Viewsยท

AI anxiety in the workplace

AI is moving faster than the people working with it can keep up with. What that is doing to people in tech right now, and why most workplaces are responding the wrong way.

I've been writing code, advising clients and building products for over 25 years.

The last two years have been different.

AI is moving fast enough that even people inside the industry are struggling to keep up. Every week brings a new model, a new capability, a new piece of work that used to take a day and now takes a minute. The pace is genuinely unprecedented. That is coming from someone who lived through the dot-com bust, mobile, cloud, and the social media wave.

My LinkedIn is full of people being so certain about AI. Every other founder is building the next unicorn thanks to AI. Investors are looking for the next big thing in AI, and software engineers building 10X whilst sat on the beach all via their phones.

The marketing is good. AI is here to stay, and it will change a lot of things for the better.

But we have a habit of noticing the consequences of new technology too late.

The damage social media did to young people's mental health was visible to anyone paying attention by about 2015. Society responded, properly, almost a decade later. Short-term lets were good for travellers and bad for the residents of Barcelona, Lisbon and Amsterdam long before any of those cities acted. The pattern is consistent. New technology arrives, it disrupts, and no one notices till it is too late.

Disruption comes at a cost. The fear with AI is that some of the cost will be paid in mental health.

We do not yet know the long-term effects on critical thinking or on judgement. Although the signs are not good, the real impact will be seen when more data is available.

What we can already see, though, is the effect on people working in tech right now. That part of our future is something that is happening now.

In the trenches of software engineering these problems manifest in junior developers thinking maybe they should have made a different career choice. Senior developers not sure if they should mentor their juniors (the ones lucky enough to find a job) on how to code or simply how to use Claude Code.

If you are a CFO, CTO or CEO seeing all the hype about AI, the risk of simply ignoring such a groundbreaking technology is too great. Incorrect implementation without considering the wellbeing of your team is even greater, but not always seen.

A founder who is certain about every aspect of his start-up is now asking AI "is my company viable in 2026, 2027, 2028" at one in the morning. Getting two conflicting answers on two different nights and being more unsure of himself than before asking the question.

A clear example of this is happening in San Francisco. In February, the writer Nikunj Kothari published an essay called Token Anxiety, subtitled "Scenes from San Francisco, February 2026". I originally came across this via a video on YouTube via The Prime Time. Kothari's description of dinner parties this year went as follows:

"A friend left a party at 9:30 on Saturday. Not tired, not sick. He wanted to get back to his agents. Nobody questions it anymore. Half the room is thinking the same thing. The other half is probably checking on the progress of their agents at the party."

He goes on to describe how the new flex among developers is not what you have built. It is what is working while you are sitting there not working. The CEO of Y Combinator has publicly given up drinking because of Claude Code, because he needs his brain "maximally pristine" to ship ten thousand lines a day. Anthropic doubled its revenue projections in two months. Some engineers are spending tens of thousands of dollars a month on AI tools. One processed the equivalent of thirty-three times the entire content of Wikipedia in a single week.

The writer Jae Kaplan put it bluntly in a response to the piece: a tool that produces randomised, low-quality output and demands constant attention is a slot machine. Token anxiety is real to engineers who previously took great pride in the code they hand wrote.

We have now gone beyond taking your work home with you. You are taking it to bed and to the toilet with you.

AI anxiety is different to other workplace anxieties. Most workplace worries decay with time and reassurance. You worry about a layoff round, the layoffs happen, you survive, the anxiety eases. AI anxiety doesn't decay because the thing causing it keeps advancing. Each new model release brings it back. The reassurance you got six months ago no longer applies. It compounds rather than passes.

A lot of workplaces are getting the response wrong.

"AI won't replace you. People using AI will replace people who don't." It has become the standard reassurance in tech, and it is the most anxiety-generating sentence anyone is currently saying. It tells people the bar is moving and that they are personally responsible for keeping up. It does the opposite of reassuring.

Mandatory AI-usage metrics tell people they are being measured on something they don't yet understand. Town halls where leadership talks "AI vision" without naming what is actually changing leave people more uncertain than they were before. The companies pretending none of this is happening are in a worse position than everyone else.

We are not going to slow the technology down. That ship has sailed, and no one is saying we need to slow it down. What we can do is talk about it honestly, and start measuring the thing that is actually breaking before it breaks.

AI anxiety is real, and organisations need to learn to measure and understand it. Rather than ploughing millions into AI implementations and boasting about how many people have been laid off last quarter, companies would do better to understand the risks of AI mis-adoption on the wellbeing of their staff.

Laying off 20% of your workforce has an adverse effect on the remaining 80%. Just because you survived the cull doesn't mean you are any more at ease.

Those are the things we should be tracking, and they are not the things most workplace wellbeing platforms track.

That is the gap we are working on at Occlio. It is also why I am writing this newsletter.

The point of The Measured Workplace is to open an honest conversation about what is actually happening to people at work as the AI wave keeps moving, and about what serious measurement of workplace mental health looks like.

As I was finishing this piece, Business Insider published a feature on developers walking around in public with their laptops held half-open, just to keep their AI agents running. Through airports. Down school hallways. Across ice rink car parks. One of them, a head of product with two young daughters, codes through their weekly skating practice. When the parents flood into the changing room at the end, he goes with them, laptop ajar. He is untying his ten-year-old's skates while glancing back at the screen. Is it done?

First published on LinkedIn on 13 May 2026 as issue 1 of The Measured Workplace, the Occlio newsletter.

More in ViewsAll articles