An AI story I hear often sounds like this: a company adopts AI, announces layoffs, and the two get reported as cause and effect. One side calls it cruelty dressed up as progress. The other calls it efficiency. Then everyone moves on to the next announcement.

Both reactions are first-order. The current arrangement of work is unquestioned as a given and arguments focus on who absorbs the shock. I'm incline to zoom out — to the parts of the story that don't make headlines — and work backward to the jobs, because I think current narratives are insufficient; not informative, useful, or realistic.

Bullshit jobs exist and are common

David Graeber wrote about "Bullshit Jobs": work that even the people doing it personally feel lacks meaning and/or value. The report nobody reads. The approval that approves nothing. The meeting that exists to prepare for the meeting.

Here's the uncomfortable question underneath that term: these jobs didn't appear by accident, and they don't persist by accident. Bullshit jobs persist because a job does more than produce output. It distributes income, health insurance, stability, and political goodwill. Organizations — especially public ones — carry an unstated goal alongside their stated mission: preserve and provide employment. No statute says so. No performance plan mentions it. But watch what actually shapes decisions about headcount, office locations, consolidation, and automation, and you can see it operating.

I say this as someone who spent more than fifteen years in public service and will draw a public pension. This isn't an outsider's contempt. It is a personal accounting, informed by years of observations and my experience. The unstated jobs goal isn't a conspiracy; it's what emerges when workers reasonably defend their livelihoods, managers gain standing from headcount, elected officials protect payroll in their districts, and voters expect government to provide a stability that private employers don't.

When people fight to preserve jobs that automation could replace, they are not confused. They are defending status quo, because nobody has shown a credible road out of it. The local maximum we're stuck on — where hollow work persists because the paycheck attached to it is load-bearing — isn't a misunderstanding to be corrected. It's a rational position to be made unnecessary.

That's the third-order story: not "will AI take jobs," but "what happens to a society whose main system for distributing income and belonging was quietly running a jobs program?" What happens when the pretense stops working?

AI as a tool for motivated hands

Before going further, I want to reject a lazy assumption on my own side of the argument: that AI does administrative and clerical work worse, and automation is therefore just cost-cutting with a quality loss pushed onto the public. Sometimes that's true. Sometimes it is not.

Here's what I actually expect to happen, because it's already happening: the motivated employee picks up the tool first. The analyst who always resented spending four hours reformatting data before one hour of actual analysis. The caseworker who wants to spend time with people, not files. The clerk who knows the backlog is an embarrassment and now has a way to clear it. These people don't experience AI as a threat. They experience it as the removal of toil the can now offload to improved technology. Working smarter. Their work products get better. The service gets faster. And their week fills up with the parts of the job that required a human all along.

This is the second-order effect hiding inside every agency and office: the gap opens up not between humans and AI, but between people who use the tool to level up their contribution and people whose entire role was the toil. For the first group, automation is a promotion in everything but title. For the second, it's an audit — one that most organizations have avoided giving for decades.

The public sector should want this audit. Delivering the same service with less labor and the same people doing more valuable work is what public service is supposed to mean. The only reason to fear it is the unstated goal — and the fear is legitimate exactly to the degree that we've made losing a job catastrophic. Which brings me to what jobs were actually doing for us.

What the job was carrying

A job is a bundle. It carries income, obviously. But it also carries structure, identity, colleagues, and an answer to "what do you do?" Most of all it carries something we rarely name: the experience of being needed. Someone depends on your work, notices it, is helped by it.

Look at retirement research and you'll see how real this is. Plenty of people with fully secure incomes deteriorate without the structure and purpose work gave them. Income transfers don't transfer meaning. This is why "just give everyone a check" is an incomplete answer, however necessary some version of the check may be.

But notice where the meaning actually came from. Rarely the task itself — few people found transcendence in the spreadsheet. It came from needed contribution, and employment happened to be the container that organized it, made it visible, and paid for it. Parents have that experience without a job title. So do volunteers, coaches, caregivers, and open-source maintainers. The job was the container, not the contents.

The second-order opportunity, then, is not "save the jobs" and not "cut the checks." It's build new containers: organized, supported, witnessed ways to contribute — care, teaching, craft, stewardship, civic work, and the enormous backlog of things worth doing that were never profitable enough to become jobs in the first place. Anyone who has spent time in local government knows how long that backlog is.

Now the shock, in its proper place

Only now do I want to talk about the layoff — the thing the story usually starts with.

The shock is real. A layoff is a household losing income, insurance, and plans at once, because the job bundled all of them together and the severance cut the whole bundle. Nothing above makes that less painful, and "you've been freed from a bullshit job" is the winner narrating the loser's experience.

But the shock is a symptom, not the disease. It hurts as much as it does because we routed income, belonging, and contribution through a single instrument and made losing that instrument a fall off a cliff. The first-order framing — automation causes layoffs, layoffs cause suffering — is true, and it keeps our attention exactly where the deeper problem wants it: on defending the bundle, instead of on unbundling it.

The gains from automation are real; that's the whole premise. The question is whether they visibly finance the transition — a floor under income, and investment in those new containers for contribution — or concentrate quietly while the displacement disperses loudly. If people can see the road off the local maximum being paved with the surplus the automation created, the fortress stops being necessary. If they can't, they will vote to bring the jobs back, including the hollow ones, and they won't be wrong to.

Technology moves in quarters. Institutions move in decades. A displaced person's life moves in years. Any automation that's a triumph on the first clock and a catastrophe on the third hasn't found a higher maximum — it's just standing on someone.

The layoff headline is the smallest, truest part of a much larger story. Third order: the jobs system was carrying weight it never admitted to. Second order: the tool is already lifting the toil out of real work, and the people using it are showing us what work could become. First order: the shock — which is what we get whenever we change what a job does before changing what a job has to be.

That last part is the actual work. Not resisting the tool, and not celebrating the layoffs. Changing what a job has to be, so that leaving a hollow one stops being a leap off a cliff, and losing one stops being a catastrophe.