Operation CPG Leaders Still Don’t Trust AI With Forecasting

At an interesting panel on AI in CPG operations a recurring theme reitirated just how misunderstood operations really are — even inside large organizations. Several panelists noted that many executives underestimate the complexity of running global supply chains, where thousands of decisions are interlinked across geographies, vendors, regulations, and timelines.

That complexity has only increased. Geopolitical conflicts, shifting tariff regimes, and rapidly changing regulations now introduce macro-level instability that operations teams must absorb in real time. For COOs, uncertainty is no longer episodic — it’s structural.

Despite the uncertainty the panelists unanimously agreed that forecasting models are the only viable tool to try to remain slightly ahead of the curve. These models aren’t perfect, but they’re the only mechanism available to stay ahead of demand shocks, supply disruptions, and inventory risk.

As the discussion was about AI tools you would expect AI to excel here. Yet the entire panel reached the same conclusion: they have not encountered an AI forecasting model they trust.

AI is being used — but tactically. Panelists described leveraging AI for productivity and narrow operational tasks, such as resolving shipping disputes with carriers or streamlining internal workflows. These are bounded problems with limited downside.

What AI is not being trusted with is the heavy modeling that keeps the business running. The reason is simple: if forecasting breaks, the entire operation can grind to a halt. Trust matters more than theoretical accuracy.

TheMarketAI Take

We’ve touched on this before: AI struggles to scale to high-volume, high-stakes environments while maintaining low variance.

Operations forecasting sits at the intersection of messy real-world data and rigid computational requirements. The inputs are complex, obscure, constantly changing, and deeply contextual. This is where probabilistic systems face their hardest test.

The panel was cautiously optimistic that AI will eventually solve this. That may be true. But for now, AI in CPG operations lives at the edges — helpful, incremental, and contained — while the core still runs on models humans trust enough to bet the business on.

We’ll wait and see.