AI is meant to help workers do their jobs. Some Walmart employees say they’re the ones providing assistance in the relationship.
Walmart’s massive bet on AI includes equipping in-store workers with a suite of tools to make them more efficient. But in some cases, workers are finding themselves frequently correcting mistakes and training the tech, writes BI’s Dominick Reuter.
As the largest US private-sector employer, it’s a great example of the challenges that can arise when trying to effectively get AI tools used by your staff at a massive scale.
Workers told Dominick the tools don’t factor in complications that could arise from a simple assignment. Restocking an aisle involves more than putting merchandise on a shelf. You might first need to clean up a mess or remove items past their expiration dates.
In other cases, the tool inadvertently creates more work.
Spark, Walmart’s delivery service, updated a feature that helps drivers navigate stores when picking up items. But tracking workers’ real-time locations ended up wasting more of their time, sometimes prompting them to pick up items like ice or frozen foods first.
Walmart seems comfortable with hiccups being part of the process.
Executives expect employees to help shape the tools they’re using. CEO John Furner previously said that the AI-for-all approach, rather than a more centralized, top-down directive, will be more effective.
That strategy has elicited some big wins. Walmart logistics manager Leo Garcia told Dominick how he built an AI app that helped truckers get home faster while also reducing the number of empty trailers.
Garcia’s tool is the ideal result of this type of approach: Employees creating solutions to real-world problems instead of management forcing unhelpful tools onto them.
But whether it’s Walmart or another company, putting parts of AI development in the hands of its on-the-ground employees comes with two hurdles:
Maintaining buy-in: Putting tools in employees’ hands to elicit feedback makes sense in theory. But you’re banking on workers continuing to share their thoughts on the tools. Anyone who’s ever set up a survey knows how difficult it can be to get people to respond.
Add in the fact we’re talking about AI — something plenty of people hold a great deal of resentment and fear toward — and you can see how people might have some reservations.
Everyone’s got an opinion. Too much feedback on how a tool should operate is a separate challenge. How do you decide what to listen to and what to ignore? For a chain that has more than 4,600 stores across the US, there are bound to be some discrepancies. A Philadelphia Walmart is serving a different clientele and has different needs than one in New Orleans.
Walmart’s VP for associate tools told BI the company wants to give stores flexibility to use tools how they see fit. Good luck navigating that when stores are sharing diverging feedback for the same tool.
