MHI Solutions Improving Supply Chain Performance
Augmenting Supply Chain Talent with Intelligent Tools
COMPANIES ADOPT AI to boost efficiency, but many fall short because they overlook a critical factor: people.
“People put too much focus on the technology and forget to focus on the people training, the people upskilling, on changing the process as well as the organization and the mindset,” Bart De Muynck, a supply chain industry expert and advisor, said.
To achieve the best results, companies should involve employees who will be affected by AI early in the planning stage of its implementation. When employees experience how AI can simplify their work, they’re more likely to adopt the technology.
SET THE STAGE FOR COLLABORATION
To get human‑AI collaboration off to a good start, companies should first demystify the technology. “Some of the companies that we’ve worked with actually do workshops to teach people the basics of AI,” Andrea Morgan‑Vandome, chief innovation officer at MHI member Blue Yonder, said. One critical element is explaining how AI arrives at the answers it produces, because people are more comfortable with the technology when they understand how it works. Education helps drive excitement about bringing it into the workplace.
Companies also need an overall AI strategy and a culture of innovation that encourages employee input in its implementation. “It’s not just leadership saying, ‘We’re going to use AI; we’re going to push it down your throat.’ No, you have people come up with ideas, you upskill them, you empower them. That is a lot more successful, and people then feel a lot more valued in the organization,” De Muynck said.
Employees should be involved in change management—an essential yet often overlooked step in successful AI implementation. The people who do the work every day are best positioned to review current processes, identify mundane tasks that AI can handle and differentiate them from more complex tasks that require human intelligence or institutional knowledge.
“What I always recommend is committing to use AI for some very basic tasks like brainstorming or writing emails,” Vince Castillo, assistant professor in the Fisher College of Business at Ohio State University, who includes AI in undergraduate and executive education classes, said. “You can also use AI for background research and for getting up to speed on certain topics.” In these early stages, AI is best used for repetitive, frequent tasks.
For example, many companies still require a truck driver to get a physical document signed and returned when they make a delivery. “Maybe there’s an AI application where the driver can take a photo of the document so that all of the content can be extracted. Then there’s AI that can pull that information, store it and go into it for any sort of analysis,” Castillo said.
The data can be accessed and used for reports more quickly than if the driver had to return to an office and hand in a physical document.
UNDERSTANDING AI INTERACTIONS
There are many types of AI, and to work successfully with any of them, humans need to be aware of their capabilities and limitations.
Castillo is allowing supply chain college students to interact with one type of AI. Shortly after ChatGPT’s 2023 release, he began incorporating generative AI (GenAI) into the curriculum for the machine learning class he teaches. Students can use AI however they want as long as they are transparent about it.
“I wanted to see what they could learn and figure out about using the technology,” Castillo said.
For one assignment, students build an interactive supply chain map that incorporates data on factories, ports and customer markets worldwide. Their work must include elements such as node linkages, different shipping modes and the demand for different SKUs during specific time periods.
“It’s really a fun, exciting assignment and it’s a tough one. Before generative AI, I never even attempted it, because for that kind of product you usually need programming skills,” Castillo said. “My goal has been to teach the students how to lead the generative AI model to write the code for them.”
The exercise changes the skill sets students need; they think in terms of strategic orchestration rather than learning programming.
They also begin to understand that AI isn’t always the best answer. For example, students who let AI write the code for some analyses often get tired of correcting AI’s work and decide to write the code themselves.
“I realized that’s how students are going to continue to learn in this world of AI,” Castillo said. “They need to learn not only what AI is and what it can do for them, but also where it is limited. We’re also going to be able to see what skills are still going to be relevant in this new age of AI.”
Castillo said his students have learned that GenAI is great for brainstorming and for gaining surface‑level knowledge. But the answers it provides are often generic, and students who let the machine do their thinking for their reports don’t do well.
“There’s a very real risk of what we call cognitive outsourcing, where you can get something that on the surface level appears to be very valid, but the moment you start looking at whether it actually fits the context, it all falls apart,” he said. “Students have developed a healthier respect for what generative AI can do and a better understanding of where they have to step in and still be that human in charge.”
Castillo also teaches an executive education class on AI. Participants have a wide range of AI experience, but all are eager to learn more about what it can do. While this class offers a more strategic, high-level view of AI’s capabilities, executives also get some hands‑on experience. In one exercise, they learn how to use a large language model (LLM) to access information in logistics routing guides that can contain hundreds of pages.
Castillo said some executives are frank about their desire to use AI to reduce headcount, but he tries to get them to view the technology differently. “I understand there’s a temptation for that, but a better way of thinking about generative AI, and where the real value propositions are, is in augmenting, not automating,” he said. “There are processes that should be automated, but the intent is to augment some human users and make them much more capable or more efficient at things that they’re already doing.”
These classroom lessons will benefit both students and executives when they begin using real‑world AI applications in their jobs.