The Systems Engineering Research Center (SERC) co-organized an AI-focused workshop with the International Council on Systems Engineering (INCOSE) on October 11-12 and provided a dynamic platform for an international assembly of experts to exchange ideas and insights. The two-day virtual event, AI4SE & SE4AI, followed the annual in-person workshop the SERC co-hosted in September with the Army DEVCOM Armaments Center Systems Engineering Directorate.
“We’ve been at this for about four years now, and it’s been amazing to watch the growth in terms of research and applications and tools in this domain,” said SERC CTO Tom McDermott, referring to the burgeoning enthusiasm for AI and its intersection with system engineering. “We had about 20 presentation abstracts the first two years that ballooned to 34 last year, and then we had 72 this year. We have this opportunity to do both a live and a virtual event because there’s so much interesting work going on, and many of you want to present that.” More than 50 people presented their work at the two events.
Several speakers touched on AI’s potential to streamline processes, boost efficiency, and guide decision-making in system engineering. Others focused on the challenges of applying system engineering principles to AI systems.
“As we delve into this realm, we find ourselves confronting a significant gap in the field of system engineering for AI,” said Ms. Niloofar Shadab, a Ph.D. candidate at Virginia Tech. Her session on “Closed Systems Engineering for Artificial Intelligent Systems” presented research she conducted with SERC researchers Dr. Peter Beling (Virginia Tech) and Dr. Alejandro Salado (University of Arizona).
To address the “chasm” and the “lack of emphasis on the scalability and scoping of intelligence within engineering practices,” Shadab’s team proposed a solution based on closed systems principles. Their approach offers a fresh perspective on engineering AI systems by considering their closed nature and the intricate coupling between the AI system and its context.
Dr. Barclay Brown, associate director of AI research at Collins Aerospace and chair of INCOSE’s AI Systems Working Group, was one of many speakers to delve into the significance of large language models as vital components in AI applications. In his session, “Hiring Trained Animals: Generative AI Patterns and Practices for Systems Engineering,” Brown proposed implementing and combining techniques like retrieval augmented generation, prompt engineering and fine-tuning as integral elements that bridge the gap between data and language to enhance system intelligence.
“Large language models can be a component in an application, a component in a system — and the patterns for using large language models continue to evolve,” said Brown, who also serves as INCOSE’s CIO. “These patterns keep emerging, and we’re trying to figure out what’s practical and workable.”
The integration of various AI techniques and algorithms was also discussed in the “Streamlining Requirement Management with AI” session by Michael Jordan of SPEC Innovations. His research underscored the transformative potential of AI in requirement management, acting as a catalyst for meaningful integrations and decision support while providing “contextual guidance” throughout the process. “However,” he insisted, “it’s vital to emphasize that AI is not here to replace humans in this process, but to assist and complement ensuring a symbiotic relationship where human validation is still required.”
As the boundaries between human expertise and machine capabilities blur, the pivotal question guiding systems engineering research has become clear: How can practices be aligned for a future where humans and machines seamlessly collaborate to fulfill their objectives? “That’s why we’re here,” said McDermott. “That’s why we have a working group in INCOSE, and that’s why we are doing workshops like this.”
Download the presentation slides from this workshop on the event page. Download the presentation slides from the September workshop on that event page. Follow SERC on LinkedIn for regular updates on systems engineering research.