This site depends on JavaScript to run. Please enable it or upgrade to a modern browser that supports it.

ASCM Insights

AI and the Supply Chain: An Interview with MIT’s Yossi Sheffi


Editor’s note: Yossi Sheffi, Ph.D., is the Elisha Gray II professor of engineering systems at the Massachusetts Institute of Technology, where he also serves as director of the MIT Center for Transportation and Logistics. ASCM Editor-in-Chief Elizabeth Rennie recently had the opportunity to interview Sheffi, ahead of his keynote address at ASCM CONNECT 2023: North America, September 11-13, in Louisville, Kentucky. Learn more and register for the premier supply chain event of the year.

Rennie: Artificial intelligence (AI) is already enhancing supply chain careers. But how can industry professionals ensure that ongoing advancements continue to prioritize jobs that involve deep, meaningful work?

Sheffi: The creation of deeper, more meaningful work depends on businesses recognizing and taking advantage of the unique qualities that humans bring to their jobs. These include flexibility, creativity, critical thinking, communication skills, resilience, and other assets that machines cannot provide. Most involve simply understanding context — an area where humans alone excel.

Rennie: Can you share some examples? And how can supply chain professionals tap into these uniquely human skills in order to advance in their organizations and careers?

Sheffi: First and foremost, a lifetime of experience in the physical world gives people the ability to detect changes or discrepancies between normal and abnormal situations. For example, during the financial crisis of 2008, companies worried about the financial health of their suppliers. Many asked for financial data from suppliers, but these numbers could be manipulated and only provided a lagging, infrequently updated view of conditions. To augment the data, many companies sent people to spot-check key suppliers’ production of parts or materials. Simply by walking through a supplier’s offices and factories, the visitor could gauge its financial health through observing the existence of too much or too little inventory, the bustle or silence of the facilities, and the emotional states of the workers.

People are also more adaptable than robots when faced with unstructured conditions and environments. Any robot or software system is built and optimized for a specific set of tasks or a specific domain. However, change — whether disruptions, new knowledge, new products, competitors’ actions, economic cycles, and so forth — can render the machine’s appropriateness moot, and then a person has to take over the task. Moreover, in the social context of an organization and supply chain, crisis management teams can create new organizational structures and new collaborations to deal with one-off challenges, such as the 2011 earthquake in Japan, the COVID-19 pandemic, and the loss of manufacturing capacity due to a flood or a factory fire. Basically, computers can and do collaborate, but they do so through programmed protocols, while people can adaptively and quickly create new ways of collaboration when and as needed.

Rennie: You’ve mentioned a lot of extreme supply chain disruptions: COVID, severe weather, the financial crisis. How can AI systems help supply chain organizations with their risk tolerance?

Sheffi: AI-based systems can, in many cases, offer a range of possible actions. Each one may come with a probability that a certain objective will be met — things like gaining the maximum profit, creating the safest outcome, having the lowest emissions and so on. But even if there’s only a single objective, and no social or moral considerations are present, a decision may still be a matter of judgment regarding risk. Should you take the high-risk-high-reward option or the safe one? Or maybe something in between — not totally safe, but potentially with a better-than-minimum reward? While rules can be programmed, the most appropriate choice may be different than what the rules suggest. Again, risk-tolerance is all about context and is therefore something that people are likely to be involved in for the foreseeable future. So, while the machine crunches the numbers, final decisions are made by people. In general, the more consequential the decisions are, the more likely it is that people will call the shots.

Rennie: So, if the more consequential decisions need a human touch, what about activities that touch the customer? There are a lot of AI-based solutions interacting with customers and really controlling the customer experience these days — some much more successfully than others. In general, do you think that’s an effective choice?

Sheffi: While a growing number of AI applications are being used in consumer-facing environments, they cannot replicate human empathy and communication. For example, few contract negotiations can be successfully completed without both sides understanding each other, developing rapport and appreciating each other’s point of view. Such qualities are particularly evident in cases in which systems fail and people need to overcome their difficulties by working together.

But, as you say, there are some AI-based programs starting to surpass those limits. A class of emotionally intelligent (EI) chatbots could help businesses improve their ability to share information, collaborate with internal and external resources, and address the needs of customers. Some programs can even capture human cognitive states and emotions by analyzing facial and vocal expressions.

Rennie: Before we wrap up, I’d like to go back to the previous point you made about context. Here at ASCM, we often talk about making an impact through ethical supply chains. Obviously, ethics demand taking into consideration context. Can AI help with that?

Sheffi: Many supply chain tasks involve value judgments, moral understandings and subjective elements, which not only change over time, but also vary from one team to the next. In many cases, machines may be able to sift through large amounts of data and present options for actions, but again, people have to make the ultimate decisions, especially in cases where the implications of those decisions matter. This is especially true when context changes and decisions must be made in a different environment. For example, when prioritizing the response to a disaster, should preference be given to customers, employees, suppliers, shareholders or the community? People, embedded in the human experience of personal life, — family, friends, coworkers, customers, and communities — are likely to be better than machines at understanding and judging a response that’s both economically viable and socially responsible.

Listen to the latest episode of The Rebound with Yossi Sheffi, Yossi Sheffi and the Magic Conveyor Belt. 

About the Author

Elizabeth Rennie Editor-in-Chief, SCM Now magazine, ASCM

Elizabeth Rennie is Editor-in-Chief at ASCM. She may be contacted at

Use of Cookies

We use cookies to personalize our website’s content and ads, to provide social media features and to analyze our traffic. We also share information about your use of our site with our social media, advertising and analytics partners who may combine it with other information that you’ve provided to them or that they’ve collected from your use of their services. You consent to our cookies if you continue to use our website.