Most AI change programmes start with the same assumption.
Employees will resist. So the plan focuses on them: training sessions, communication plans, adoption dashboards, champions in every team.
I believe that assumption is backwards.
In many organisations, employees are more ready for AI than their leaders think. The real slowdown often happens at the top.
Leaders demand change from everyone but themselves
Here is a pattern I have seen many times.
The leadership team announces an ambitious AI programme. Money is approved. A vendor is chosen. The launch is exciting.
A few months later, the results are not what was promised. Everyone two levels down can see why. But nobody says it out loud.
Why? Because the leaders who championed the programme are too invested in it to admit it isn’t working. And everyone else is afraid to challenge an idea the bosses have already backed.
So the project stays “active”. Nothing moves. Nobody stops it, and nobody fixes it.
Resistance that looks like confidence
This kind of leadership resistance is hard to see. It doesn’t look like crossed arms and refusal.
It looks like confidence. Decisions that keep getting delayed. Pilots that never scale. Rollouts approved in the boardroom that quietly stall in the teams.
Underneath, many leaders feel AI as a threat to their identity. For years, their judgment and experience were their main value. Now a machine can do part of that. So instead of honestly asking, “Is my judgment still the advantage it used to be?”, they defend it.
What leaders can do
1. Use AI yourself, visibly. If leaders don’t use it, nobody will believe it matters.
2. Make it safe to say “this isn’t working”. Reward people who bring problems early. Thank them publicly. The fastest way to save a failing project is to hear the truth sooner.
3. Review your own role honestly. Ask: which parts of my work could AI do? Where does my judgment still add the most value? How should my role change?
4. Separate your ego from the project. Backing an idea and then admitting it needs to change is strength, not weakness.
5. Set a clear “stop or change” point. Before any AI project starts, agree what results you expect by when, and what you will do if you don’t get them.
Example: A company’s leadership team launches an AI tool for its sales team. Adoption is low. The managing director’s first reaction is to blame the salespeople. Instead, she sits with three of them for a day. She learns the tool doesn’t connect to their CRM, so they have to enter everything twice. The problem was never resistance. It was a leadership decision made without asking the people doing the work. She fixes the integration, and usage rises within weeks.
The bottom line
AI adoption doesn’t usually stall because of employees or culture.
It stalls because leaders ask everyone else to change, while protecting themselves from the same honest examination.
If you want your team to embrace AI, go first. In using it, in questioning it, and in changing yourself.
What would your team tell you about your AI projects if they knew it was completely safe to be honest?