Shared Goals

By the time this column reaches you, the INFORMS elections will be complete and the new board members announced. Congratulations to the new members – as of this writing, I do not know who won; hence, I am not including any names! INFORMS appreciates your dedication to our profession and is very pleased to welcome you to your new roles on the board. I also appreciate the members who raised their hand to be considered as a candidate. I would also like to thank all the members who voted in this election – this is an important right and responsibility that you have as a member of INFORMS. So, my appreciation to all of you as well.

If you are interested in volunteering to serve on the board, there will be three positions up for election in 2023: President-elect, VP of Membership and VP of Marketing, Communications and Outreach. For more information, you can visit informs.org/nomination.

In this issue, I would like to share some of my thoughts following the second AI/OR workshop and some musings about ways in which our profession can make a difference as we face myriad societal challenges.

In-person AI/OR Workshop

In my April column [1], I talked about the plan to have two more AI/OR workshops following the success of the first virtual workshop held in September 2021. I am excited to report that the second workshop was held in person on August 16-17 in Atlanta. All the organizers (INFORMS, CCC and ACM SIGAI) believed that this workshop should be in person to allow for robust networking and discussion opportunities. And, indeed, the outcome clearly surpassed our expectations. All the speakers and participants had ample opportunity to debate and discuss the various topics on the agenda [2]. We will share a report-out of the workshop when it is ready. In the meantime, here are a few highlights.

The goal of this series of workshops is to establish a joint strategic vision for artificial intelligence (AI) and operations research (O.R.) that will maximize the societal impact of AI and O.R. in a world that is facing myriad significant challenges. This year, four sessions focused on the foundational elements of trustworthy AI: fairness, explainable AI/causality, robustness/privacy and human alignment/human-computer interaction. The speakers and participants were selected from computer science and O.R. communities to foster a healthy exchange of ideas between the two groups.

I am excited about the ideas discussed during the wrap-up session and hope that they will lead to future collaborations along the following lines:

• Some grand challenges for research areas that would spur collaborative research.
• Summer school for cross-disciplinary learning.
• Hackathons involving teams from AI and O.R.

Funding Opportunities in the AI/OR Space

One of the goals of the AI/OR initiative has been to promote funding opportunities from agencies such as the National Science Foundation (NSF) and National Institutes of Health, or federal agencies such as the U.S. Department of Transportation (USDOT), Department of Defense, etc. We made progress in this regard at the workshop:

• We invited Dr. John Abowd from the U.S. Census Bureau and Dr. Murat Omay from USDOT to share opportunities for research in their respective organizations. The workshop provided a good platform for the participants to learn about the large number of problems that can benefit from our combined expertise and for their organizations to understand the background and experience of the researchers in our fields. Our goal is to hold the third workshop in the Washington, D.C., area in early 2023 to increase participation from more federal agencies to learn about potential research collaborations that can help address multiple societal problems.
• One of my personal hopes has been that the AI initiative will provide ideas for cross-cutting funding opportunities from NSF and other agencies. I would like to give a shoutout to members of the INFORMS NSF Liaison Committee who accepted our invitation to attend this workshop and were excited to pull together a visioning workshop proposal involving a multidisciplinary team. Being in person enabled these members to meet with workshop participants and plan next steps. If you are interested in this topic and are reading this column before the 2022 INFORMS Annual Meeting, I encourage you to attend the panel discussion on Tuesday on AI/OR: Research and Funding Opportunities [3].

Key Takeaways from the Workshop

I personally took away many interesting ideas from the talks and breakout sessions at the workshop. Here are a few that are top of mind:

• The AI and O.R. communities have many shared goals but often use different terminology. We can benefit from a common language and more cross-discipline training. This workshop (as well as the first one) clearly shows the potential value in collaboration between the two groups. The AI community has focused on scalability and automation, whereas the O.R. researchers place greater emphasis on optimality guarantees and “humans in the loop.” Another key difference is the way each community has traditionally approached the decision-making problem: AI takes a data-driven approach and O.R. favors the model-driven one. However, there is increasing emphasis on a model- and data-driven approach in both communities, which is necessary for innovative advances in AI research.
• Given the overall theme of the workshop, many of the talks referred to “explainable AI,” especially in the context of trustworthy AI. An interesting question that came up during the Q&A sessions was why there is no “explainable O.R.” I would say that we have many tools in the O.R. arsenal that are widely used to show that the solutions we find are optimal or near optimal. We rely on sensitivity analysis, shadow prices, simulation and “what-if” analysis to show that our model gives solutions that are better than other possible solutions to the problem. As practitioners, we have always needed to demonstrate that our model fits the problem and “explain” why our solution is the best. Isn’t that explaining O.R. models, or “explainable O.R.”?

Societal Responsibilities

The past couple of years have been challenging, with complex issues that have affected our lives in many ways: slow emergence from a (continuing) pandemic, a completely unexpected war that is still ongoing, and many political conflicts that have caused us to evaluate how we make daily decisions, including whether to attend a conference in person or virtually.

The INFORMS Board and staff are acutely aware of these concerns and will be engaged in discussions over the next few months on choices that are within our control. I find comfort from knowing that our members have the knowledge, expertise and experience to look at complex issues and use data- and model-driven approaches to recommend solutions to these problems. The challenge comes from the fact that any given topic has multiple points of view. Many social issues are multifaceted and cannot be easily expressed in terms of clear preferences, nor can they be assigned cost/benefit values using a common currency. However, our community has tackled many complex societal issues and provided valuable input to rulings related to many serious decision problems, including organ donation decisions, transportation security, Federal Communications Commission (FCC) regulations and more.

The key is for us to become engaged in these issues and tackle them with the tools we know best. As a community, let us strive to offer support as O.R. professionals and provide a data- and model-driven, balanced view to policymakers and those in charge of making the critical decisions that affect so many people. Let’s look for ways to engage and make a difference! Our tools are powerful and have been used in many ways to save lives, save money and solve problems. If there are decisions that are being made but don’t seem fair, use your expertise to explore the issues, gather data and analyze the situation; you may even be spearheading some new areas of research that can make a difference!

I will end my column with best wishes to everyone for the upcoming Annual Meeting. I hope to see many of you in person in Indianapolis!

References

1. Kulkarni, 2022, “Internal and interdisciplinary growth,” OR/MS Today, April 13, https://doi.org/10.1287/orms.2022.02.12.
2. https://cra.org/ccc/events/artificial-intelligence-operations-research-workshop-ii/#agenda
3. https://tinyurl.com/bda7azwu