For years, the big worry about AI at work was jobs. Would the tool take ours, or someone else's? That question has not gone away, but a quieter one is now sitting beside it. In a 2026 Resume Now report, 63% of workers said they expect AI to make the workplace feel less human this year, and 57% named the loss of human skills as the biggest workforce issue, a little ahead of job displacement at 49%.
Read that again slowly. Most of us are not afraid of being replaced. We are afraid of working somewhere that no longer feels like people. And because we expect it, we may be quietly helping it along, handing over the warm parts of work first because they are short, repetitive and easy to hand over.
That is a human issue because a prediction is not a fact. Whether work feels human is decided by a handful of small moments, and those moments are still ours to keep. The warmth in a team is not lost to AI. It is given away, or it is kept, one ordinary choice at a time.
------------- Context -------------
When a new tool arrives, we naturally reach for the tasks that feel like a chore. They are the short ones, the repeated ones, the ones where nothing seems to be at stake. A welcome message for a new starter. A thank-you after a project. A note saying well done. They take five minutes each, they come around often, and they are easy to ask a tool to write.
The trap is that the same features that make these tasks easy to hand over also make them valuable. They are short because warmth does not need many words. They are repeated because relationships are built through repetition. And they feel low in stakes only because the cost of getting them wrong is slow and invisible, a little less goodwill each time.
This is where the survey result is useful. A workforce that expects work to become less human is a workforce that may stop defending the human parts. If we believe the outcome is decided, we stop treating each message, each hello and each thank you as a choice. The prediction becomes the permission.
The better framing is that the human quality of a workplace is a set of habits, not a feature of the technology. AI does not decide whether we welcome people properly. We do. The technology only makes it easier to skip, and easier is exactly when it helps to decide in advance.
That matters because trust, loyalty and the sense of belonging all grow from moments that were plainly meant by a person. Once they become routine output, the people receiving them can tell, and the warmth that made them worth sending is gone.
------------- A Forecast Is Not a Finding -------------
Imagine a design studio of six people. The owner is busy, so she asks the tool to write the welcome note for a new junior designer. It is polite, correct and slightly generic. The new starter reads it, says thank you, and says nothing more about it.
Nothing looks wrong, and nothing has been lost that anyone can point to. But the junior designer has just formed a first impression of the place, and the first thing it told her was that nobody there had five minutes to write to her. A few weeks later she is slower to ask questions and quicker to message the tool instead.
The owner never decided that her studio should feel less human. She simply followed the forecast, which said that this is where things are heading, and so this must be what is normal.
What we expect shapes what we allow. A forecast tells us what is likely if nobody chooses otherwise, and that is all it tells us. It is not a finding about what has to happen.
This is a human outcome worth naming. When we treat the prediction as a prompt to decide instead of a verdict to accept, the welcome stays a welcome, and the new starter knows who actually wrote it.
------------- The Warm Tasks Are the First Ones We Hand Away -------------
There is an uncomfortable pattern in how teams adopt AI. The tasks that go first are not the heavy ones, which need care. They are the light ones, which feel safe. And the light ones are very often the relationship ones.
Take an account manager who finishes a long project with a client. The thank-you message used to take her ten minutes. She wrote it, added the one detail about the client's launch day that only she would remember, and sent it. Now she asks the tool to produce it in 20 seconds. The message is fine. It is also the same one that every other supplier sent that week.
The time saved is nine minutes and forty seconds. What it cost is harder to count: the small signal that someone paid attention, which was the whole purpose of the message.
We tend to measure the saving because it is visible and the loss is not. But in relationship work, the effort is the message. A thank you that took no thought says so, and a note that took ten minutes says that someone chose to spend them.
That does not mean the tool has no place here. It can remind her of the dates, help her recall what the client cared about, or tidy a rough draft. It simply should not be the one speaking. The human outcome is that the client knows a person was thinking of them.
------------- Three Moments Are Enough to Keep a Workplace Human -------------
The practical question is how to protect the human parts without banning anything or returning to doing everything by hand. The answer is to choose very few moments and to protect them on purpose.
We suggest three. The welcome, because it sets the tone for everything a new person believes about the place. The feedback, because honest, kind and specific feedback is where people feel seen. And the thanks, because recognition is the cheapest and most powerful form of belonging a team has.
A team lead preparing a performance conversation shows how this can work. She uses AI to organise her notes from the past quarter, so nothing is forgotten and the examples are accurate. That saves her an hour of sorting. Then she closes the tool, walks into the room and says the words herself, including the difficult sentence, in her own voice.
The tool did the tidying. The person did the caring. Nobody receiving that feedback would feel it had been processed.
This is the kind of line worth drawing. AI may prepare, remind, organise and polish. A person speaks the welcome, delivers the feedback and signs the thanks. The result is that people keep what they came for, which is to be seen by another person, and the efficiency we gain elsewhere does not cost us that.
------------- The Human Standard Has to Be Said Out Loud -------------
The last step is the simplest and the most often skipped. A standard that lives only in one person's head will not survive a busy week. If we want the human moments to hold, someone has to say so, and the team has to agree.
It can be one sentence, written at the top of a shared note or said at a Monday meeting. In this team, welcomes, feedback and thank-yous are written and spoken by a person. Anything else, we are free to use the tools for. That is all.
A small group that does this finds something unexpected. People relax. Once the line is drawn, nobody has to wonder whether it is rude to use a tool for the weekly report or whether they are expected to hand-write every message. The things that matter are protected, and everything else is simply permitted.
It also gives leaders something to model. When the owner of that design studio writes the next welcome note herself, it takes seven minutes and it is a little clumsy, and the new starter keeps it. That is the human result, and it is a trust that no efficiency saving buys back.
A spoken standard turns a worry about a predicted future into a small, visible practice in the present. The team is no longer waiting to see whether work gets colder. It is deciding that it will not.
------------- Practical Moves -------------
First, write down the three moments that matter most in your work, such as welcoming someone, giving feedback and saying thank you. Choosing them in advance means you will not decide under pressure.
Second, for each of the three, put your own name on the sending, speaking or signing. Let AI prepare notes or jog your memory, but make sure the words that reach the person are yours.
Third, add one detail only you could know to every thank-you, such as a date, a small struggle or a moment that went well. It takes two minutes and is the part people remember.
Fourth, look back at the last week and find one warm task you handed to a tool without thinking. Take it back next week and notice what changes in the reply you get.
Fifth, share the standard with your team in a single sentence, and ask each person to add the moment they most want to stay personal. A shared line is far easier to keep than a private one.
------------- Reflection -------------
Most of us expect work to feel less human, and the expectation is understandable. The tools are fast, the days are full, and the easiest thing is always to hand over the small warm tasks first. But a prediction describes a direction, not a destination. What happens depends on what we protect.
That is why the human question matters as much as the efficiency question. The welcome, the feedback and the thanks are small, and they carry a great deal of trust. When a person plainly stands behind them, they build the kind of loyalty and belonging that no tool can produce on our behalf.
In the end, the AI shift worth paying attention to is not whether workplaces become colder. It is whether we notice the moments that make them warm, and keep them in our own hands. Three moments, protected on purpose, are enough to prove the forecast wrong.
Which three moments at work do you most want to stay person to person, and have you ever said so out loud?
When did a message from a real person last change how you felt about a team or a client?
And what is one warm task you have already handed to a tool that you would quietly like back?