According to Supply Chain Dive, only 12% of supply chain professionals report that their organizations are currently making use of AI in their operations. A slightly larger proportion—28%—said that they were using predictive analytics. And an even larger group of 60% of respondents claimed that they were going to implement AI within the next five years—but even so, those are pretty dismal adoption rates when it comes to technology that has the power to be truly transformative up and down the value chain.
Now, part of the low rate of response might be a factor of the confusion surrounding what actually comprises AI. For some, the phrases “artificial intelligence” and “machine learning” (ML) evoke images of complex, custom-coded R programs that require trained data scientists to work with—or shiny, modern software solutions that cost an arm and a leg. But in reality, AI crops up in smaller, less showy ways all of the time. Supply Chain Dive quotes Stefan Nusser, the VP of product at Fetch as saying: "In my mind, any data-driven, model-based machine learning approach—that to me is AI.” With this definition, we’re betting that some respondents might realize that they’re using AI or ML after all.
It just goes to show that—especially with regard to new technology paradigms—knowledge is power. With that in mind, here are a few things to know about AI and ML when it comes to the global supply chain.
1. ML Can Quantify Uncertainty in the Supply Chain
While AI often gets used as an umbrella term for a nebulous set of technologies ranging from metaheuristics and path search to neural networks, machine learning refers to something more specific: algorithms that take in vast quantities of data to find patterns and correlations that would be invisible to the naked eye. As such, one of the most obvious uses for ML in the supply chain is taking in historical data to create forecasts of everything from future product demand to possible freight prices. Crucially, this process can also be used to gain a better understanding—and, indeed, to quantify—the uncertainty inherent in your forecasted outcomes. This, in turn, gives you the ability to build in elasticity for plans that have a high degree of uncertainty and reduce that elasticity for plans with less uncertainty. This gives planners the power to reactively reduce capital commitments without increasing risk, for example.
2. AI Makes Predictive Maintenance Possible
One of the biggest drains on the resources of manufacturers is machine idle time. An unexpected machine breakdown can set your production plans back considerably and ultimately cost you significantly. An unexpected truck breakdown can do very much the same. While you can certainly do preemptive maintenance on a rotation to try and stave off these unplanned stoppages, it’s difficult to find the right approach and frequency without the help of AI. With AI, however, you can monitor machine usage on an ongoing basis, uncover the hidden signs of an impending breakdown, and proactively schedule maintenance to prevent the breakdowns from happening. As you can imagine, this has the potential to radically reduce downtime and thus increase OEE, giving a boost to your bottom line in the process.
3. ML Can Correlate Lead Times, Throughput, and More to Aid Digital Twins
When we mentioned proactive production planning and scheduling adjustments above, the baseline assumption was that a manufacturer considering AI integration would already have some sort of digitized planning solutions in place—or, at the very least, have moved away from Excel spreadsheets. If that’s the case, we expect that more than a few readers will already be utilizing digital twins (i.e. digital representations of your factory or supply chain created for the purpose of running planning simulations). But did you know that machine learning can make your digital twins that much more powerful? Indeed, because ML algorithms can find hidden connections amid lead time information, throughput data, and other production information, it can more accurately correlate the factors that result in particular outcomes on your factory floor. In this way, you improve your simulations, and thus the plans that are based on the those simulations.
4. AI and ML Have the Power to Increase Supply Chain Resiliency
Supply chain resiliency is, above all else, a matter of responding to change in a way that’s productive and retains as much value as possible. This requires you to have the right systems, processes, and structure in place to monitor supply chain conditions and execute planning adjustments without delay. But it also requires you to perform replannings and create new production schedules, route and tour plans, etc. as rapidly as possible. Artificial intelligence presents one of the most promising paths forward in that department. Why? Because AI can perform analyses and generate planning scenarios radically faster than a human planner can do unaided—meaning that you’ll lose less time when you see a new situation emerging. Of course, this isn’t just a matter of machines taking over for human effort and expertise—user-centric or explainable AI, for instance, is designed to help walk planners through complex reasoning more quickly, rather than spitting out optimizations from a black box. In this way, humans can still bring their own experience to bear in helping to keep things on track in the face of the unexpected.
5. These Technologies Are Already Here
All of the different applications for AI and ML in the supply chain are, of course, important to keep in mind as you consider new technology deployments and sketch out the future of your supply chain. But the most important thing to know about these technologies is that they’re already here. They’re not science-fiction—they’re not “five years down the road.” They’re currently embedded in technology solutions that are already on the market, which means that decision makers at supply chain businesses already have the opportunity to integrate AI into their business processes. The longer you wait to seize this opportunity, the less chance you have of leveraging these technologies as a competitive advantage and gaining a first mover advantage.
