Turning productivity gains into value that is visible to employees requires concrete decisions from leadership.
Your employee may see that AI is useful. Believing that it is good for their own future is another matter. Preparing a report faster, pulling meeting notes together or improving a customer reply can increase the value of everyday work. Yet the same person may wonder how their own role will change in the coming year. It is not surprising that these two thoughts exist side by side. Experiencing the benefit of a tool is not the same as trusting the outcomes of the transformation.
Benefit and concern
Microsoft’s UK research, published on 21 October 2026, offers a starting point for thinking about this distinction. Conducted by Censuswide, the research covers 5,000 employees and 1,000 senior business decision-makers. Of the employees who use AI in their work, 35 percent are worried that the technology is reducing their own job security. The research also reports positive effects of AI use on work and careers. We cannot assume that the findings on benefit and those on concern come from the same people. From a management perspective, my conclusion is this: we need to understand the tool’s usefulness and the employee’s confidence in the future separately. [1]
It is easy to jump quickly from here to the conclusion that “employees are resisting change.” Yet we first need to look at which question has been left unanswered. Is it that the employee does not understand the new tool? That they do not know who will bear responsibility when they make a mistake? That they cannot find time to learn? Or that they cannot see how they will benefit when their own contribution grows? Different reasons can lie behind the same behavior. Fitting reluctance to use the tool into a single label also narrows the options available to the manager.
How the capacity gained will be used
In AI transformation there is a topic that management often does not discuss openly: how the capacity gained will be used. When a team completes the same work in less time, the time that is freed up does not automatically turn into a meaningful benefit. That time can go to better service, a postponed improvement, a new skill or additional workload. Each of these is a different management choice. When no choice is made, employees’ assumptions fill the gap. Asking the communications team for a more persuasive message cannot take the place of a decision that has not yet been made.
That is why, when launching a pilot, I suggest asking not only “How much faster will we get?” but also “What work will we direct the time we gain toward?” What quality difference will we create for the customer? Which of the team’s recurring burdens will we reduce? Which skill will we invest in developing? The answers to these questions should not get lost in a large strategy document. The owner of the pilot should be able to set out, in a few clear sentences, the expected change and how it will be tracked.
The opportunity to learn
The second topic is the opportunity to learn. Giving an employee a license is not the same as creating the conditions in which they can learn to use it. If you leave learning to the person’s evenings while their workload stays the same, you end up assuming that everyone with access to the software has an equal opportunity. Yet the manager needs to make room in the calendar, define a safe space for experimentation and explain how they will approach early mistakes. Whether learning is an organizational priority shows in these small decisions more than in any presentation.
The future of the role
The third topic is the future of the role. No manager can guarantee that everyone’s job will stay the same in a period of uncertainty. There is no need to give such an assurance. But it is possible to separate what is known today from what has not yet been decided. Which tasks are expected to change? Which decisions will remain with people? When and how will those affected be informed? Ensuring that an employee does not learn about a decision affecting their future through rumor is one of the most basic forms of transparency management can provide.
Accountability
The fourth topic is accountability. If an output prepared by AI is wrong, who will check it? What approval will be needed before it goes to the customer? When an employee rejects a recommendation, will they be able to justify doing so? While these boundaries remain unclear, saying “experiment more boldly” can invisibly shift the risk onto the team. Safe experimentation is not built with a technical environment alone. Authority, support and how the manager behaves in the face of mistakes are also part of that environment.
At the next pilot meeting
To apply this approach, take up a single workflow at your next pilot meeting. Identify the step in that work that takes the most time today. Write down which step AI will change. Then look for the expected benefit in three separate places: the customer experience, the organization’s capacity and the employee’s work. If no measurable change can be found under a given heading, treat that as a gap; do not paper over it with statements of intent.
For example, the time it takes to prepare customer replies may shrink. But if the accuracy of the replies falls, the gain in speed does not count as success on its own. Likewise, if the employee gets through more cases but constantly has to do extra checks, it cannot be said that the workload has truly decreased either. That is why time, quality and workload should be tracked together. When the first results do not meet expectations, changing or stopping the implementation should also be one of the pilot’s legitimate outcomes.
You do not have to resolve this entire conversation in a large meeting. You can start with the owner of the workflow and a few of the people who do the work. First, record the current situation. Then define the boundaries of a small trial. Fix the evaluation date from the start. Do not collect employees’ feedback only with the question “Are you satisfied?” Ask where they carry out extra checks, what they are able to do better and what uncertainty remains. That way you can see the part that usage data cannot explain.
The real test for leadership
This is where the real test for leadership begins. Being able to keep the employee’s question on the table while explaining the benefit the company will gain: How will my job change, and how will I be supported through this change? The answer to this question may not always be reassuring. But it can be clear, consistent and trackable. Trust, too, is not built only by painting a positive picture of the future; it grows as the decisions made and the promises kept confirm one another over time.
At your next AI meeting, discuss the usage rate and the time saved. Then add this question: Through which concrete decision have we made the value of this transformation visible to employees? If your answer is not yet clear, that is exactly where the leadership agenda to work on lies.
Source and method note
- Microsoft UK, 21 October 2026. Censuswide research; fieldwork period 2-12 October 2026. Read the research announcement.
This is a UK sample; the results cannot be generalized to all countries. The research was published by Microsoft. Because the sample coverage of the benefit and workplace-concern measures may differ, the percentages have not been compared directly. The management recommendations are the author’s interpretation; the survey does not establish cause and effect.
Image: original illustration generated with AI.