The AI confidence gap: why UK leaders and their teams see different futures

Most leadership teams believe their AI programme is going reasonably well. New research suggests many of their people see it differently, and that the difference matters more than the technology.

What the research found

Microsoft commissioned Censuswide to survey 1,000 senior business decision-makers and 5,000 employees across the UK. The findings, [reported by Technology Record on 21 October 2026](https://www.technologyrecord.com/article/uk-could-be-a-world-leading-ai-economy-within-five-years), show a striking split:

- 84% of leaders believe the UK can become a world-leading AI economy within five years. Only 41% of employees share that confidence. - Nine in ten leaders say their organisation has a clear AI strategy. - 71% of employees say AI has had a positive impact on their work, and 69% say it has improved the quality of what they produce. - 57% of employees worry AI will reduce job opportunities, and 35% of those already using it are concerned it has lowered their own job security. - Both groups put AI skills and training among their top three priorities.

It is worth noting that this is vendor-commissioned research, released alongside Microsoft's own report on AI diffusion. The headline numbers are still useful, because the pattern they describe will be familiar to anyone who has led a change programme.

Why the gap is not about enthusiasm

The easy reading is that employees are sceptical and need persuading. The data does not support that. Most employees are already using AI and say it helps. The gap is not in enthusiasm. It is in experience, and in trust about what comes next.

Leaders see AI through strategy papers, investment cases and steering groups. Employees see it through their working day. Those are two honest views of the same change, and both are incomplete on their own.

Four things the gap is telling leaders

1. Confidence is not the same as experience

A leader's confidence is usually built on the plan. An employee's view is built on what actually happens: the extra checking of AI output, rules that are not yet clear, and expectations that quietly rise once a task looks faster. None of that shows up on a dashboard, so it is easy for a leadership team to mistake a good plan for a good experience.

What to do: spend time with the work itself. Ask people where AI has made their job harder, not easier, and treat the answers as data rather than resistance.

2. The worry is about futures, not tools

The 57% figure is the one to sit with. People are not mainly worried about whether the tools work. They are worried about what the tools mean for their role, their progression and their job security. No training course answers that on its own.

What to do: say plainly what saved time will be used for. If the intention is better work rather than fewer people, write it down and act on it. If the answer is not yet known, say that too. Uncertainty handled honestly builds more trust than reassurance that turns out to be hollow.

3. The best evidence is a few levels down

Service management has understood this for a long time. As John Sansbury writes in The Journey to Service Excellence, "the best people to understand how to improve the working environment are those on the frontline or at the coalface, who are often dealing with consumers on a daily basis."

The same is true of AI. The people using it every day know where it saves time, where it creates rework and where it quietly changes the job. Ian Hughes makes a related point 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."

What to do: build regular listening into the programme, not a one-off survey. Go beyond the polite first answer to the real concern underneath it.

4. There is already common ground

The most hopeful finding is that leaders and employees agree on what comes next. Both put AI skills and training among their top three priorities. Most change programmes would be glad of a starting point like that.

What to do: match support to where each person actually is. Some people need confidence to start; others are already capable and need permission and boundaries to go further. A single training plan for everyone tends to suit no one.

The question leaders should be asking

The useful question is not "how do we get our people more excited about AI?" It is "what would our people need to see to believe this is good for them, and how would we know if we weren't hearing it?"

That is a leadership question, not a technology one, and it is where the gap closes.

How Esm can help

The [AI Enablement Toolkit for Leaders](https://exchange.esm.institute) gives leaders practical tools for exactly this gap:

- The change listening guide helps leaders hear the concern behind the first answer. - The AI change charter sets out honestly what will and will not change for people. - The naive-to-native adoption ladder matches support to where each person is. - Protect the apprenticeship keeps routes for people to learn and progress as tasks change.

Alongside the toolkit, Esm's training and enablement helps leaders and teams build AI skills together, so confidence grows on both sides of the gap.

The toolkit cannot settle the wider national debate about AI and jobs. What it can do is help one organisation be clear and fair about how it treats its own people, and that is where trust is won or lost.

Tags: AI adoption, leadership, change management, employee experience, AI skills

Sources

- https://www.technologyrecord.com/article/uk-could-be-a-world-leading-ai-economy-within-five-years