The Signal Comes From Above.

I keep hearing the same complaint from middle managers: "My team won't adopt the AI tools." And I keep asking the same question: what did your leadership do with the savings from the last AI deployment?

Usually there is a pause. Then: "They cut headcount." Or: "It went to margin." Or, most revealingly: "I don't actually know."

That answer tells me more about the adoption problem than any engagement survey ever will.

The one fork that determines everything

When AI frees capacity in an organization - automating tasks, compressing cycle times, reducing manual work - that freed capacity is a single pool. Leadership faces one fork: spend it on building new value, or pocket it as margin improvement. That decision cascades through the entire organization.

This fork is not a philosophical debate. It is a financial decision that every person in the organization reads and acts on. You cannot spend the AI dividend twice. It either funds scale or it funds the P&L. The instinct is to bank first because the return is measurable now. But banking first poisons the cooperation that scaling depends on.

The fractal problem

Here is the part most AI strategies miss. The fork does not just happen at the top. It happens at every level. A team lead receives freed capacity from AI and faces the same choice: invest it in higher-value work, or protect it (hoard headcount, slow-walk adoption, keep the buffer). A middle manager sees the same fork: use the AI savings to fund a new initiative, or hold them as budget slack.

And at every level, people look up to decide which way to go.

This is what I call the fractal incentive: extraction at the top propagates as withholding at every level beneath. It does not matter how well you design the team-level incentive. If leadership is extracting, the signal overrides everything below it.

You cannot fix AI cooperation at the level where it is failing. The signal comes from above.

Why this is invisible to most diagnostics

Standard AI readiness assessments measure skills, infrastructure, data maturity, and sometimes culture. They do not measure where the AI dividend went. They do not ask Finance to show the allocation. They do not overlay headcount data with deployment timelines. They do not check whether the person accountable for an AI workflow has the authority to override it.

Without that data, a diagnostic cannot distinguish between two organizations with identical readiness scores but completely different structural realities: one where leadership is reinvesting and cooperation is building naturally, and one where leadership is extracting and cooperation is decaying despite every change management effort thrown at it.

The extraction path is not just bad for culture. It is frequently bad math.

How to read the signal

There are four questions that surface the extraction signal. None of them appear in a typical readiness assessment.

The uncomfortable conclusion: if your leadership is extracting - banking the AI dividend as margin rather than reinvesting it - no team-level intervention will produce genuine adoption. Not training, not change management, not incentive redesign at the function level. The signal from above is too strong.

The first step is measuring it. Most organizations have never tried.