Every conference I have been to in the last twelve months has the same moment.
Someone on stage, talking about AI, reaches for the Industrial Revolution. The crowd nods. The point gets made. Whatever the point was, the comparison did the heavy lifting, and nobody asked the next question.
I have been asking the next question for a year, and the answer keeps getting more uncomfortable.
The Comfort Reading and the Horror Reading
There are two ways to use the Industrial Revolution as an AI analogy, and both of them are wrong by themselves.
The comfort reading goes: "We have been here before. People panicked about looms and trains and tractors and computers, and every time we adapted, and every time the new economy was bigger and richer than the old one. Relax. We will adapt again." This is the version executives use right before they cut a department.
The horror reading goes: "The Industrial Revolution destroyed entire communities. It took generations to recover. Real people lived in real misery for decades while the productivity gains accrued somewhere else. AI is going to do that again, only faster." This is the version doom-scrollers use right before they ask whether their kid should bother with college.
Both of these are technically true. Both of them are doing the same thing, which is using history as a feeling instead of as data. The interesting part of the Industrial Revolution is in the middle, not at either end.
What Actually Happened to the People in the Middle
There is a phrase that economic historians use that almost nobody outside that field has heard. Engels' Pause.
From roughly 1780 to 1840, British output per worker climbed sharply. The factories worked. The machines worked. The economy got more productive at a rate the world had never seen. And the wages of the median British worker barely moved. For sixty years.
Robert Allen formalized the math on this in 2009. The pie was growing. The slices were not. The eventual broad wage gains everyone points to when they reach for the comfort reading? Those happened in the 1860s and 1870s. The workers of 1820 did not get to enjoy them. Their grandchildren did.
The lesson of the Industrial Revolution is not that things work out. The lesson is that things work out for the people who get to define the new categories, and brutally for the people who get defined by them.
Sit with that for a second before you decide which reading you have been using.
They Re-Defined School to Match the Factory
Here is the part that should haunt anyone in education.
The American high school as we know it did not exist in 1900 in any meaningful way. By 1940, it was nearly universal. Goldin and Katz documented the whole arc in "The Race Between Education and Technology," and the through-line is brutally clear: the high school movement was a deliberate, decades-long re-architecture of public schooling around the skills that mass-production employers needed.
Literacy and numeracy were not new in 1910. The decision to require them at scale, on a schedule, with credentialing, on the public dime, was. School got rebuilt to match the factory, and it took thirty years, and the kids who graduated from the new system walked into a world that was waiting for them. The kids who aged out of the old system mostly did not.
The previous tech transition rewrote what counted as an education. There is no reason this one will not.
If you are an educator reading this and wondering why I keep coming back to "are we training students for a world that exists," it is because the last time we asked that question seriously, we built the entire modern school system around the answer.
The Luddites Were Right About the Wrong Thing
The word "Luddite" got loaded with a meaning its original users would not recognize.
The Luddites of 1811 to 1816 were not anti-technology. They were skilled textile workers who had specific labor agreements with specific employers, and those employers started using specific machines in specific ways that broke those agreements. The Luddites smashed the machines that were used to break the deals. They left alone the machines that were used inside the deals.
Brian Merchant's book "Blood in the Machine" walks through this in detail, and the part that stuck with me is that the Luddites lost the argument because they were arguing about contracts, and the people who beat them framed it as being about progress. Once "anti-Luddite" came to mean "pro-progress," the actual workers' grievance disappeared from the historical record for almost two hundred years.
The 2026 version of this argument is happening right now. Every "AI is just a tool, get used to it" framing is doing the same rhetorical move the Industrial Revolution winners did. It is not about whether the tool is good. It is about who gets to decide how the tool is used, and against whom.
Same Story, Different Cameras
Here is what this looks like outside the white-collar bubble in 2026.
The alarm monitoring industry is running its own miniature Engels' Pause right now. AI video analytics can scan thousands of camera feeds in parallel, filter out the squirrels and the headlights and the drifting clouds, and route only the events that look like a real human doing a real thing to a human operator. Companies in the space report false-alarm reduction rates north of 90%, which means an operator who used to react to four hundred alerts a shift now reacts to forty.
The job did not disappear. It got redefined.
The task that used to define the role (watch the screen, react to motion) is gone. The task that defines the role now (decide whether a flagged event is real, talk to the customer, dispatch the right response, escalate when something is off) is the part the screen-watchers were never paid extra for. Same job title. Different portfolio. The wage trajectory is moving, but it is moving up for the operators who can do the new portfolio and flat-or-down for the ones who cannot.
I talked to an operator a few weeks ago who told me her job got "more interesting and less safe" the same week. More interesting because the boring parts left. Less safe because the company suddenly noticed she could handle three times the volume, and the math on how many operators they need has changed.
That is exactly the shape of 1820. Productivity went up. The operator's wages have not, yet. Whether they do depends on choices that are being made right now, not by her.
The Honest Part
Here is the part I am not going to soften.
The "we adapted before, we will adapt again" line is technically true and morally lazy. The British workers of 1820 did not get to enjoy the wage gains of 1880. Their grandchildren did. If the AI transition takes even a decade, and there is no serious analyst who thinks it will take less, most working adults reading this will spend a meaningful part of their career inside the gap, not on the other side of it.
Telling someone whose job is being redefined this year that "history shows things work out" is the same as telling a 25-year-old loom worker in 1820 that her great-grandchildren will be fine. It is true. It is not a plan.
The lesson of the Industrial Revolution is that the productivity gains happen on one timeline and the human gains happen on a different, longer one, and the gap between those two timelines is where actual people live their actual lives. The question for 2026 is not whether we will adapt. The question is what we are willing to do for the people inside the gap, and whether we are willing to be honest that the gap exists.
What I Am Doing Differently
I keep two questions taped to the inside of my notebook now.
What is the next thing I am going to stop counting as work?
Who decides?
The first question keeps me honest about my own job. I have already stopped counting first drafts as work. I have already stopped counting research grinds as work. Five years ago, I would have argued that either of those was the core of what I do. Now they are the part the machine handles, and the work is somewhere else, and the somewhere else is harder to defend at a budget meeting because it does not have a clean unit of measure.
The second question is the one I want educators, execs, and parents to start asking out loud, because the Industrial Revolution answer was "the people who owned the factories," and that answer made the gap longer and meaner than it had to be. The 2026 answer is still being written, and most of the people who will be affected by it are not in the room where it gets decided.
If you are reading this and you are in one of those rooms, the question is for you.
If you want to think through what your own portfolio of tasks is going to look like on the other side of the gap, that is what I am building over at bensaibrain.com
Sources
Created with ❤️ by humans + AI assistance 🤖