Walmart workers correct AI tools they're expected to use daily

Walmart's frontline workers are finding themselves in an unexpected role: training the AI systems designed to make their jobs easier.

As the largest private-sector employer in the United States, Walmart has deployed a suite of AI tools to its in-store workforce. But employees report frequently correcting the technology's mistakes rather than benefiting from increased efficiency, according to Business Insider's Dominick Reuter.

The tools often fail to account for real-world complications. When an AI system assigns a worker to restock an aisle, it doesn't factor in that they might first need to clean up a spill or remove expired products. In some cases, the technology creates additional work instead of reducing it.

One example: Walmart's Spark delivery service updated a feature to help drivers navigate stores by tracking their real-time locations. The system began directing workers to collect perishable items like ice cream first—the opposite of efficient order fulfillment.

Why it matters

Walmart's approach represents a broader shift in how companies deploy AI. Rather than imposing tools from the top down, the retailer expects employees to actively shape the systems they use. This strategy can produce breakthrough solutions—one logistics manager built an AI app that helped truckers get home faster while reducing empty trailer trips. But it also exposes two fundamental challenges that any organization scaling AI will face: maintaining employee buy-in when workers fear automation, and managing conflicting feedback from thousands of locations with different needs.

The bottom-up AI strategy

Walmart CEO John Furner has championed what he calls an "AI-for-all" approach. The company believes employees closest to problems are best positioned to solve them, rather than relying solely on centralized directives.

This philosophy has produced notable successes. Logistics manager Leo Garcia created a tool that addressed a specific pain point for truckers, demonstrating how frontline innovation can deliver measurable business value.

Two persistent obstacles

Companies pursuing similar strategies face significant hurdles.

First, maintaining participation. Soliciting employee feedback requires workers to continuously engage with tools they may resent or fear. AI remains controversial among workers who worry about job displacement, making sustained buy-in difficult.

Second, managing divergent input. With more than 4,600 U.S. stores, Walmart serves vastly different markets. A Philadelphia location has different operational needs than one in New Orleans. When stores provide conflicting feedback on the same tool, deciding which input to prioritize becomes complex.

Walmart's vice president for associate tools told Business Insider the company wants to give stores flexibility in how they use AI systems. That flexibility, while valuable, complicates the process of incorporating feedback at scale.

The details were first reported by Dominick Reuter at Business Insider.