Opening the “Black Box”: How to make AI explainable for supply chain planners

This is an excerpt of the original article. It was written for the September-October 2023 edition of Supply Chain Management Review. The full article is available to current subscribers.

September-October 2023

Best of the best. Best in class. The elite. Whatever terminology you use to describe the top performers in industry, they all have one thing in common: Companies try to emulate them. That is not easy, of course, but honors such as the annual Gartner Supply Chain Top 25 provide a roadmap for firms hoping to reach the upper echelon. As we do each year here at Supply Chain Management Review, our September/October issue dedicates significant real estate to the Gartner Supply Chain Top 25. Why do we do this? Because our mission is to help inform you, the supply chain practitioner, in all the best ways to make your own supply chains more efficient and…

Browse this issue archive.

Already a subscriber? Access full edition now.

Need Help?

Contact customer service

847-559-7581   More options

Not a subscriber? Start your magazine subscription.

However, in most companies the human is still in charge of supply chain planning and is ultimately held responsible if anything goes wrong. Insufficient inventory? Delayed deliveries? Promotion overlooked? Managers are unlikely to accept the justification that an AI system made the decision. Therefore, human planners need to comprehend the workings of AI, grasp why it makes certain decisions, and intervene if they believe they possess superior insights. However, if interventions become too frequent or turn out to be wrong, the benefits of new planning systems become marginal. As much as AI needs to continue to improve, human planners also need to learn to collaborate effectively with AI and develop trust in it. Understanding and confidence is crucial for an effective partnership between human and AI.

The ability to explain is vital. The AI system must articulate the rationale behind a specific decision or recommendation to ensure alignment with the planner. Creating this understanding for the human may prove challenging given that decisions are often based on thousands of variables, relationships are often non-linear, and calculations are conducted through many layers. In response, the field of Explainable AI (XAI) emerged and made significant advancements in areas with the most urgent needs. For example, in healthcare, AI assists in diagnosing diseases or recommending treatments. In finance, AI automates loan approvals. In human resources, job applications are pre-screened by AI. In each instance, users need to trust AI recommendations, necessitating clear explanations for the decisions made.

In this article we explore how explanations can guide supply chain planners in understanding, and thus accepting, AI-driven decisions. We shed light on the key methods used for creating explanations and provide practical examples to illustrate their application. Our aim is to bridge the gap between planners and AI to enable more seamless, effective AI-enhanced supply chain planning.

AI in SC planning

Many of the major supply chain planning providers have been embedding AI into their solutions. Software suites like SAP Integrated Business Planning (IBP), BlueYonder Luminate, Kinaxis RapidResponse or o9 Digital Brain use machine learning algorithms and AI to improve demand forecasting accuracy, better manage inventory, and optimize supply chain networks.

SC
MR