You close the laptop. Everything on the list got done, and some of it got done faster than it would have a year ago. By any measure a manager would use, that was a good day.
And nobody really spoke to you.
There were messages. There were two calls that were honestly just status updates. But there was no moment where you turned to somebody and said a true thing about your week. You sit there a second longer than you meant to, wondering why a good day feels like that.
Feeling lonely at work while the work itself goes better than ever now has a number attached to it. And the number is stranger than almost anyone reporting on it has made it sound.
Lonely at Work and More Satisfied With Life, in the Same Survey
A study published this month in Preventive Medicine Reports looked at how working with AI lines up with distress, loneliness and life satisfaction. It used the US Census Bureau’s Household Trends and Outlook Pulse Survey from March 2026, a nationally representative sample that generalises to roughly 152 million employed US adults.
Of those workers, 51.7% said they use AI on the job. Just over half. Whatever this is, it is not a niche anymore.
Then came the two findings that do not seem to belong in the same paragraph:
- Workers who used AI more reported higher life satisfaction.
- Workers who used AI more also reported being lonely more often.
Not two different groups. The same direction of use, pointing at both results at once. More satisfied with life, and lonelier, in the same week.
How Big Is “Lonelier”? Smaller Than It Sounds
This is where most coverage stops, and it is exactly where it gets interesting.
Both effects are small. Life satisfaction rose by about a tenth of a standard deviation, which is a real signal but a gentle one. The loneliness result put heavier AI users roughly 13% more likely to report feeling lonely, and the range the researchers published around that estimate stretches from about 2% to about 25%. The bottom end of that range is very nearly nothing at all.
And there is a third finding almost nobody repeats, which changes the whole shape of the story: AI use showed no meaningful link to moderate or severe psychological distress. Not more anxiety. Not more of the heavy stuff.
So this is not “AI is making workers miserable.” The data says something much narrower and much odder. People using AI at work are doing fine. They are reasonably happy with their lives. They are not in distress. They are just a bit more alone in it.
One more honest caveat, and the researchers themselves led with it: this is a snapshot of a single month. It cannot tell you which came first. Lonelier people may be reaching for AI, rather than AI making people lonelier, and nothing in a one-moment survey can separate those. The authors called for studies that follow people over time — the same problem we ran into reading the research on whether talking to AI makes you less lonely, where one study measured an evening and the other measured a year.
The Worry Workers Name Themselves
Ask workers directly and they will tell you what they are afraid of, and it is not the thing the headlines assume.
The Workday Foundation surveyed 2,150 full-time employees across seven countries, all of them at large organisations and all actively using AI. Their top concern was not losing the job. It was 43% naming reduced human-to-human interaction — the colleagues, not the paycheque.
The same research found a third of employees said they rarely or never talk to co-workers beyond what the task requires.
But the detail that stayed with me is the reason people gave for taking a question to AI instead of to the person sitting near them. It was not that AI is faster. It was that AI felt less judgmental than asking a colleague.
Hold onto that one. It explains more than any of the percentages do.
Then a Much Bigger Survey Said the Opposite
If the story ended there it would be tidy, and it does not end there.
Gensler’s 2026 global workplace survey covered more than 16,400 office workers across 16 countries — several times the size of the studies above. About 30% of them qualified as heavy AI users. And those heavy users reported stronger relationships with their teams and more meaningful friendships at work than the people barely touching the tools.
The time data is the part that really bites. Heavy AI users spent less of their working week alone than late adopters — about 37% against 42% — and more of it learning and socialising.
Less time alone. Not more.
Two careful surveys, run in the same year, on overlapping populations, pointing in opposite directions. The easy move is to pick the one you already agreed with. The more useful move is to ask what could possibly make both true.
The Tool Is Not the Variable
Here is the reading that fits all of it.
If AI reliably caused loneliness, the biggest users would be the loneliest, everywhere, in every survey. They are not. In Gensler’s data the heaviest users are the most connected. Same tool, opposite outcome — which means the tool is not what is deciding.
What AI actually hands back is two things. One is time. The other is the removal of a reason to ask somebody.
The time is neutral. Gensler’s heavy users appear to have spent theirs on people: learning, talking, being in the room. Other people spent theirs on more tasks, and the day quietly sealed over.
The removed reason is not neutral at all. A workplace runs on small acts of needing each other — how do I phrase this, where do I find that, does this read badly to you. Each one is trivial and slightly inefficient, and each one is also a door. Take away a hundred of those small requests and nobody has been excluded from anything. Nobody is being frozen out. There is simply no longer any occasion to turn around.
That is why the loneliness in this research sits next to satisfaction rather than distress. Nothing bad happened. Something just stopped happening. It is a close cousin of what we found looking at whether it is really lonely at the top — the loneliest person in the building is rarely the one everybody assumes, and the gap is never where the saying puts it.
What Makes Asking a Person Expensive
Go back to the reason workers actually gave. Not speed. Not accuracy. It does not judge.
That is a quietly enormous admission, because it means the obstacle was never the colleague’s calendar. It is the risk of being thought less of. Asking a human being a question tells them something about you — that you did not know, that you are struggling, that you are behind. A chat window costs nothing because nothing is at stake in it. It cannot think less of you.
It also cannot think more of you. That is the half of the bargain nobody notices until much later.
And notice what people are really reaching for, because it is not “a listener with no opinion.” Nobody in history has felt truly accompanied by something that does not care. The thing we actually want is rarer and much harder: to be seen all the way down, by someone whose opinion genuinely matters, and not to be written off for what they find.
That idea is far older than any of this technology. It sits near the centre of what people have believed about God for thousands of years — that you can be completely known, including the parts you carefully keep out of the office, and still not be condemned. Not because nothing is at stake. Because love got there first.
Which has a very practical edge to it, even on a Tuesday afternoon. What the frictionless window is offering you is not safety. It is the absence of risk, and those turn out to be different things.
Four Things That Cost Nothing
None of this requires quitting a tool, a budget, or anyone’s permission.
- Notice what you stopped asking. Think of one question you would have taken to a person a year ago and now type into a box. Once this week, take it to the person instead. The answer may be worse. The day will be better.
- Spend the time it gave back on a person, not on more tasks. This is the single thing that seems to separate the connected heavy users from the lonely ones. The hour has to land somewhere. Decide where.
- Say the thing a machine cannot be told. Not “help me word this message.” Something closer to “this week has been heavy.” You only need one person and one sentence.
- Let somebody see you not know something. The judgment you are avoiding is almost always smaller than the loneliness you are buying to avoid it.
And if you are genuinely unsure how alone you have been feeling lately, our free How Lonely Am I? check takes a couple of minutes, and nobody sees the result but you.
Back to the Closed Laptop
The work got done. That part is real, and the research says you are probably not imagining the satisfaction either.
It is only that a good day used to come with a few dozen unnecessary moments attached to it, and the unnecessary moments were quietly doing something. They are the easiest thing in the world to automate away, because every single one of them looks optional right up until they are all gone.
Tomorrow there will be one question you could ask a person instead. That is the whole of it.
A Question Worth Sitting With
Do you think AI is actually making workplaces lonelier — or is it just removing the small, slightly inefficient reasons we used to have for talking to each other, and we were never going to protect those anyway? Tell us which one you think it is in the comments below. We read them.
Share This
- Half of US workers now use AI on the job. A new study found heavier users are slightly more satisfied with life AND slightly lonelier — with no rise in actual distress. Nothing bad happened. Something just stopped happening. https://bgodinspired.com/index.php/personal-growth-and-life-skills/lonely-at-work-ai-trade-off/
- The reason workers gave for asking AI instead of a colleague wasn’t speed. It was that AI doesn’t judge them. That one line explains the whole thing. https://bgodinspired.com/index.php/personal-growth-and-life-skills/lonely-at-work-ai-trade-off/
- One survey says heavy AI users are lonelier. A bigger one says they’re the most connected people in the building and spend less of the week working alone. Both are right — which means the tool was never the variable. https://bgodinspired.com/index.php/personal-growth-and-life-skills/lonely-at-work-ai-trade-off/
Q&A
Does using AI at work make you lonely?
Current evidence is mixed and does not establish cause. A 2026 study in Preventive Medicine Reports, using a nationally representative US survey from March 2026, found that workers who used AI more often reported feeling lonely more often — about 13% more likely, with a published range running from roughly 2% to 25%. Because the survey captured a single moment in time, it cannot show whether AI use led to loneliness or lonelier people were more drawn to AI. A separate Gensler survey of more than 16,400 office workers in 16 countries found the opposite pattern, with heavy AI users reporting stronger workplace relationships.
What did the 2026 study on AI use at work actually find?
It found three things. Workers reporting greater AI use at work had slightly higher life satisfaction, about a tenth of a standard deviation. They also reported a modestly higher rate of loneliness. And there was no significant association with moderate-to-severe psychological distress. The sample generalised to roughly 152 million employed US adults, 51.7% of whom reported using AI at work. The authors called for longer-term studies, since a cross-sectional survey cannot establish cause and effect.
Why do workers say they prefer asking AI instead of a colleague?
Because it does not judge them. In research by the Workday Foundation covering 2,150 full-time employees across seven countries, workers described AI as less judgmental than asking a colleague. The same research found 43% named reduced human-to-human interaction as their top AI-related concern, ahead of losing their job, and that about a third of employees rarely or never talk to co-workers beyond what the task requires.
Do all studies agree that AI makes workers lonelier?
No, and the disagreement is substantial. Gensler’s 2026 global workplace survey of more than 16,400 office workers across 16 countries found that the heaviest AI users reported stronger team relationships and more meaningful friendships at work. Those workers also spent less of their working week alone than late adopters, roughly 37% against 42%, and more time learning and socialising. One way to read the conflict is that AI frees up time and removes small reasons to ask colleagues for help, and the outcome depends on what replaces them.
How can I use AI at work without feeling more isolated?
Pay attention to what the tool displaced rather than to how much you use it. Take at least one question a week to a person instead of a chat window. Deliberately spend some of the time the tool saves on people — the survey evidence suggests heavy AI users who stay connected spend less of their week working alone, not more. Tell a colleague one true thing that is not a task update. And allow yourself to visibly not know something in front of someone, since avoiding that small risk is the specific trade most people are making without noticing.