Loyola University Chicago

Loyola Business Leadership Hub

Housed in the Quinlan School of Business

Supply chain leaders discuss the power of A.I.


Left to right: Ray Nair, VP of Supply Chain at CDW; Brook Ruckinski, Sr. Director of Inventory Management, W.W. Grainger; Tom Siegmund, Founder and Chairman, Harvest Sporting Group.

By students: Matt Hyatt, College of Arts and Sciences Class of 2024 and Christian Piantanida, Quinlan School of Business Class of 2022

If the COVID-19 pandemic has taught us anything, it is to expect the unexpected. Stochastic events like the pandemic are precisely what business executives around the world fear and why many companies are striving to be flexible and nimble. Not only do these events impact each business individually, but they have the potential to disrupt entire distribution networks and supply chains.

To understand how innovative technologies can be leveraged to detect and respond to supply chain anomalies, Loyola University Chicago’s Business Leadership Hub hosted the conference Designing Flexibility to Address Uncertainty in the Supply Chain with AI in June 2022. The event brought together senior supply chain executives from a variety of companies including Amazon, CDW, Dollar General, Griffith Foods, Kohl’s, Kroger, McKinsey, Microsoft, W.W. Grainger, and Walgreens, among others. Quinlan faculty who participated included Cerag Pince, Associate Professor; George K. Thiruvathukal, Professor of Computer Science and Deptartment Chair; Maciek Nowak, Professor and Interim Dean; and Michael Hewitt, Professor and Faculty Director of the Supply Chain and Sustainability Center. Discussions centered on global supply issues and how companies can build sense-and-response systems by leveraging data and AI.

The sold-out conference was outstanding, and we look forward to the second annual AI in the Supply Chain event planned for May 18, 2023, in Chicago. For information on that event, please contact Steven Keith Platt, Director of Analytics and Lecturer of Analytics and Applied AI,  Quinlan School of Business- splatt1@luc.edu.

Here are some key takeaways from the conference:

AI must be interpretable for it to be useful. 

Some speakers said their teams opt for simple solutions over complex ones, especially in cases where accuracy may come at the cost of explainability. People do not trust what they don’t understand, and a confusing AI model may be rejected. Simplicity makes all the difference in adopting intelligent forecasting tools.

Consistency is also essential for stable businesses. A few speakers agreed that to promote stability and predictability, they would opt for a model with 50-80% accuracy over a model with 70-75% accuracy.

Learn AI processes over algorithms.

AI is becoming increasingly used in the workplace. One day, it might be considered a soft skill, similar to the adoption of Excel. While it may not be necessary to learn every strategy in building intelligent systems from the ground up, grasping the general process and limitations of AI will keep businesses from losing ground and help leaders work in partnership with their data scientists and engineers.

The same solution will not always work.

Rajiv Malik, EVP of Sourcing and Product Services at Kohl’s Corporation, explained how fashion trends follow this rule: what was trendy last season will likely be out of fashion the next. In other words, the same solutions will not always work. This rule applies to companies and their supply chains. To be robust and prepared for change, you must be aware, innovative, and adaptable.

Know when to deviate away from the rule.

Blindly following AI may not lead to good decision-making. But it can provide a great generalization of trends. Anomaly detection—the identification of rare events—is a helpful tool for spotting when things go awry. It allows management to know when to change course even if they choose not to follow algorithmic decisions.

Plan for uncertainty. 

The keynote speaker, Warren Powell, Co-founder and Chief Analytics Officer of Optimal Dynamics and Professor Emeritus at Princeton University, suggests that managers need to understand the complexity and variability within a supply chain to begin making informed decisions. Due to uncertainty about inputs and events, the traditional point forecasting methods no longer suffice. Therefore, a distributional forecasting method is more appropriate for making effective supply chain decisions. “You cannot prepare for uncertainty unless you model uncertainty,” says Powell, “Distributional forecasting is the foundation of planning into an uncertain future.” 

The jungle of stochastic optimization.

Powell defines “The Jungle of Stochastic Optimization” as 15 fields that deal with decision-making under uncertainty. These fields are fragmented, exhibiting their own unique tools and methods for specific problems. To successfully address sequential decision problems involved with supply chains, teams require individuals with a broad understanding of all 15 fields. “Real applications require skills that span a wide range of problem settings,” says Powell. However, finding these individuals to hire could pose a challenge since most universities still teach the unique, highly specialized AI techniques instead of also teaching a broad range of application skills.

Go deeper: