When a company introduces AI tools, leaders often read hesitation as resistance to change.
That explanation is convenient. It makes the problem an employee’s attitude, something training or enthusiasm can fix.
But if you ask someone to show you which parts of their job a machine can do, they may hear a different question: Are you going to help us prove we need fewer people?
That is not technophobia. It is a reasonable question about their livelihood.
Imagine a manager asked to promote new software licenses and find ways to increase adoption. In a one-on-one, the manager asks an analyst to suggest tasks that might be automated. The analyst says he will look into it, then does not bring it up again.
It would be easy to label that response disengagement. It might also be self-protection. If the analyst understands the work well enough to identify what can be automated, he may worry that his knowledge is becoming evidence against his role.
The manager may have the same worry. That is the part leaders often leave out of the adoption message.
You cannot credibly tell employees that a new tool will make their work more productive while asking them to ignore the possibility that it could also change how many people are needed. A cheerful presentation does not resolve the uncertainty. And a promise that no one’s job will change is not useful if you do not have the authority to make it.
So start with work you can put on the table yourself.
A manager might take a recurring administrative task—a report assembled from several dashboards, for example—and try to automate it in front of the team. Show the messy parts. Let the first attempt fail. Explain what still needs judgment, where the output needs checking, and what you do not yet know.
This does not prove that anyone else’s job is safe. It does give the team a different place to begin: not with a demand to expose their own work, but with a leader willing to examine their own.
That example may make it easier for people to offer ideas. It may also reveal that the tool is less capable, more cumbersome, or more dependent on human oversight than the sales pitch suggested. Any of those discoveries is more useful than pretending adoption is a test of loyalty.
There is a limit to modeling, though. A manager automating a disliked task can still look like a demonstration of how fewer people might do more. Employees are not wrong to notice that. If you know what the organization intends to do about staffing, say it plainly. If you do not know, say that too. Give employees a real voice in how the tools change their work, rather than asking them to supply ideas while decisions happen elsewhere.
And do not confuse participation with reassurance. A team can experiment and still be uncertain about what the experiment means for them.
The first work AI should replace is yours—not because that guarantees anyone else’s future, but because asking other people to take the first risk while you keep your own work untouched is a poor way to earn their trust.
Before you ask your team what a machine could do, identify one task you are willing to test yourself. Then tell them what you know, what you do not know, and what decisions are not yours to promise.
That will not remove the fear. It will show whether you are willing to face it with them.
