Companies love to say they are becoming AI-first. A new global survey from Infosys, covering 2,603 employees, shows that message is not reaching the people who actually run the business day to day: middle managers.
The numbers are stark. Only 22% of middle managers said they are actively involved in their company's AI transformation. Senior executives report 49% involvement, and even junior employees come in at 25%, ahead of the managers above them.
Middle managers are not just less involved, they are also less trusted with the tools. Just 13% say they have real freedom to choose how they use AI, compared to 22% of senior leaders. One in five middle managers admit to hiding their true level of AI use from their own company, and they are far more likely to downplay it than to exaggerate it.
None of this is happening in a vacuum. Separate Gallup research tracking workplace AI use since 2023 shows the gap between leaders and everyone else has kept widening, not narrowing, even as overall AI use climbs. Leaders now report frequent AI use at roughly two and a half times the rate of individual contributors, and that gap has grown every quarter.
The timing could not be worse for middle managers specifically. Gartner projects that by the end of 2026, one in five companies will eliminate more than half of their middle management positions, often citing AI as the reason those roles are no longer needed. Oracle alone has already cut tens of thousands of roles, with middle management the primary target.
That creates a genuine trap. The layer of the company most disengaged from AI training is also the layer most likely to be cut to pay for AI investment. And the layer being cut is exactly the one whose job is translating a CEO's strategy into what actually happens on the factory floor, in the call center, or at the sales desk.
A separate LinkedIn survey of 1,252 C-suite leaders adds another layer to this. Half of those executives admit they do not have a clear picture of what skills or roles their own company will need as AI use grows, and most say they are moving faster than they can measure the results.
Put together, this is not a story about middle managers being slow or resistant. It is a story about leadership setting a direction, cutting the people meant to carry it out, and never checking whether the message landed. Any company serious about AI adoption should start by asking its middle managers a simple question: do you actually have the same access, freedom, and training your executives do? For most companies right now, the honest answer is no.