The AI disconnect: why leaders and employees are living different stories

Two meetings about the same rollout

Picture two meetings in the same organisation on the same afternoon. In the boardroom, the AI programme is going well. Usage is up, the pilots look promising and the communications plan has been delivered on time. In a team meeting two floors down, people are quietly wondering whether their jobs will exist in two years, and nobody has asked them.

Both meetings are describing the same programme. Only one of them is reported upwards.

New research from Brunswick, published on 5 November 2026, puts numbers on that picture. It helps explain why AI programmes can look healthy on a dashboard and feel uncertain on the floor.

What Brunswick found

Brunswick surveyed more than 2,000 US employees and 65 senior leaders at companies with 250 or more employees. Some caveats first. The sample is American, and Brunswick is a communications advisory firm, so this is research from a firm that advises leaders on how to communicate. The leader sample is small. Even so, the gaps are large.

Adoption itself is not the problem. Nearly six in 10 employees use AI at work at least a few times a week, and only 8% reject it outright. If you measure usage, you will be pleased.

The gap shows up in belief. 92% of leaders believe AI will create new opportunities. Only 53% of employees agree. That is a gap of nearly 40 points. On communication, 95% of leaders say their company talks openly about the benefits of AI. Only 57% of employees feel the same.

Brunswick's sharpest line is this: "Leaders who believe their workforce is aligned have no reason to change course, nor can they fix a problem they do not believe exists."

Why the gap stays hidden

The usual measures of an AI programme are counts: licences issued, prompts run, people trained, messages sent. Each of these is an output. None of them tells you whether people understand where the organisation is going or whether they can see a place for themselves in it.

Service management has a useful habit here. It insists that value is judged by the person receiving a service, not the person providing it. Esm applies that principle to AI communication. A message is not a service until it has been received and understood. Measured from the sending end, the communications plan is complete. Measured from the receiving end, almost half the workforce did not get it.

There is also a more personal reason the gap stays hidden. Ian Hughes put it plainly in What Stays the Same: "The most dangerous person to lead an adoption programme is the person who found it easy. They cannot see the obstacle, because they never hit it."

Senior leaders have usually had more time, more support and more say in how AI arrives. It is easy, from there, to assume everyone else has had the same journey. As the same book says, "The first mistake almost every leader makes is to assume the team's journey will look like their own."

The motivated middle

The most interesting finding is about who is leading adoption. It is not the youngest staff. Employees with five to 14 years of experience reported the highest levels of use, excitement and confidence. 67% of them use AI frequently, against 52% of those with fewer than five years and 48% of those with 15 years or more. 78% of mid-career staff say they understand how to use AI at work, against 65% of early-career staff.

Mid-career people know the work well enough to see where judgement matters and where a shortcut is safe, and they are still open to changing how the work is done. They are the natural translators between a strategy and a Tuesday morning.

That matters for another reason. The Conference Board, reported the same week, found that 38% of leaders expect fewer middle managers and 36% expect fewer frontline managers and project leaders, while only 4% expect a smaller C-suite. If the layer most likely to carry adoption is also the layer most likely to be thinned, organisations may be removing the people who close the gap.

What employees want to hear

Brunswick tested five messages about AI. The one that worked best framed AI as a partnership: people bring empathy, ethics and creativity, and AI brings speed and scale. 75% said it would make them feel their company was on the right track. Messages that led with disruption and job losses landed below 50%.

The point is not that leaders should pick the friendliest slogan. The difference, Brunswick says, was whether employees could see a role for themselves in the story. People also asked for plain explanations, clear policies and an honest account of what is still uncertain. PwC's Pete Brown made a similar point in HR Magazine this week: "Workers aren't expecting leaders to sugar-coat everything."

Two questions worth asking this month

Esm would suggest two honest questions before the next AI update goes out.

1. If you asked each team what the AI strategy means for their own work, what would they say, and are your measures about what was sent or what was understood?
2. Who are the mid-career people already using AI well, and will any restructuring plan remove the very people who translate strategy into daily work?

Neither needs new technology. Both need leaders to listen from the receiving end.

The honest summary

Adoption numbers can look excellent while trust quietly drains away. The gap is not a failure of the tools. It is a failure to notice that the people doing the work are living a different story from the people describing it. The fix starts with leaders accepting that their own experience of AI is not the typical one, and then going to find out what the typical one is.

To see how other leaders are working through this, the conversation continues on ESM's Exchange.

Tags: ai adoption, leadership, employee experience, communication, trust, change management

Sources

• https://review.brunswickgroup.com/article/ai-workplace-disconnect/

• https://itbrief.news/story/ai-adoption-outpaces-leadership-change-report-says

• https://www.hrmagazine.co.uk/content/news/exclusive-interview-how-to-create-an-ai-capable-workforce