Two years of constant disruption show the value of integrating AI in the supply chain

Unprecented and continued supply chain disruption has become a major business problem that places huge demands on organisational resilience and agility

Many industries face significant upheaval from the spike in energy prices, war in eastern Europe, skills shortages, commodity and shipping cost inflation, port delays, congestion, container shortages and Chinese Covid-19 lockdowns. The seemingly neverending problems represent disruption to the point of disorientation.

The consequences are visible in poorly-stocked shelves and warehouses, drops in manufacturing output, delays in on-time delivery to customers, and loss of revenue. As organisations attempt to satisfy demand they may feel forced to engage in heavy expenditures on expedites like air freight, adding to their cost pressures (and carbon footprint, a growing area of pressure for companies).

While there is no way of preventing unexpected events, integrating AI into supply chain systems can help improve supply chain efficiency and resilience, enabling organisations to manage the disruption far more effectively.

For example, by automating mundane tasks so planners focus on the complex exceptions, their productivity can be increased. Additionally, they can use AI to predict demand increasing the accuracy of the signal that kicks of the rest of the supply chain planning steps. AI can also be fused with optimisation and custom heuristics to address supply allocation, such as maximizing revenue from existing inventory by determining what could be built with the least additional budget.

Organisations integrating AI and advanced analytics have advanced warning of likely events and can respond far more quickly. Take demand-sensing, for example, which uses AI to enhance short-term forecasting by incorporating external signals along with the usual sales history inputs. These external signals could include data from the downstream supply chain, market information, social media and commodity price indices. To give one example of how this works – planners with these inputs can act fast to secure raw materials that become available at a smaller increase in price before the rest of the market piles in.

In a control tower approach, an AI application will monitor disruption signals and prescribe recommendations based on what it has learned from the efficacy of previous interventions.

AI provides a variety of techniques to deal with a wide range of events. A machine learning approach called clustering, for example, will aggregate sales patterns into groups used to adjust forecasts. These clusters offer guidance on how to order and replenish.

As the significant gains AI delivers become better understood, adoption has increased. The MHI Annual Industry Report for 2021 showed an increase in AI adoption from 12% to 17%, with almost another quarter (24%) of respondents expecting to implement AI within a couple of years.

The steps businesses require to maximise the immense potential of AI integration

Integration of AI into existing supply chain management systems does, however, involve many careful steps and wider considerations.

One of the main areas of attention is data. AI needs large amounts of data to make predictions, and while supply chains typically have no shortage of data, it needs to be accessible, prepared for analysis, and attention paid to increasing its quality. AI-adopting businesses also need to ensure they have good processes behind decision-making, overhauling them in preparation. They should also address the data literacy of their workforce, so they can use the technology to best effect.

The human element remains vital. AI research progresses daily, but even as systems become smarter, human input is needed. AI applications and skilled employees should complement each other, rather than create what some call a “lights out” or fully autonomous supply chain that replaces the human. We know AI can find patterns in massive amounts of data beyond the cognitive capacity of homo sapiens, but it lacks the three C’s: it cannot derive meaning from context, collaborate by building and maintaining relationships, or provide a conscience. The goal should be to focus planner domain expertise on situations with high volatility and complexity that we cannot predict, much like the last couple of years.

Human supervision is also necessary as ESG (environmental, social and governance) requirements become more urgent. AI will do what we program it to do, but it has no conscience interested in avoiding modern slavery and deforestation or opting for the most environmentally-friendly use of energy. It needs a guiding hand to address these important challenges.

The implementation of AI should also focus on transparency so it is not a mysterious and threatening black box solution. AI models typically trade precision for interpretability, which means they may be more accurate but with no explanation as to how the results were derived. Planners need AI accompanied with tools to increase interpretability, so they have choices about the decisions organisations make. And they need democratisation in the form of new developments such as automated machine learning that can deliver AI to people without data expertise but who can interact with tools that generate insights for them to use.

In conclusion we can say that the last two years have demonstrated how traditional spreadsheet approaches to supply chain planning and operations struggle when one set of severe and unforeseen disruptions follows another. Businesses need AI so they can see ahead, respond faster and more effectively, equipping their supply chain leaders to make the right decisions and seize opportunities more quickly than competitors. The world is changing fast, and businesses with complex supply chains must adapt.

About the Author

Polly Mitchell-Guthrie is VP of Industry Outreach and Thought Leadership at Kinaxis. Previously, she was Director of Analytical Consulting Services at the University of North Carolina Health Care System and worked in various roles at SAS Institute, in Advanced Analytics R&D, as Director of the SAS Global Academic Program, and in Alliances. She has an MBA from the Kenan-Flagler Business School of the University of North Carolina at Chapel Hill, where she also received her BA in Political Science. She has been very active in INFORMS (the leading professional society for operations research and analytics) and co-founded the third chapter of Women in Machine Learning and Data Science (now more than 90 chapters worldwide).

Featured image: ©Blue-Planet-Studio