| |
| Asana Hosseini Dolatabad | |
| North Carolina State University |
Artificial intelligence is rapidly transforming supply chain management, automating decisions that once required significant human expertise. Yet as algorithms become more capable, a critical question emerges: can optimization alone build supply chains that are sustainable, resilient, and socially responsible? This article argues that human judgment remains indispensable in the AI era. Drawing on recent operations management research, it examines three key areas where humans continue to add irreplaceable value — defining sustainability objectives that algorithms cannot set on their own, providing contextual understanding that historical data fails to capture, and maintaining accountability in an environment where algorithmic authority is quietly growing. Rather than viewing AI as a replacement for human decision-making, this article suggests that the future of supply chain management lies in designing effective collaboration between humans and intelligent systems.
Artificial intelligence is changing supply chains at an incredible pace. Tasks that once relied heavily on managers’ experience and intuition—such as forecasting demand, managing inventory, or coordinating suppliers—are increasingly being supported by algorithms. Companies are investing heavily in machine learning, large language models (LLMs), and automated decision systems in the hope of becoming faster, leaner, and more responsive.
"As this transformation accelerates, however, a bigger question begins to emerge: if algorithms become good enough, what role will humans still play?
"
A recent vision statement published in Manufacturing & Service Operations Management by Cohen et al. (2026) argues that this question is becoming central to the future of operations management. The paper discusses how AI is reshaping supply chains across five interconnected layers: intelligence, execution, strategy, human, and infrastructure. Yet what makes the paper especially interesting is that it does not treat AI simply as a technical tool. Instead, it raises a broader concern: modern supply chains may become increasingly efficient and automated, while at the same time becoming more dependent on human judgment, trust, and oversight.
This article focuses on the “human layer” of that discussion. The key issue is not whether AI can replace managers entirely. In many cases, it probably can automate large parts of operational decision making. The more important question is whether optimization alone is enough to build supply chains that are sustainable, resilient, and socially responsible.
The Sustainability Paradox
One of the most common assumptions about AI-driven supply chains is that better forecasting should naturally reduce waste. If firms can predict demand more accurately, they should be able to produce only what customers actually want. From an operations management perspective, however, the story may not be so simple.
AI systems dramatically improve responsiveness. In industries such as fast fashion, firms can react to changing trends almost instantly. This strengthens what Cachon and Swinney (2011) describe as “quick response” systems, where companies rapidly replenish inventory and adapt products to uncertain demand. But faster response does not necessarily eliminate waste—it may simply move waste somewhere else.
Research by Long and Nasiry (2022) suggests that as firms become faster and more responsive, they often rely more heavily on upstream flexibility. To stay ready for sudden demand changes, firms may hold larger amounts of raw materials, unfinished products, or excess production capacity. As a result, stores may end up with fewer unsold garments, while significantly more fabric, semi-finished inventory, and production resources are wasted earlier in the supply chain.In other words, AI may reduce visible waste downstream while quietly increasing invisible waste upstream.
This is exactly where human judgment becomes critical. AI systems cannot decide on their own what should be optimized. Algorithms simply optimize the objective function they are given. If a company prioritizes speed, profit, and rapid trend adaptation above everything else, AI will pursue those goals extremely efficiently—even if the result is greater environmental pressure and more unsustainable consumption patterns.
That is why humans still matter. Managers define the objectives, constraints, and trade-offs that shape algorithmic behavior. Sustainability does not automatically emerge from optimization. It has to be intentionally built into the system.
The Context Gap
Another reason supply chains still require humans is that real-world operations contain far more context than algorithms can easily capture.
AI systems are trained on historical data and statistical patterns. But supply chains operate in environments filled with uncertainty, incomplete information, and rapidly changing conditions. A manager may recognize rising political tension in a supplier region, understand local labor issues, or anticipate operational disruptions long before those risks appear in formal datasets.
Humans therefore continue to provide something algorithms struggle with: contextual understanding.
Cohen et al. (2026) emphasize that AI contributes pattern recognition and predictive capabilities, while operations management still depends heavily on structural thinking, trade-offs, and contextual reasoning. Even highly accurate systems can fail when environments suddenly change in ways the data did not anticipate.
Ironically, more advanced AI systems do not always make managerial work easier. As algorithms become integrated into operational decisions, employees must spend more time interpreting outputs, evaluating recommendations, and deciding when systems should or should not be trusted. In many cases, the challenge is no longer whether AI can make decisions, but whether humans know when to rely on those decisions.
Trust, Responsibility, and Algorithmic Authority
One of the most interesting parts of the paper involves the problem of trust calibration between humans and algorithms. Research discussed by Cohen et al. (2026) suggests that people often react inconsistently to AI systems. Sometimes employees lose confidence in algorithms after seeing even a small mistake. In other situations, workers become overly dependent on algorithmic recommendations simply because disagreeing with the system feels professionally risky.
Over time, this can quietly shift authority inside organizations from human judgment toward algorithmic recommendations. Employees may follow AI-generated decisions not because they genuinely trust them, but because relying on the algorithm feels safer than challenging it. In highly sensitive supply chain decisions—such as supplier selection, sourcing, or disruption management—this can become problematic. Decisions that are environmentally harmful, strategically weak, or ethically questionable may receive less scrutiny simply because they came from a system that appears objective or data-driven.
The Changing Role of Managers
The paper also discusses a future in which organizations may rely on multiple AI systems working together. Instead of a single algorithm making decisions, firms may increasingly use one system to generate recommendations and another system to evaluate them.
For example, one AI system might recommend a low-cost supplier, while another system reviews whether that supplier creates sustainability risks, geopolitical exposure, or policy violations. In this structure, one algorithm acts almost like a decision maker while another behaves more like a reviewer or supervisor.
Even in this type of system, however, humans remain essential. Managers still define goals, decide which trade-offs matter most, and intervene when systems conflict or fail under unexpected conditions. The role of managers may therefore shift from directly making every operational decision to supervising and guiding networks of intelligent systems.
Conclusion: Optimization Alone Is Not Enough
A simple example from retail workforce scheduling illustrates why human involvement still matters. Algorithms can create mathematically optimal schedules based on labor costs, staffing constraints, and demand forecasts. Yet store managers often adjust these schedules using local knowledge that never appears in the data. They may know which employees work well together, which workers are unreliable, or when customer traffic tends to become unusually heavy. Interestingly, these human adjustments often improve actual operational performance.
This example captures a broader lesson about AI-enabled supply chains. Supply chains are not purely technical systems. They are sociotechnical systems shaped by human judgment, organizational incentives, environmental constraints, and contextual understanding.
AI will almost certainly continue improving forecasting accuracy, responsiveness, and operational efficiency. But the future of sustainable supply chains may ultimately depend less on eliminating humans from decision making and more on understanding how humans and intelligent systems can complement one another effectively.
"AI can optimize supply chains, but it cannot decide what should truly matter. Human managers remain essential because they define priorities, balance trade-offs, and recognize risks and context that data alone cannot capture.
"
References
Cachon, G.P., Swinney, R., 2011. The value of fast fashion: Quick response, enhanced design, and strategic consumer behavior. Management science 57, 778–795.
Cohen, M.C., Dai, T., Perakis, G., Agrawal, N., Allon, G., Boute, R.N., Cachon, G.P., Chen, Z., Cohen, M., Cristian, R., et al., 2026. Om forum—supply chain management in the ai era: A vision statement from the operations management community. Manufacturing & Service Operations Management.
Long, X., Nasiry, J., 2022. Sustainability in the fast fashion industry. Manufacturing & Service Operations Management 24, 1276–1293.
Acknowledgements: We would like to thank Ronak Tiwari for taking time to review this article. Photo credit goes to assistance of ChatGPT for the header and footer photos.
