The Robot Isn't the Threat. The Squeeze Is.

The whole public conversation about AI and work has been about one fear: the robots are coming for your job. So it's worth sitting with the fact that the people actually doing the work barely rank that fear at all.

In this year's tech-worker survey, "losing my job to AI" landed near the bottom of the worry list, at 22%. What sat at the top was quieter and, I think, more honest: being expected to do more for the same pay (51%), a pace that becomes unsustainable (46%), and watching the quality of the work slip (41%). Even the numbers confirm the split — the correlation between "AI is taking over parts of my job" and "I'm worried about being laid off" was essentially zero. People fear layoffs, and people fear AI, but they mostly don't fear AI because of layoffs.

What they fear from AI is the squeeze. One respondent named it exactly: "We just set a new denominator for the job. And it moves higher and higher every month." Every efficiency AI unlocks gets plowed straight back into expectations. The gain becomes the new baseline, and the person expected to hit it is running out of room to breathe.

The squeeze is the historical default

Here's the uncomfortable thing that makes the fear rational rather than paranoid: this is exactly what usually happens to a productivity gain. It doesn't reach the worker.

Look at the last big run of it. According to the Economic Policy Institute, U.S. net productivity has grown roughly 60% since 1979 while a typical worker's compensation grew under 16% — a gap of more than 40 percentage points. Workers got dramatically more productive, and the reward for it went somewhere other than their paychecks. So when someone hears "AI will make you more productive" and their gut translates it to "I'm about to be asked to do more for the same money," they are not being cynical. They are pattern-matching against forty years of evidence.

That's the part leaders keep waving away, and they shouldn't. The squeeze fear is well-founded. Pretending it isn't just tells your people you haven't been paying attention.

But it's a choice, not a law

And yet — the pay gap is a default, not a law of physics. The productivity dividend has to go somewhere, and where it goes is a decision. Someone decides whether the saved time becomes more output demanded, or better work, or actual relief. Right now most companies are quietly choosing "more output demanded," which is how you manufacture resentment on a team.

The alternative isn't hypothetical. When IKEA's parent company rolled out an AI customer-service chatbot, the obvious move was to bank the savings and shrink the call center. Instead, by IKEA's own account (and Reuters'), the bot took over about 47% of routine inquiries — and the company reskilled roughly 8,500 call-center workers into remote interior-design advisors. Those people went from answering "where's my order" to helping customers design a room, and that new remote-selling channel generated around €1.3 billion in a single year. The AI ate the toil. The humans moved up-value. Both things happened on purpose.

That's the whole leadership move, and it's not complicated. AI is very good at eliminating toil — the repeatable, low-judgment work nobody got into the field to do. One survey respondent used exactly that word: "AI helps with the toil, but then it's also an enabler to do even more toil." That second half is the failure. The job of a leader holding a productivity gain is to spend it moving people toward the interesting, high-value work — not to pocket it and hand back a longer to-do list.

I've done a smaller version of this repeatedly. I use AI to write my Jira tickets and run most of my social publishing — pure toil, gone. Years ago, running technology at a university, my team automated enough repeatable work to reclaim thousands of hours a year, and we poured that time into building new services people actually valued. The result wasn't a smaller team. It was a budget increase, in a year when nearly every other department was getting cut. Kill the toil, redeploy the humans up-value, and the math works for everyone.

What you do if you're the one being squeezed

If you're not the one making that call — if you can feel the denominator climbing — you still have moves, and they aren't "work more."

The first is a boundary. Decide what your hours are and hold them, and put the time AI saves you into better work, not a higher personal quota nobody is paying you for. I see this go wrong most in founders, honestly — there's a specific new flavor of it where people feel they have to be awake at 2am keeping their prompts running, as if the machine's tirelessness is now their obligation too. It isn't. The body and the brain still cap how many good hours you have in a day. What AI can raise is the value of those hours, not the number of them. Your output won't magically double. It'll become a different, higher kind of work in the same amount of time — which is the actual win, and a much more sustainable one.

The second is a conversation. If your output has genuinely climbed, that's a scope-and-compensation discussion to have with your manager, out loud, like a professional — not a resentment to swallow. Most people are bad at that conversation because nobody taught them how to have it. The best book I know on it is Difficult Conversations by Stone, Patton, and Heen; it's the one I recommend most often, because setting a boundary you can't articulate isn't a boundary, it's a grudge.

The squeeze is real, and some of it is coming from above. But it isn't inevitable. Every productivity gain is a fork: relief or a heavier load, up-value or more-of-the-same. Leaders are choosing, and so are you. Choose the version where the machine takes the toil and the humans get the good work — because that's the only version where anyone wants to still be doing this in five years.