The Computing Community Consortium (CCC) is very pleased to announce the release of the final report in the workshop series on enabling research collaborations between Artificial Intelligence (AI) and Operations Research (OR) experts. This workshop series, first proposed in 2020, recognized the significant progress made in Artificial Intelligence research in recent years, especially in areas such as machine learning. However, it also noted the advances in optimization and decision-making made within AI, areas where the deep expertise of Operations Research is invaluable. Recognizing the complementary strengths of these two fields and the potential for significant societal impact through their synergy, the workshop series aimed to bridge this gap.
“From the outset, we envisioned this series as a crucial step towards fostering a deeper understanding and collaboration between the AI and OR communities,” said Ramayya Krishnan, co-author, Dean of the Heinz College of Information Systems and Public Policy at Carnegie Mellon University, and 2019 INFORMS president. “We believed that, by bringing together experts from both fields, we could unlock innovative solutions to complex problems. Operations Research brings a rich history of modeling, optimization, and statistical rigor that is crucial for grounding the exciting advancements in AI.”
Sven Koenig, co-author, previous CCC council member, and Chancellor’s Professor and Bren Chair in AI at the University of California, Irvine, added, “Techniques from AI and Operations Research are often complementary. Knowing about techniques from both disciplines helps one to build better intelligent systems for addressing real-world challenges, especially those that need to make smart decisions under complex constraints, because one has more tools available to model the involved decision problems, solve them efficiently, effectively, and robustly, and provide explanations to users.
CCC held three workshops in this series, in 2021 (held virtually due to the pandemic), 2022, and 2024. The first workshop focused on identifying barriers to collaboration between AI and OR researchers and providing recommendations for overcoming these barriers. The second workshop envisioned strategies for employing OR methodologies to develop AI systems that are trustworthy and robust. The third and final workshop of the series focused on specific application problems whose solutions would benefit from a joint AI and OR approach. These “Challenge Problems” were selected from an open call to the AI and OR communities, and were combined into seven topic areas, including:
1.) AI/OR to address the opioid epidemic and causal inference
2.) Large language models (LLMs) and OR
3.) Multi-agent interactions, learning, and risk mitigation in AI-based supply chains
4.) OR for addressing data issues
5.) Integrating OR and AI through the optimization lens
6.) AI/OR trainers for trustworthy training of non-expert humans and AI agents
7.) The tradeoff between optimality and explainability
“It was such a great experience to highlight the risks of multi-agent interactions in supply chain markets during the third workshop. AI and Operations Research, working together, are key to understanding and mitigating these risks,” said Gianluca Brero, co-author and Assistant Professor of Information Systems and Analytics at Bryant University. “Contributing to the final report felt like a natural extension of the fruitful discussions we had at the workshop, and I’m really looking forward to seeing the impact of this report on fostering future research collaborations at the intersection of AI and OR.”
The final report, Making a Case for Research Collaborations Between Artificial Intelligence and Operations Research Expert, outlines recommendations from all three workshops. These recommendations target all stakeholders engaged in AI and OR research, including funding organizations, research institutions, conferences, non-profits, and researchers themselves. Read these key recommendations in the Strengthening AI and OR Collaboration Key Recommendations one-pager.
“Our hope is that this report serves as a catalyst for real change,” stated Radhika Kulkarni, co-author, retired Vice President of Advanced Analytics R&D at the SAS Institute and 2022 INFORMS President. “We believe the recommendations provide a practical roadmap for building stronger bridges between AI and OR, ultimately leading to more impactful research and solutions for pressing societal challenges.”
We would like to thank each of the final report authors, listed below, and we hope this report will be just one step in a larger effort to foster meaningful and impactful collaborations between the AI and OR communities.
Report Authors: Radhika Kulkarni (SAS Institute, Inc. (retired)), Gianluca Brero (Bryant University), Yu Ding (Georgia Institute of Technology), Swati Gupta (Massachusetts Institute of Technology), Sven Koenig (University of California, Irvine), Ramayya Krishnan (Carnegie Mellon University), Thiago Serra (University of Iowa), Phebe Vayanos (University of Southern California), Segev Wasserkrug (IBM Research), and Holly Wiberg (Carnegie Mellon University).
