Scaling AI for Mission Challenges: The U.S. Department of Defense (DoD) needs to scale AI for mission challenges to enhance operational efficiency and national security in an increasingly complex and technology-driven battlefield.

The SEI positions the U.S. Department of Defense (DoD) at the forefront of new developments at the interface of computer hardware and software. Advances in computer hardware have been a major force in the “Cambrian explosion” of AI and will continue to maximize the efficiency—and mission effectiveness—of AI in the near term.

With increased memory and processing power, it is possible to scale AI up, handling larger AI models and datasets and speeding training and inference. At the same time, mission challenges frequently require the DoD to scale AI capabilities down to run on edge devices in disrupted, degraded, intermittent, and low (DDIL) environments like battlefields. As experts in advanced computing, the SEI has a deep and current understanding of the rapidly evolving landscape of computer hardware as well as the skills and experience to adapt AI solutions for both large-scale and edge environments.

Advanced Computing Lab

Our Advanced Computing Lab identifies, evaluates, and applies the latest in AI computing technologies to solve DoD and national security problems. We collaborate with customers, stakeholders, government organizations, and the defense industrial base to improve existing capabilities and identify future capabilities. Our work spans the entire stack: from algorithms to assembler, across the full computing spectrum, from big iron to the edge.

Enabling AI at the Edge

The SEI is leading the software engineering community in envisioning the future of software systems at the edge. The increasing pace of innovation in hardware and AI solutions is enabling a range of software systems to be deployed closer to their users, a paradigm referred to as edge computing. While industry is primarily focused on the hardware and networking aspects of edge computing, much less attention is being given to the software stack that enables mission capabilities, despite the operational uncertainty that is typical of tactical environments.

To fill this gap, the SEI is working to define and improve the software stack for edge systems as well as tools for developing and deploying edge systems. Three primary areas where progress must be made to provide reliable capabilities at the edge, and in turn enable deployment and use of AI and machine-learning (ML) capabilities at the edge, include resource adaptation at the edge, rapid deployment to the edge, and data and data architecting for the edge. More information about the challenges and recommendations for these areas can be found in Engineering of Edge Software Systems.

Heterogeneous Computing Lab

The SEI developed the heterogeneous lab to investigate the new opportunities and growing challenges while using heterogeneous hardware and software in product development. The lab team is exploring techniques and developing prototypes for secure and safe solutions that can take advantage of the multifaceted computing hardware now available that have CPUs, GPUs, FPGAs, and additional processors integrated into a single computing device. Current areas of focus include memory safety, security during boot and normal operation, and developing system and software verification techniques and practices for safety-critical systems operating on heterogeneous platforms.

Quantum Computing

The SEI is exploring the implications of quantum computing for software engineering.