Formalized artificial intelligence (AI) engineering practices will help national defense and security agencies adopt AI in a way that is repeatable and scalable. As part of its efforts to grow the field of AI engineering, the SEI is bringing together members of the AI community. The SEI, Duke University, SRI International, and MIT Lincoln Laboratory organized the AI Engineering Symposium in March 2022 for AI researchers and practitioners.

The symposium, part of the Association for the Advancement of Artificial Intelligence (AAAI) Spring Symposium Series, focused on human-centered, scalable, and robust and secure AI. The symposium’s goal was to evolve the state of the art; foster critical relationships; and gather lessons learned, best practices, and workforce development needs in the area of AI engineering.

The AI Engineering Symposium drew participants from fields including robotics and computer science and from organizations including MIT Lincoln Laboratory, University of Maryland Applied Research Laboratory for Intelligence and Security (ARLIS), the Air Force Research Laboratory (AFRL), Clemson University, and Carnegie Mellon University. Representation from academia, industry, and the defense and national security spheres enabled the kind of interactions needed for a cross-domain AI engineering discipline.

The event proceedings expand the AI engineering body of knowledge and practice with papers on adaptive autonomy, the DevOps lifecycle, human-AI interaction measurement, synthetic training images, hazard analysis processes, kernel density decision trees, human-AI teaming, and measuring beyond accuracy.

The SEI is leading a National AI Engineering Initiative with funding and guidance from the Office of the Director of National Intelligence (ODNI). AI engineering combines the principles of systems engineering, software engineering, computer science, and human-centered design to create AI systems in accordance with human needs for stakeholder outcomes. Events like the AI Engineering Symposium establish shared language across multidisciplinary researchers, outline the field’s progress and open questions, foster collaboration, and provide a forum to address the challenges of applying theoretical concepts in complex Department of Defense settings.

Read the AI Engineering Symposium proceedings at https://resources.sei.cmu.edu/library/asset-view.cfm?assetid=884163.