The scoreboard, not the mindset
Your team is not resisting AI. They are reading the scoreboard.
Microsoft surveyed 20,000 knowledge workers who already use AI at work. 65% fear falling behind if they do not adapt quickly. And 45% say it still feels safer to focus on this quarter's goals than to redesign how they work. Only 13% say they are rewarded for reinventing their work with AI when the short-term results do not immediately land.
Microsoft calls it the Transformation Paradox. Put more bluntly: we are asking people to take a risk we have quietly priced as a mistake.
From inside the org chart, the resistance disappears
If you are measured on this quarter's deliverables, and an AI experiment that does not pan out counts against you, the rational move is not to experiment. That is not a mindset problem. That is someone correctly reading the incentives in front of them.
Which is why the usual fixes underperform. Generic AI training and appeals to be braver both point at the person — when the block is in the reward system around them.
The tell is in the same data
Leaders are twice as likely as their employees to say reinvention is rewarded regardless of outcome: 21% against 10%. The people setting the incentives believe the safety net exists. The people who would have to fall into it do not.
That shape should be familiar. It turns up wherever a policy is announced from above and experienced from below — the intent is real at the top and unverifiable at the bottom, and only one of those groups ever finds out which.
A test you can run this week
It takes one conversation. Name, plainly, what happens to someone on your team whose AI experiment fails.
Not the policy — the actual answer. What it does to their review, their next assignment, how it gets referred to six months later. If the real answer is "it goes against them," you have found your adoption problem, and it was never their mindset.
The fix is not a bigger training budget. It is making one failed experiment survivable in public, once, where people can see it. Until that has happened the safety net is a claim. After it has, it is evidence.
Redesigning how the work gets done — and how it gets measured — is the actual lever: See how we approach it →
Keep reading: You will lead AI labor · AI Transformation
This is the thinking behind a Trusted AI Culture.
Sources: Microsoft, 2026 Work Trend Index (20,000 knowledge workers). All figures are scoped to people who already use AI at work; non-users were screened out of that survey.