Artificial Intelligence (AI) Division
Leaders in defense and national security want to obtain the leap-ahead capabilities AI offers. At the same time, it is difficult to get AI right. As many as 85% of current AI deployments fail—failures largely due to the difficulty of replicating, verifying, and validating rapidly developed and deployed AI systems.
The SEI AI Division addresses the need for leap-ahead AI capabilities that are reliable, responsible, safe, fair, and transparent. As part of Carnegie Mellon University, a world leader in AI, we are building on our continuing legacy as world experts in cultivating a discipline of software engineering and pioneers in cybersecurity to help our customers acquire, build, and deliver AI systems that address mission needs.
What We Do
The SEI AI Division conducts research in applied artificial intelligence with a primary focus on AI Engineering, addressing questions related to the practical design and implementation of AI. As our government customers adopt AI and machine learning to provide leap-ahead mission capabilities, we are helping surface leading practices through a community-wide movement to mature the discipline of AI Engineering, leveraging defense and national security problems as a context for definition. In our work, we build real-world, mission-scale AI capabilities and research and define the processes, practices, and tools to support operationalizing robust, secure, scalable, and human-centered AI systems.
Our Virtual Labs
In the AI Division, we accelerate collaboration through virtual laboratories that enable us to work closely with researchers at other organizations, stakeholders, and customers.
Advanced Computing Lab
Hardware is a key enabler for AI, and the hardware landscape is evolving rapidly. 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.
Adversarial Machine Learning Lab
Our Adversarial Machine Learning Lab is working to make machine learning as secure as possible for the DoD and Intelligence Community. We organize our work into a find-fix-verify paradigm, where we find machine learning vulnerabilities by developing new adversarial attacks, fix vulnerabilities by developing defenses and mitigations to known attacks, and verify, within a given system, that vulnerabilities have been properly mitigated via adversarially focused test and evaluation. Our portfolio of work ranges from open collaborations with our colleagues at Carnegie Mellon University to restricted collaborations with DoD and IC sponsors.
Leading the Movement Toward Robust, Secure, Human-Centered, and Scalable AI
We currently lead a community-wide movement to mature the discipline of AI Engineering for Defense and National Security. As part of the National AI Engineering Initiative, we engage collaborators to conduct and shape research and to identify advances in techniques and practices relevant to the AI Engineering discipline.
With our collaborators in the defense and national security community, we have identified three pillars of AI Engineering:
- Human-centered: AI designed to align with humans, their behaviors, and their values.
- Scalable: AI infrastructure, data, and models that can be reused across problem domains and deployments.
- Robust and Secure: AI that works as expected even when deployed outside of closely controlled development, laboratory, and test environments.
Our current partners include
- U.S. Office of the Director of National Intelligence
- U.S. Department of Defense Joint Artificial Intelligence Center
- U.S. National Security Agency
- U.S. Under Secretary of Defense for Research and Engineering
- University of Maryland Applied Research Laboratory for Intelligence and Security
Transforming AI Research into Practice with a Network of Collaborators
The SEI AI Division is uniquely positioned to bring together the research of academia, the creativity of industry, and the mission of government while leveraging the SEI’s own acknowledged expertise in software engineering, cybersecurity, and emerging technology. We are able to bring together experts on the cutting edge of AI research, transform research into best practice, and implement solutions to shape the future of AI.