Zoom:https://illinois.zoom.us/j/89319046900?pwd=xlVOJYfQUbkMpXa5a9PR0Y4tfcCQtT.1
Refreshments provided
Abstract:
Recent advances in large language models (LLMs) have sparked excitement about their potential to transform engineering practice and education. Today, industry pioneers dream of developing general-purpose AI engineers capable of designing and building humanity's most ambitious projects such as starships and Dyson spheres. However, how close are today's models to this vision? This talk presents my research toward building such general-purpose AI engineers. First, I will introduce EngDesign, a comprehensive benchmark developed by a team of 65 researchers that I organized to systematically evaluate the engineering design capabilities of state-of-the-art LLMs across nine engineering domains. EngDesign pioneers a simulation-based evaluation paradigm that moves beyond textbook knowledge to assess genuine engineering design capabilities and shifts evaluation from static answer checking to dynamic, simulation-driven functional verification. Our findings show that while current models demonstrate impressive progress, none yet reach engineering AGI. In the second half of my talk, I will go deeper into the specific area of control engineering and showcase ControlAgent, an agentic system that bridges LLMs with real-world control design through tool integration, simulation-in-the-loop, and hardware-in-the-loop testing. This case study demonstrates potential solutions for deploying LLM systems for real-world engineering design problems. Finally, I will conclude by outlining the key challenges and opportunities on the path toward building truly general-purpose AI engineers.
Bio:
Bin Hu serves as an Assistant Professor in the Department of Electrical and Computer Engineering at the University of Illinois Urbana-Champaign, and holds an affiliation with the Coordinated Science Laboratory. His current areas of interest include: (1) engineering AGI by integrating LLMs with domain expertise, and (2) learning-enabled control and autonomy. Bin earned his B.S. degree in Theoretical and Applied Mechanics from the University of Science and Technology of China in 2008, and his M.S. degree in Computational Mechanics from Carnegie Mellon University in 2010. He received his Ph.D. in Aerospace Engineering and Mechanics from the University of Minnesota in 2016. Between July 2016 and July 2018, Bin worked as a postdoctoral researcher in the Wisconsin Institute for Discovery at the University of Wisconsin-Madison. In the past, Bin received the NSF CAREER Award, the Amazon Research Award, and the O. Hugo Schuck Best Paper Award.
Faculty Host: Dilek Hakkani-Tur
Meeting ID: 893 1904 6900
Password: csillinois
