What to tell your team when they ask if AI will replace them

There's a question your employees are carrying around right now, and most of them are afraid to say it out loud.

Will AI replace me?

They're not asking it in the all-hands. They're asking it on the commute. In the group chat. In the quiet minute before they open their laptop. And when they go looking for an answer, they look up — at you, at leadership. You're the one who's supposed to know how it really is.

That's the role. The team has always looked to the front of the room for a read on the future, and for most of business history that worked, because the leader genuinely did know more.

There was also time. Change used to move slowly enough that even a big, slow-moving company could keep up — time to design new roles, retrain people, move someone into a different job instead of out the door. A leader could honestly say "this will change things, and here's how we'll get you through it," because there was, in fact, time to get through it.

That time is mostly gone. AI moves faster than any company's planning cycle was built for. And what most companies see when they look at AI isn't a careful rethink of how work gets done — it's a way to cut costs and move faster. That's the race they're running, and everyone in the building can feel it. Which is why "will AI replace me" cuts deeper now than it would have two years ago. Your team isn't only asking whether their work will change. They're asking whether anyone will take the time to bring them along — or whether they're just a cost to trim.

So here's the situation you're actually in. Your team needs an answer to the biggest question in their working lives. They're looking straight at you. And you don't have it.

Not because you're unprepared. Because nobody has it. AI is moving faster than anyone can plan for, the vendors are selling a confidence they can't back up, and the honest answer to what AI will do to any one team is we don't know yet. For the first time in most leaders' careers, the person at the front of the room and the people in the seats are looking at the same fog, at the same time.

There's no shortage of input, either. Every few weeks one of the Big Four puts out another report on workforce disruption — which roles get automated first, which functions shrink, how fast. You read those. So do your employees. The forecasts are sitting in everyone's feed at the same time.

So one thing is already clear: you cannot stand up and say business as usual.

That's the real problem this piece is about. Not how to deliver good news, and not how to deliver bad news. How do you communicate to your team about AI when you genuinely don't hold the answer yourself?

Saying "business as usual" when it isn't is gaslighting

I've watched this pattern play out in enough organizations to know how it goes. Something big is shifting — a merger, a restructure, now AI — and the message that comes down to the team is some version of "nothing to see here, stay focused, business as usual". Meanwhile budgets are being cut and hiring is frozen and the news is full of it.

Here's what that does. The leader is saying something they don't believe, and they need the team to believe it anyway. When someone tells you the thing in front of your eyes isn't happening — when they ask you to doubt yourself rather than doubt them — that isn't reassurance. It's gaslighting.

And it leaves your team with only two ways to read you: you can't see what they see, or you can and you're handling them. One makes you look out of touch. The other makes you look dishonest.

Neither is a leader anyone wants to follow into an uncertain year.

Now, your instinct is right, you want to shield the team from worry, and you want to look like you have a grip on things. But a hollow reassurance does the opposite of protect. It tells your most capable people that the official channel can't be trusted on the one topic they care about most. And once they believe that, they stop coming to you for the real picture. They go to the group chat instead.

"I don't know yet" is the strongest thing you can say, if you say it well

Here's where most leaders flinch. They agree, in principle, that honesty matters. Then they say: but if I admit I don't know whether AI changes this team, won't I look weak? Won't people panic?

No. And the fear is worth taking apart, because it's misdiagnosed.

People don't panic at the words "I don't know." They panic at a gap they can't measure. When the official story doesn't match what they can feel, and nobody will say how wide that gap actually is, their imagination fills it in — and imagination always reaches for the worst case. An unknown that's been named and sized is something a person can carry around. The hidden one is what keeps them up at night.

A named uncertainty is smaller than a hidden one. Every time.

So "I don't know" is the right answer. But on its own it's not enough — and this is where leaders get it wrong in the other direction. Some hear "be honest about uncertainty" and turn into a fog machine. Everything is changing, nobody knows anything, we'll adapt as we go. That is abdication, and it leaves the team with nothing solid to stand on.

The honest answer has a structure. "I don't know" only works when it's surrounded by things you state plainly that you do know — and there is always more of that than the panic suggests. A clear AI message sorts into three buckets, said out loud:

• What's decided. Your principles — whether the intent is to augment people or replace them, said plainly. What's protected this quarter. What is not on the table.
• What's being worked on. The pilots running, the workflows under review, the questions a team is actively chasing.
• What's genuinely still open. The things you have not figured out — named specifically, not gestured at.

The decided bucket gives people something to stand on. The open bucket, once spoken aloud, stops being a monster and becomes a list.

Vague honesty reassures no one. Specific honesty is what changes mindsets.

You don't owe your team all the answers. You owe them a path of getting to it

Here's the move that turns this from a hard conversation into a good one.

When you can't hand your team the answer, the instinct is to feel you have nothing to offer. That's not true. You have something better than an answer, and it happens to be the only honest thing you can offer right now: a credible way of finding out.

Think about the gap between these two statements. "We don't know what AI will do to this team — we'll keep you posted." And: "We don't know yet. So here's the intent we're holding, here are the three things we're testing this quarter, here's what each test would actually tell us, and here's when we'll know more." The first is a shrug. The second is a plan. Neither one has the answer — but only one of them is leadership.

This is what a leader can stand behind honestly: not a forecast, but a process. Say plainly what you're aiming for — whether the goal is to grow people's skills or to cut jobs — because your team will assume the worst until you tell them otherwise. Then treat the open questions as experiments with real dates. We're testing whether AI can take the repetitive half of this role. We'll know by March. If it can, here's what we'd want the rest of the role to become. You're not promising an outcome. You're promising a method, and a date.

And here is the part that matters most. Do not run those experiments over your team's heads. Run them with your team.

The people whose jobs are in question are not bystanders here. They're the ones who know which parts of their work are tedious and which parts are the actual craft. They know where AI would genuinely help, and where it would quietly break something nobody noticed mattered. A leader who designs the AI experiments alone — in a room with vendors and strategy decks — is throwing away the best source of insight in the building. Worse, it tells employees the future is being decided about them rather than with them. That signal does its own kind of damage.

The leaders who get this right do something that looks almost backwards. They put the people most afraid of AI closest to the AI decision. They ask the team to help pick which tasks to test. They let the people doing the work decide what "better" should mean. That doesn't just produce smarter experiments. It changes how the whole thing feels — because someone helping to shape the change is no longer just waiting for it to land on them.

You can't give your team certainty, but you can give them a say.

There are structured ways to run this. An AI Workflow Sprint is one — it brings the team into redesigning how their own work gets done with AI, instead of having that redesign handed to them.

Turning this into questions you can use

So here are five questions to sit with before your next AI conversation with your team:

1. What am I about to say that I don't quite believe myself? That's usually the line worth cutting.
2. What does my team already know that I'm talking around? They read the same news you do.
3. Can I actually fill the three buckets — decided, in progress, still open? If they come out thin, that's a thinking gap, not a wording one.
4. Have I said out loud whether the goal is to augment people or replace them? If not, they've already guessed.
5. Who's in the room when the experiments get designed? Worth checking the people who do the work are on the list.

None of these need a perfect answer. They're just worth a quiet moment before you next conversation.

The leaders who come through this period with their teams' trust intact won't be the ones who called AI correctly. Nobody will. They'll be the ones who were straight about what they didn't know, clear about what they were optimizing for, and willing to build the answer together with the people it affects most.

Helping teams do exactly that — get to clarity early, and as a group — is most of what we do at Design Sprint Academy.