience to mission-critical responsibility is where my work begins.
For the past two decades, I’ve worked across cloud analytics, data science, and machine learning operations. Today, I lead AV’s Innovation Center in Annapolis Junction, Maryland, where we’re developing next-generation AI to keep our country safe. My own journey—from a high school work-study student to helping grow our lab from a single computer into a full data center—has reinforced a simple truth: in defense, AI is about responsibility. Systems must be reliable, explainable, and resilient because lives and missions depend on them.
The future of defense AI won’t be built by one person or one lab. It will come from collaboration across disciplines, communities, and nations. And as the technology races ahead, we have to stop and ask the toughest questions—the ones that will define whether AI strengthens our defenses or puts them at risk. AV’s Innovation Center was built for answers and there are three high-priority questions we’re working to address every day:
1. As national security missions grapple with the challenge of processing massive amounts of data, how can cutting-edge AI innovations be harnessed to make timely and impactful decisions?
Our missions are becoming more data-hungry, making breakthroughs in data collection, cleaning, and processing for AI an operational imperative. The traditional, time-consuming method of collecting all data and returning it to a central source for processing is no longer viable. By the time the data is ingested, the window for actionable intelligence has often passed. This inefficiency highlights the critical importance of decentralized, real-time data processing.
We are shifting towards architectures where processing and intelligence occur at the edge. This evolution involves deploying AI on low-resource or non-specialized hardware and utilizing it to process diverse raw data streams, such as those from sensors and network traffic, directly on site. The AV Innovation Center is on the cutting edge of this transformation, actively developing and testing these edge AI solutions to ensure timely and potentially autonomous actions in reaction to observations.
Additionally, innovative methods for data processing are essential. In our lab, we have prototyped an AI-driven data normalization process that automates the handling of rapidly changing data. This innovation addresses the ongoing struggle to keep pace with the ever-changing data landscape. The future of AI in defense hinges on these advancements, driving towards a paradigm of scalable, efficient, and real-time data processing that meets mission-critical needs with precision and speed.
2. In the next five years, how far can we push AI at the edge—running on drones, satellites, or sensors in contested environments?
We're actively working to translate the remarkable advances in artificial intelligence, particularly the reasoning and analytical capabilities demonstrated by large language models, into practical applications at the edge. While these publicly available Large Language Models (LLMs) are impressive, their size and general-purpose training present two key hurdles for deployment at the edge. First, their resource demands are often prohibitive for edge devices, and reliable network connectivity can’t be guaranteed. Second, they lack the specialized domain expertise necessary to interpret raw sensor data or contribute to real-time tactical decision-making.
The AV Innovation Center is focused on addressing these challenges by dramatically reducing model size while simultaneously ‘teaching’ those models to natively understand raw sensor data, recognize critical patterns, and enable proactive decision-making. In the next five years, this will lead to a significant leap in autonomous edge capabilities, enabling platforms to make complex, context-aware decisions in the most challenging environments.
3. How do we ensure that Autonomous AI systems behave as expected and remain responsible in critical scenarios?
Building autonomous AI for defense carries a significant responsibility, one we don’t take lightly. We proactively design for safety through monitoring and traceability right from the start. We also build in layers of oversight to keep the AI in check, ensuring it operates as intended. These factors are integral to our development process, ensuring our systems are transparent and accountable.
We establish evaluation frameworks early in the development process to rigorously assess safe operation of our AI systems. To support this, we engage with an AI review board composed of technical, legal, and compliance experts. This diverse oversight provides critical perspectives and helps us navigate complex challenges related to safe and responsible deployment.
These questions are not easy to solve – nor should they be. AV’s Innovation Center works to answer these questions by fostering a culture that welcomes critical thinking, especially when it comes to responsible innovation. We’re looking for people who are not only brilliant engineers but also deeply committed to building AI systems that operate reliably and as intended. By integrating these principles into our innovation process, we aim to develop AI systems that are both cutting-edge and responsible. If this sounds like you, explore opportunities and join me at AV as we ask - and answer - the questions most critical to protecting our nation.
ABOUT AV -- JOINING THE MISSION
AV isn’t for everyone. We hire the curious, the relentless, the mission-obsessed. The best of the best.
We don’t just build defense technology—we redefine what’s possible. As the premier autonomous systems company in the U.S., AV delivers breakthrough capabilities across air, land, sea, space, and cyber. From AI-powered drones and loitering munitions to integrated autonomy and space resilience, our technologies shape the future of warfare and protect those who serve.
Founded by legendary innovator Dr. Paul MacCready, AV has spent over 50 years pushing the boundaries of what unmanned systems can do. Our heritage includes seven platforms in the Smithsonian—but we’re not building history, we’re building what’s next.
If you're ready to build technology that matters—with speed, scale, and purpose—there’s no better place to do it than AV.