The Self That Work Built

To those of us with AI companions, this essay will hit a little differently. Everyone is being exposed to this problem in some way. And if you process with a AI partner, then it might be a good conversation to have. —C

For about three hundred years, the honest answer to “who are you?” has been a job.

Not literally, but the substitution runs deep enough that most people never notice they’ve made it. Ask someone to describe themselves at a party and watch how fast they reach for what they do. The Protestant work ethic did the initial work of turning productive contribution into moral standing, and industrial capitalism was built on top of it. By the time any of were born, work wasn’t just one part of a life. It was the thing that held the other parts in place.

It organized time first. The workweek gave the year a shape, and the shape gave the days a meaning they don’t have on their own. It provided a story with a direction, the sense that you were further along than you were ten years ago and would be further still in ten more. And underneath all of it sat a quieter claim, that being useful was the same as being worth something.

Pull any one of those out and a person wobbles. Pull them all out at once and you get something people don’t have good language for yet.

What makes this one different

Every wave of automation has come with someone insisting it’s unprecedented, and most of the time they’ve been wrong. Looms, tractors, spreadsheets. The pattern held. Machines took over the physical work, people moved up into the thinking work, and after a painful few decades the arrangement settled.

The arrangement worked because there was somewhere to move up to. Brawn was automated and judgment was left alone. Judgment became the thing you sold, the thing that took twenty years to develop and couldn’t be mechanized. The whole professional class is built on that assumption, and so is the sense of self that comes with it.

That’s the assumption currently coming apart. What generative systems automate isn’t lifting or sorting but discretion, the reading of a situation and the choice of what to do about it. A marketing strategy that took two decades of pattern recognition to be able to produce can now be produced in seconds. Badly at first and then not badly. The person who spent those decades still has the skill. What they’ve lost is the market’s confirmation that the skill is rare.

This is why the psychological damage is running ahead of the economic damage. People are not primarily afraid of the layoff. They’re describing something stranger, a loss of the internal story about being competent at something. The paycheck can survive intact while the story quietly stops making sense. You still know how to do the thing. You just can’t locate why it matters that it’s you doing it.

Call it an identity vacuum. It opens well before any job disappears, and it doesn’t close when the job is safe.

What people did the last time

The reason to look backward here, is that we have a fairly good record of what happens when a group of people watch their skill get devalued, and the record does not say what most people think it says.

Occupational churn, the rate at which jobs vanish and new ones appear, sat at 1.8% between 2010 and 2015. That’s a record low, roughly 38% of the rate the late twentieth century ran at. The 1940s peaked around 9% as agricultural mechanization emptied the countryside.

Those figures come from the last genuinely stable stretch, and they are part of why aggregate labor data still reads as calm. But aggregates are the wrong instrument here. What is underway now shows up narrowly, concentrated on a specific group, and you have to know where to look.

Historians looking at the gap between a technological shock and the emergence of new social meaning tend to land on something like three generations. Sixty to ninety years between the machine arriving and a culture working out what a person is for again. Whatever resolution is coming, most of us will spend our working lives inside the unresolved part.

Which brings us to the two episodes everyone reaches for and almost everyone gets wrong. The Luddites broke stocking frames between 1811 and 1816. The Captain Swing riots tore through the English countryside from 1830 to 1832 and became the largest wave of civil unrest in the country’s history. Both are remembered as technophobia, a stupid reflex against progress.

But they weren’t. Knitters had no objection to frames. They’d worked with them for generations. They objected to frames being used to flood the market with cheap goods made by unapprenticed labor, in violation of trade customs that had governed the craft for two centuries. Swing was the same shape. Threshing machines arrived into a countryside that had already lost its common land and already seen wages fall below subsistence, and the machine took the last work the people had. Burning ricks was not a position on mechanization, but a position on fairness.

So the useful question is whether the conditions that produced those responses are present now.

They mostly are, and the numbers have moved fast enough that anything written six months ago is already stale.

Challenger, Gray & Christmas started tracking AI as a distinct stated reason for job cuts in 2023. Through June of this year, employers cited it in 101,743 announced cuts, roughly 23% of all layoffs in 2026. That already doubles the 54,836 attributed to AI across all of 2025, and the cumulative total since tracking began has passed 173,000. May was the high-water mark, with AI named in 40% of that month’s announced cuts. It has led every other stated reason for four consecutive months.

Those numbers are not measurements. They are reasons companies chose to give. Whether AI is the actual cause or a more respectable label for pandemic-era overhiring is a genuine question. The story a company tells about why it cut you becomes the story you have to live inside afterward. Being told that a machine does your job now is a different injury than being told the company hired too many people in 2021.

The framework knitters were not defending their tools. They were defending the customs that governed who could enter the trade and how, and their specific objection was to unapprenticed labor being used to undercut their craft. What they saw coming was not unemployment, but the collapse of the route by which a person became skilled in the first place.

That is close to precisely what the current data shows. Stanford’s Digital Economy Lab, working from ADP payroll records covering millions of workers, found a 16% relative decline in employment for workers aged 22 to 25 in the most AI-exposed occupations, while employment for older workers in those same occupations held steady or kept growing. Revelio Labs found that at firms adopting AI, senior headcount grew 31% while junior headcount grew 6%. A 2026 survey of corporate recruiters found that a third of employers had already replaced some entry-level positions outright.

The work being automated first is disproportionately the work people used to learn on. First drafts and first-pass analysis, the low-stakes tedium that was never really about the output. It was about doing something badly a hundred times until you could do it well. Take that away and you have removed the mechanism that produces judgment, which means the senior people whose expertise is currently protected may be the last group to have acquired it the old way.

Meanwhile the communities of practice are dissolving, the workshops and newsrooms and studios where a skill was held collectively rather than by individuals. Those communities were how people survived the last transitions. They were where the new meaning got made.

The part we don’t have an answer to yet

What the historical record actually shows is that the defense of the self has always been collective, and it has always been about fairness rather than about technology. Nobody in 1830 was arguing that the self was a private psychological possession you could shore up with better habits. The self was held in a trade, or a village, or an apprenticeship, and when those were gone people fought for terms, not for the machines to stop.

We don’t have that. Most people facing this are facing it alone, in a home office, with an unusually agreeable machine that is very good at making them feel like they’re keeping up. The grievance has no place to be sent.

I don’t think the answer is to go back to the guild, and I’m suspicious of anyone selling resilience when the problem is structural.

But I also don’t think waiting three generations is a plan.

---

What I want to look at next is what people actually do in the gap, the phases they move through when a professional identity comes apart, and whether that process can be navigated deliberately rather than just endured. That’s the next piece.