MIE Seminar: Dr. Theodore Allen, Professor, Department of Integrated Systems Engineering & Courtesy Professor, Computer Science and Engineering, Ohio State University, "AI Safety and Data-Driven Policies through OR Externalizations"

Content

Abstract 

For many years, the discipline of operations research (OR) has helped people to make better decisions in part through externalizing the decision process so that all can approve objectives, assumptions, and associated data. This presentation describes two important roles that the interaction between large language models (LLM) and OR models can play in constraining AI and possibly human agents. First, we illustrate the formulation and solution of planning problems demonstrating how the LLM can not only generate desirable plans but support transparent decision formulations. Second, we show the roles that optimal classification trees (OCTs) can play in making predictions interpretable. These are enabled by our recent advances in formulating and solving OCTs at unprecedented scale to optimality as dynamic programs. Examples from aviation and health care are used to motivate and explain the methods.

Bio

Theodore T. Allen is a Professor in the Department of Integrated Systems Engineering and Courtesy Professor in Computer Science and Engineering at The Ohio State University (OSU). He serves as Associate Director of the Ohio State Institute for Cybersecurity & Digital Trust (ICDT) and as Faculty Director for Industrial Cybersecurity at the Center for Design & Manufacturing Excellence (CDME). Dr. Allen joined OSU in 1996 after completing a full-time internship at Ford Motor Company. He earned his Ph.D. in Industrial and Operations Engineering from the University of Michigan, Ann Arbor, in 1997. His research focuses on the design and analysis of experiments (DOE), including innovations in simulation optimization and metaheuristic techniques. His work has been recognized for both methodological contributions and large-scale practical impact. A 2022 Edelman Academy entry highlighted his development of a two-color ant colony optimization method for vehicle routing, which enabled DHL to realize $205 million in cost savings and reduce carbon emissions by 140 kilotons. His application of DOE principles and work on combating misinformation are estimated to have influenced approximately 400,000 individuals to receive COVID-19 vaccinations. Dr. Allen has received several honors for his scholarship and leadership, including the USCYBERCOM Cyber Recon 2025 Analyst Group Best Paper Award and the 2024 Honda Partnership Award, reflecting his sustained contributions to both academic research and collaboration with national and industry partners.