AI Transformation Dies in Middle Management
A rational person declining is not a mindset problem.
By Rahul Jindal · 3 min read
Executives sponsor AI transformation. Individual contributors adopt it. It dies in the layer between them, and we keep acting surprised.
Gallup put a number on the gate in its Q2 2025 workforce study, 19,043 employed US adults. Among people whose organisation is actively implementing AI, only 28% strongly agree their manager supports their team using it. Where the manager does, employees are 2.1 times more likely to use AI weekly. The tooling is bought, the mandate is announced, and roughly seven in ten people are looking at a manager who has not backed it.
Look at it from a director's seat.
They are measured on this year's numbers. The AI initiative costs them their best people for two quarters, because the ones who can define the workflow are the ones who understand it, and those are exactly the people they cannot spare.
If it works, the benefit is a headcount reduction that lands on some budget line, possibly not theirs, probably next year, and the transformation office gets the credit.
If it fails, it failed in their function.
That is a rational person declining. Not resistance to change, not fear of technology, not a mindset problem. A correctly-read incentive structure.
Most of what we prescribe for this misses. Change-management workshops do not move a number on a scorecard. Executive mandates produce compliance, which looks like participation and produces nothing.
What works is duller: change what the layer is measured on. Let the function keep the savings it generates instead of harvesting them centrally. Fund the backfill so lending your best person costs you nothing. Put the pilot in the director's own objectives, weighted like everything else.
One company I worked with did exactly one of these, letting functions keep the first year of savings. Volunteer pilots went from two to eleven in a quarter. Nothing about the technology changed.
Harvard Business Review reached the same place in September 2026, writing that the common failure point is not the boardroom, the vendor shortlist or the training portal, but the layer that has to turn ambition into everyday behaviour.
We keep calling this a culture problem, because culture problems are always somebody else's to fix.
It is an incentive problem, and incentives have owners.
What is your middle layer paid to do?
Part of The Seven Conversations, a function-by-function series on enterprise AI transformation.