Will AI take our jobs? How to talk about your team’s fears

Why “AI will streamline our work” can sound like a redundancy announcement

An employee hears in a meeting that a new tool will halve the time it takes to prepare proposals. The manager is thinking about deadlines and the quality of the work. She starts calculating how many specialists will be needed once the company can get through twice as much work in the same time. Both interpretations are understandable. The change affects not only tasks, but also people’s sense of being useful, the stability of their income and their position in the team. The less information there is about the company’s plan, the more room there is for speculation.

It doesn’t help to reply: “Don’t exaggerate – nobody has anything to worry about.” It is better to name the question behind the anxiety: “You want to know whether changing how we work will mean changing the number of jobs. That’s an important question.” A sentence like this doesn’t prejudge the outcome, but it shows that the board and the manager understand what is at stake. Uncertainty is a serious burden even when, in the end, no cuts are made.

What research says about employees’ fears of AI

In a Pew Research Center survey published in 2025, 52% of employed Americans said they were worried about the future impact of AI on work, and 32% expected fewer job opportunities for themselves. These are opinion survey results from the US, not a forecast that one in three jobs will disappear, nor a measure of the situation in Polish companies. They do show, however, that the anxiety is not just an individual “dislike of new things”. Research on job insecurity has long linked it to stress and less favourable attitudes towards work.

You also need to watch out for the opposite extreme: the fact that certain tasks are being automated does not automatically mean an entire profession will disappear. In one department, AI will make it easier to produce a first draft, while a person checks that it meets the client’s needs. In another, the number and type of assignments will change. So rather than trading global forecasts, the manager should set out a map of the specific tasks, the data available and the decisions that have actually been made.

Why even an enthusiast for new tools may be worried

The person who uses AI most often may also be the one who best understands how fast things are changing. Today he is showing colleagues how to speed up data analysis; this evening he is wondering whether, a year from now, the company will still be paying for his skills. If a manager takes enthusiasm as a sign that there are no worries, they will miss an important part of the conversation. Likewise, silence does not mean agreement: someone may not want to reveal in front of colleagues that they are worried about their loan repayments, their skills or their position.

It is worth separating the public conversation from the individual one. At a team meeting, ask: “Which tasks raise the most doubts? Which rules on use are we missing?” In a one-to-one, you can add: “What is the most uncertain part of this change for you?” Don’t assume that people’s answers depend on age or length of service. Someone with many years of experience may want to try out the tool, while a young specialist may be afraid of losing the chance to learn the basics of the profession.

What to say when someone asks: “Are you going to make people redundant?”

The hardest moment comes when employees want an unequivocal promise that the manager cannot make. An honest answer might be: “There’s no approved decision on job cuts linked to this project. But I’m not going to promise something I have no control over. What I do know is that the trial will cover two processes, and we’ll discuss the results with you in a month.” If redundancies really are being considered, hiding that behind a slogan about “unlocking potential” will only deepen the disappointment later.

A simple table for the meeting helps: what we know, what we are still checking, what we cannot promise and when we will come back to the topic. The first column might contain the scope of the pilot, the second how quality will be assessed, the third the absence of any guarantees on staffing, and the fourth the date of the next meeting. This isn’t a PR trick. It is a public commitment that can be checked later. If the dates change, you need to say so as well, and explain why.

How to run a meeting that doesn’t end in slogans

A good meeting has three parts. First, the manager shows exactly what is going to change: the process, the tool, the scope of the data and how the results will be checked. Next, they open the floor to questions, including uncomfortable ones. Finally, they turn the conversation into tasks: who will run the tests, what may be shared with the tool, what criteria will be used to assess the results and where errors should be reported. A presentation of benefits on its own does not answer questions about job security or whether it is worth continuing to learn.

For people who don’t want to speak up in front of the whole team, make sure there is a way to ask questions individually. Don’t promise anonymity if you can’t guarantee it. After the meeting, send a short summary of what was agreed, the open questions and the date by which you will share further information. If someone asks about decisions on the future of jobs, don’t refer them to a course on writing AI prompts. Answer the question about the future of their job first. Only then talk about new skills.

How to talk about development without putting all the risk on people

The phrase “Everyone will have to retrain” can be heard as: “You’re on your own.” It is better to show which skills will really be needed and what the company will provide so that employees can acquire them. If AI prepares the first draft of a reply to a client, the employee may need better skills in verifying information, reading the context and explaining exceptions. These are different tasks from simply writing quickly.

It is worth setting up a pilot in which people not only learn the tool but also have a say in assessing whether using it improves the work. Errors, the workload created by checking and quality should all be monitored, not just time savings. The manager should also respond to the sceptics: “You raised three problems; we’ve solved two and we’re working on the third.” That is worth more than yet another appeal for a positive attitude.

You cannot talk effectively about fears of AI while ignoring the risk of changes in employment. You need to separate known decisions from hypotheses, ask about the specific consequences for tasks and people, and commit to sharing further information. Above all, trust is built when what you say matches what you later do.

How Empatyzer helps you prepare an honest conversation about AI-driven change

A manager can tell Em that the company is launching an AI pilot and the team has taken the news as a sign that redundancies are coming. Rather than asking for a “motivational speech”, it is worth setting out the facts: which decisions have been made, what is not yet known, what questions the team has raised and what needs to be agreed. Em will help structure the message, prepare an answer to the difficult question about jobs and rehearse a response to the accusation that management is hiding something. Where team context is available, the manager can also see which differences in working style, need for predictability and motivation may affect how the change is received. If the conversation concerns a specific person in the organisation, relationship tips help choose the questions and tone without the manager having to read that person’s private diagnosis. Micro-lessons then help reinforce these skills in the course of everyday work. The aim is not to convince everyone that AI will solve every problem, but to prepare a meeting at which people learn more, can say what they are afraid of and know when they will receive further answers.