The Survivors' Pose: From Impressionism to AI
Where did the painters go after the camera arrived?
TL;DR
Where did the painters go after the camera arrived? The textbook answer is that they invented Impressionism — they stopped trying to depict reality and started depicting perception. It’s a comforting story. We project ourselves into it every time we talk about AI and what it will do to knowledge work.
But it leaves out most of the painters. History only records survivors. This article traces four fates from the photography revolution — disappearance, escape, working around, and invention — and asks which script we are following in the AI age.
The textbook answer is that they invented Impressionism. They stopped trying to depict reality and started depicting perception. Monet painted not the sunrise but the impression a sunrise made on him.
It’s a comforting story. When the machine takes over a human function, humans find higher ground. We project ourselves into this story every time we talk about AI and what it will do to knowledge work.
But it leaves out most of the painters.
What the camera destroyed was the Academy. The dominant system of nineteenth-century European painting, governed by the Académie royale de peinture et de sculpture founded under Louis XIV in 1648, sustained by a market that paid painters to render reality with precision and high finish. The camera took that entire function and handed it to a machine. A few painters migrated and became the Impressionists. Most did not.
History does not tell us their stories. Because history only records survivors.
We use the survivors’ script to imagine ourselves. We see Monet, and we see ourselves in him. The real question runs the other way.
Which function of knowledge work has AI replaced, and what happened to the people who did not make the migration?
Part I: Nadar’s Red House
Start here because this is where the Impressionist uprising happened, and the building belonged to a photographer. The painters who held the show were his personal friends. The supposedly anti-photographic uprising was soaked in photography.
The story is more tangled than the textbooks let on.
April 15, 1874. 35 Boulevard des Capucines, Paris. A four-story building, exterior painted red, with a 15-meter metal sign across the front bearing one word: NADAR.
This was the most famous photography studio in Paris. Its owner, Félix Nadar, lent it for free to a group of painters the official Salon had rejected. Among the works on display was a painting called Impression, Sunrise. Three weeks later a critic named Louis Leroy used the word Impressionism in Le Charivari to mock the show. That word eventually named an epoch.
The textbook tells this day as an uprising. The machine arrived, the painters refused to be replaced, they revolted, they invented new art, they won.
But Degas was an obsessive amateur photographer. Manet worked from Nadar’s photographs. Monet’s other painting in the show, Boulevard des Capucines, was painted from a second-floor window inside Nadar’s studio.
This doesn’t look like an uprising. It looks like a group of painters struggling through the age of photography, holding a show in a photographer’s house.
And there’s a piece missing from the picture...
The Person We Still Do Not Remember
Sarah Goodridge was born in 1788 in a Massachusetts town called Templeton. Her family was so poor they couldn’t afford paper. As a child she drew with a stick on the sanded kitchen floor, or scratched figures into peeled birch bark with pins. In 1820 she opened a portrait miniature studio in Boston. She never married. With her brush she supported her mother, her sisters, and several nieces. For a woman in early nineteenth-century America, this was rare.
A portrait miniature was a tiny watercolor on a slip of ivory shaved paper-thin, about the size of a calling card, set in an oval frame under curved glass and worn as a pendant or tucked between the pages of a book. It served neither the rich, who had oil portraits on their parlor walls, nor the poor, who had no portraits at all. It served the middle. Doctors, lawyers, merchants, ministers. People who wanted to carry a loved one’s face with them when they traveled.
The profession was practiced disproportionately by women. Nineteenth-century society offered very few respectable ways for a woman to support herself. Teacher, nurse, seamstress, governess. And then handicraft. Portrait miniatures could be made at home. They didn’t require the woman to appear in public. For an educated middle-class woman, the craft was one of the few paths through which she could earn a living without depending on a husband or father.
On August 19, 1839, the daguerreotype process was announced publicly at the French Academy of Sciences in Paris. A daguerreotype portrait cost about one-fifth what a miniature cost.
Many miniaturists abandoned the craft. Some retrained as photo colorists, hand-painting color onto black-and-white photographs, demoted from artists to assistants. Most just disappeared.
Goodridge’s eyesight began to fail in 1850. By 1851 she had stopped painting and moved in with family in Reading. She died on December 28, 1853. Boston had had dozens of miniature studios. By the 1860s there were essentially none. A complete profession, with two hundred years of history, evaporated in twenty years.
She is not an abstraction. She was a person with a name. She lived sixty-five years. She supported her sisters and her nieces. She represents the people who, after technological replacement, neither escaped successfully nor were absorbed into the new industry.
Their story goes like this. Orders started declining in the 1840s and dried up in the 1850s. Some took secondary jobs in the new industry. Some lived off savings until they died. Some moved in with their grown children. Most of them were women. Most were middle-skill. Most had supported families with their craft.
Art history doesn’t remember them, because art history remembers the people who invented new art, not the people whose old profession was buried.
Our entire understanding of the photography revolution rests on the version art history wrote. Monet, Manet, and Nadar are in that version. Goodridge is not. So when we talk about AI today, the path that comes to mind is Monet’s. Humans find new ground.
It isn’t so. For the overwhelming majority of people displaced by technology, the actual shape of displacement is not migration. It is disappearance. No name, no story, no second act.
Is there another way of living?
Way One: Escape
The standard story is that the camera arrived, painters were forced to turn from depicting reality to depicting subjective perception, and out of this Impressionism was born.
It’s too clean. Impressionism didn’t come from the camera alone. Kant had argued since 1781 that we don’t perceive things in themselves, only representations the mind has processed. The birth of psychology in the late nineteenth century turned perception itself into a scientific subject. The invention of paint in metal tubes in 1841 made outdoor painting practically possible. The rise of the middle class created a new market for art that didn’t have to be historical or religious. The camera was one trigger among several, not the only one.
What matters more is that for the painters themselves, escape was never a warm story.
In the summer of 1868, a twenty-seven-year-old Monet jumped off a bridge into the Seine. His landlord had thrown him out and he was deep in despair. He was lucky enough to know how to swim, and in the water he changed his mind and swam back. On June 29 he wrote to his old friend Frédéric Bazille, admitting that he had been so depressed the day before that he had been stupid enough to throw himself in.
Later in life he recalled a stretch of the 1870s. He had to sell a painting Renoir had given him for 300 francs because he needed food. He owed the butcher several hundred francs. He knew the bailiffs were coming to seize his paintings, he had about two hundred, so he took a knife and slashed every one of them. When he came back to settle his accounts a few weeks later, he discovered the slashed paintings had been sold in lots of fifty for thirty francs a lot. Sixty centimes per painting. In 1875 Paris, sixty centimes bought you a kilogram of bread.
He wasn’t an exception. Through the 1870s most of the Impressionist circle survived on borrowed money, family support, debt extensions, and their wives’ dowries. Pissarro was poor his entire life. Sisley died in poverty. Monet outlived all of them, to eighty-six, which is why he lived long enough to be recognized.
But living to 86 was luck, not a strategy.
There is also a deeper question. Did Monet actually escape to a place the machine could not reach?
Look closely. Impressionism’s techniques, the outdoor sketches, the captured moments of light, the visible brushwork, the unusual cropping, all of these were deeply shaped by photography. The Impressionist market, middle-class apartments wanting wall decoration, was a side effect of the photographic age. The Impressionists themselves used photographs for reference.
Escape isn’t escape to somewhere the machine can’t reach. Escape is finding a new function the machine can’t do yet. In the 1870s that function was subjective perception. But it lived alongside photography, was shaped by photography, and evolved with photography. It was not a safe shelter. It was a new corner where market demand existed and the machine hadn’t yet arrived.
The people who found that corner, survived inside it, and lived long enough to be recognized are a minority of a minority.
Way Two: Work Around
The painter who actually made the most money in the 1870s art market wasn’t Monet. It was William-Adolphe Bouguereau.
He was an Academic painter. He didn’t rebel against the Salon. He was the Salon. Elected to the Académie des Beaux-Arts in 1876, promoted to Commander of the Légion d’honneur in 1885. His paintings often sold within days of completion. He was famous in Belgium, the Netherlands, Portugal, Spain, Italy, Romania, and the United States.
He painted mythological goddesses, religious scenes, idealized peasant girls. Extreme finish, exceptional technique, traditional composition. Degas and his circle invented a slur for the style: Bouguereauté, overrefined and oversweetened.
Here’s what the Impressionist victory narrative tends to bury. Bouguereau took photographs of his models and worked from them.
The painter the Impressionists scorned as a reactionary was quietly using photography as a tool. He wrote no manifestos against it. He didn’t throw himself into the Seine. He kept photographs in the corner of his studio, posed his models for the camera, used the prints as references, and went on selling his high-finish mythological goddesses at top prices.
Working around means quietly using the new tool to strengthen the old form. This was the shape of the actual market winner of the period.
The price for working around is that fifty years later history may mock you. Bouguereau’s reputation collapsed almost immediately after his death. He wasn’t seriously reassessed until the 1980s. He was buried for more than 70 years.
Way Three: Invent
On November 27, 1854, a man named André-Adolphe-Eugène Disdéri filed a patent in France. He had modified a camera by adding four lenses and a sliding plate divider so that a single glass plate could hold eight small portraits in one exposure. Cut them apart and paste them onto cards. He called the product the carte de visite.
This was the social currency of the 1860s. The middle classes collected the cards obsessively. Family, distant relatives, fiancés, friends, neighbors. Then celebrities: opera singers, politicians, royalty. Albums were invented specifically to hold them, a hundred or more per album. It was the Instagram of the 1860s.
By 1861 Disdéri was the wealthiest photographer in the world. His reported annual income was £48,000. His studio was the most luxurious in Paris. A German visitor called it the Temple of Photography, “unique in luxury and elegance.”
Then his own invention destroyed him. The technology was too easy to copy. Any studio with a four-lens camera could do it. Each new entrant pushed the price down. A carte that sold for twenty francs at the start cost one franc by 1875.
Disdéri had invented alone. He was defeated by a network of hundreds of mid-sized studios with marketing, capital, and distribution. In January 1872 he filed for bankruptcy. He filed again and again. By the end he was an itinerant photographer working country fairs, charging peasants a few sous for cheap portraits.
He died in Paris on October 4, 1889, at the Hôpital Sainte-Anne, a psychiatric asylum established by Napoleon III in 1867 for the indigent, the alcoholic, and the destitute mentally ill. He was penniless, deaf, and half-blind. Two days later he was buried in a free communal trench at the Bagneux cemetery.
The richest photographer in the world ended in a paupers’ ward and a mass grave. In under thirty years.
Inventing looks like the highest path. It is actually the most dangerous. The inventor of a technology is rarely its long-term beneficiary.
The First Path Is the Majority. Why?
Four people, four outcomes. Goodridge disappeared. Monet escaped to a new function. Bouguereau worked around with new tools. Disdéri invented the technology itself and was eaten by it.
Three of these are the minority. Goodridge’s is the majority. Why?
Each of the three survivor paths needs specific conditions that ordinary people don’t have.
Escape needs a new function, that function needs market demand, and the person doing it needs to live long enough to be recognized. Monet living to eighty-six was luck, not strategy. Pissarro stayed poor his whole life. Sisley died poor.
Working around needs a position of scarcity inside the old form. Bouguereau was an Académie member. Most Academic painters couldn’t match that.
Invention needs you to invent something the market will accept, then accumulate enough wealth before the market overtakes you. Disdéri failed the second half.
The ordinary person is middle-skill, has a craft, supports a family, and fills exactly the function the machine can replace. Then the clients go. The peers go. The next generation doesn’t enter the profession. The ordinary person doesn’t die dramatically. They simply fade.
History doesn’t record them, because there’s no story to tell.
Part II: The Same Script, in the AI Age
So you’re probably Goodridge. What is AI actually doing right now?
You can see it at three levels. The micro level is task replacement. The middle level is the collapse of middle-skill labor. The deepest level is the silent rewriting of how everyone thinks, including the people who aren’t replaced at all.
Micro: Technology replaces tasks, but sometimes the whole job is just a bundle of AI-replaceable tasks
The most important paradigm shift in labor economics over the past thirty years is this: technology doesn’t replace occupations. It replaces tasks.
The wording sounds like a quibble. It changes everything about who gets replaced.
A radiologist sounds like one job. But a radiologist does dozens of distinct things every day: reading X-rays, reading CT scans, writing reports, talking to patients, consulting with attending physicians, teaching residents, attending multidisciplinary meetings. Those are tasks. A job is a bundle of tasks.
When AI learned to read X-rays, which deep learning could do by around 2017, it didn’t replace the radiologist. It replaced one task in the bundle. The other tasks remained. The job survived, but with different content, different wage pressure, and a different training path on the way in.
In 2023 Tyna Eloundou and colleagues at OpenAI and Penn published a paper called GPTs are GPTs. It came out in Science in 2024. They broke 1,016 American occupations into tasks and estimated each task’s exposure to LLMs. The headline: about 80% of U.S. workers will see at least 10% of their tasks significantly affected by LLMs. About 19% will see more than 50% of their tasks affected. Nineteen percent is thirty million people.
There’s a corner the framework misses. Some jobs don’t get partially eaten. They get eliminated. Because the bundle of their tasks is exactly the bundle the machine can do.
Sarah Goodridge’s miniature painting worked this way. The core function was rendering a specific person precisely on an ivory chip you could hold in your hand. That function is 100% reproduction, 100% replaceable by photography, with no residual tasks. The job wasn’t partially hollowed out. It was eliminated.
Which knowledge jobs today have task bundles that match the AI bundle exactly? Proofreaders. Transcriptionists. Market research analysts at the execution layer. Translators of non-specialist material. Entry-level customer service. Telemarketers. Their core function is 100% rule-following cognition. AI doesn’t need to replace them. AI is them.
Middle: The middle goes first, and more thoroughly than it did in 1850
The assumption that technology replaces low-skill labor is wrong.
MIT’s David Autor proposed a different picture in 2010. Job polarization. The hardest hit isn’t the bottom. It’s the middle. Office clerks, line supervisors, bookkeepers, junior data analysts, call-center agents, bank tellers, junior paralegals. The bottom, restaurant work, cleaning, nursing, home care, and the top, financial analysts, doctors, lawyers, consultants, both hold, and in some cases grow.
Why? Middle-skill jobs happen to be the ones whose tasks are most easily rule-bound. Low-skill work needs a flexible body making in-the-moment judgments in unpredictable environments. (A janitor has to decide what to wipe first.) Top-skill work needs someone willing to take responsibility under incomplete information and persuade other people. The middle gets squeezed because the middle is exactly what’s encodable.
This is Goodridge’s actual position. She wasn’t the bottom. Street portraitists survived. She wasn’t the top either, Bouguereau was above her, and he survived. She was middle. She had skill, she had clients, she supported a family. The middle is the layer that gets hollowed out.
AI-age hollowing has a mechanism that the 1850s didn’t have. AI copies the middle worker’s skill directly into the new hire.
In 2023 Erik Brynjolfsson at Stanford, with Danielle Li and Lindsey Raymond at MIT, tracked 5,179 customer service workers at a Fortune 500 company. Their finding: AI assistance lifted overall productivity by 14%. For low-skill new hires the lift was 34%. For the highest-skill veterans the lift was nearly zero, and on some quality metrics, slightly negative.
A new hire on the job two months, with AI, performs as well as a worker without AI on the job six months.
This destroys the middle worker’s market value. If a new hire with AI matches the output of a middle worker without AI, what is the middle worker’s accumulated experience worth?
Zero. Or rather, the cost of a new hire plus an AI subscription.
It goes deeper. In August 2025 Brynjolfsson and colleagues at Stanford published Canaries in the Coal Mine. Using high-frequency wage data they found that employment of 22- to 25-year-olds in AI-exposed industries dropped by 16% relative to other workers. Young software engineers dropped by nearly 20%.
Why? If a new hire plus AI equals a middle worker, why hire a junior at all? Just hire the middle worker and give them AI.
But junior positions are the rung that lifts people from beginner to expert.
A lawyer doesn’t start with complex litigation; she starts with contract review.
A programmer doesn’t start by designing systems; he starts writing boilerplate.
When AI hollows out those rungs, the next generation has no path from beginner to expert.
Deepest: We posed for the camera. We are posing for AI now.
The 1860s in Paris saw something else happen. It wasn’t about any of the four fates above. It happened to everyone.
In 1840 no one knew how to pose for a camera, because there was no camera. Posing didn’t exist as a human bodily behavior.
By 1870 nearly every middle-class person knew the rules instinctively. Standing in front of the camera, chin slightly up, shoulders relaxed but straight, gaze a little off-axis. One hand resting on a chair-back or the edge of a coat. Nobody had taught these rules systematically.
A new bodily discipline had appeared. It was enforced by no law and promoted by no institution. It quietly, irreversibly rewrote how everyone presented themselves to others.
It wasn’t the replacement of a job. It was the silent rewriting of everyone’s daily behavior by a new technology.
Twenty-first-century cognitive psychologists studied this kind of phenomenon for fifteen years and gave it a name. Cognitive offloading.
In 2011 Betsy Sparrow and colleagues at Columbia published a paper in Science that opened this line of research. They had subjects read factual statements, telling half of them the information would be saved to a computer and the other half it would not. Then they tested memory. The group that was told the information would be saved remembered significantly less.
The brain is actively deciding what to internalize and what to delegate to external storage. Once it judges something can be offloaded, it stops internalizing it. The judgment happens below conscious awareness. Subjects couldn’t feel it happening.
In 1860 Parisians began posing their bodies for the camera. In 2025 we are posing our cognition for AI.
The second is deeper than the first. The first only changed how you appeared in front of a lens. The second changes how you think.
The last time you edited your resume, were you already guessing how an AI would read it? When you write an email, do you reflexively paste the draft into ChatGPT for polish? When you face a complex problem, how often do you stop and tell yourself, let me ask ChatGPT first?
Each of these small movements is a sample of cognitive offloading. And Sparrow’s experiment tells us it happens below conscious awareness. You can’t feel it.
The most uncomfortable part. AI doesn’t need to get stronger to replace you. It just needs you to keep codifying your work and handing it over. Every time you turn your judgment into a prompt template, every time you encode your client relationships into an automatable CRM workflow, you are helping some future version of AI take your job.
Sarah Goodridge was at least replaced by an external technology. She didn’t train the camera to replace her. Today’s middle-skill knowledge workers are, in considerable measure, being replaced by the workflows they have voluntarily codified.
Part III: You Have One Thing She Did Not
In Enchiridion 17, Epictetus wrote that we step onto a stage we did not choose and perform a script we did not write. What we can do is play our assigned role well.
But he didn’t have our situation in mind.
The stage we’re on, we didn’t choose. AI is not what we wanted. AI is here. The script we’re performing, we didn’t write. Task replacement, the hollowing of the middle, the severed pipeline between generations. None of it is ours.
But there is one thing that separates us from Epictetus’s actor. We have read the script.
Goodridge had not. She sat in her Boston studio painting miniatures until her eyesight failed in 1851, never knowing her entire profession was about to vanish.
Disdéri had not. When he patented the carte de visite in 1854 he had no idea that thirty-five years later he would die in a psychiatric asylum and be buried in a mass grave.
Bouguereau had not. At the height of his fame in 1880 he didn’t know history was about to bury him for seventy years.
We have. We have read the complete script of 1840 to 1880. Who survived, who didn’t, and what the survivors actually did.
But just reading the script doesn’t change the ending. The script is still being written. We’re already in the second act. AI replacement will continue. Autonomous AI is arriving. The middle will keep getting hollowed out. Junior positions will keep disappearing.
So what is reading the script good for?
It lets you choose how to carry yourself through it. What the people in the script did, and what their carriage translates to today. That is what reading actually gives you.
As a Worker
Let AI Do the Precision. Let Your Hand Do the Signature.
When the daguerreotype was announced in 1839, Henry Collen was the official Miniature Painter to Her Majesty, Queen Victoria’s appointed court miniaturist. Royal Academy trained. Silver Medal in 1821. Appointed in 1837. One of the highest positions in the British miniature trade.
The shock arrived. He didn’t write a manifesto against photography. He didn’t throw himself into the Seine.
He did this. Two years after the technology went public, in August 1841, he obtained a license from Talbot and became the first professional calotypist in London. Talbot took 30% of his revenue.
He didn’t abandon his craft. He produced hybrids: calotype photographs overpainted by hand using his miniaturist’s techniques. They were called photographic miniatures. The customer paid for two things at once. The camera’s precision, which Collen couldn’t beat. And his hand. Color, finish, artistic refinement, plus the stamp of the Miniature Painter to Her Majesty.
He took about a thousand portraits using the calotype process. The earliest surviving photograph of Queen Victoria is most likely a Collen calotype. In 1842 the Treaty of Nanking, which ended the First Opium War, was reproduced using Collen’s process. Almost no one else in the world could have done that job.
The hybrid lasted three years. In 1844 his arrangement with Talbot broke down and Collen withdrew from photography and went back to miniature painting. He didn’t become an Impressionist-style new artist. He didn’t become a Disdéri-style photo magnate. He didn’t get rich.
He lived to eighty-one. He died in 1879, in comfortable circumstances. When Goodridge died poor in 1853, Collen, nine years her junior, still had twenty-six more years of dignified work ahead of him.
Don’t treat AI as the enemy. But don’t shrink into AI’s executor either. Let AI do your precision and speed. Let your hand do what only you can: your judgment, your name, your willingness to be responsible for the outcome. Customers pay for two things. The part AI can do, let AI do. The part only you can do, defend.
The real lesson from Collen isn’t that hybrid strategies always work. It’s this. Try the new technology early. If the hybrid stops working, retreat to your original craft. Don’t lose anything. Live a dignified life to old age. This is more realistic than Monet’s ten years of starvation before recognition, and a great deal safer than Disdéri’s peak-then-collapse.
Do Not Bet on a Single Medium
Degas was an Impressionist, but he wasn’t Monet. He painted in oil, pastel, and gouache. He made sculpture, etchings, lithographs, monotypes. He took photographs.
When the oil market was torn apart by the Impressionist fights, he moved to pastel. When his eyesight started failing in the 1880s, he moved to sculpture, working by touch instead of sight. The wax and clay dancer figurines from his late period were made when he could barely see. He lived to eighty-three. In 1911 the Louvre acquired nineteen of his works, an honor usually reserved for the dead.
Painters who specialized only in oil either got lucky or disappeared. Degas put no chips on any single medium. His multi-medium practice wasn’t virtuosity. It was a backup plan for a changing body.
Don’t build your professional identity entirely around knowing how to use AI. AI tools turn over every six months. Today’s prompt engineer will discover in five years that they’ve trained themselves into an obsolete skill. Today’s carefully built RAG system will be replaced in three years by autonomous agents that do the same thing themselves.
Keep multiple paths open. One specialization plus a few adjacent capabilities. One day job plus a few independent projects. One company platform plus a few portable skills. Degas wasn’t more talented than his contemporaries. He was just less committed to any single medium.
Your Moat Cannot Be Something Anyone Can Buy
Disdéri ended in a psychiatric asylum and a mass grave because his value rested entirely on a technology anyone with a camera could replicate. Patents couldn’t protect him. Scale couldn’t protect him. His competitors had marketing, capital, and distribution, and they destroyed him.
Don’t build your value on a capability that can be purchased, subscribed to, or called via API. Being 10% better at prompting isn’t a moat. Being able to orchestrate autonomous agents isn’t a moat. These are Disdéri’s camera. Anyone can buy one.
A moat has to be something deeply bound to you as a specific person. Your signature (Collen). Your fluency across media (Degas). Your right to sign off, physician, attorney, certified auditor. Your client relationships, the ones that can’t be outsourced to a RAG system. None of these can be replaced by a prompt template.
As Someone Who Manages Workers
Build the Distribution Layer, Not the Tool
Nadar didn’t invent photography. He didn’t turn it into the biggest industrial enterprise of his era either. Disdéri did that. What Nadar did was turn his studio into the cultural living room of Paris.
He had started as a failed caricaturist and journalist. He didn’t learn photography until 1853. By the 1860s he was friends with Baudelaire. He flew hot-air balloons and became the most famous balloon photographer of the nineteenth century. He photographed the Paris catacombs using artificial light. He wrote books and novels. He lent his studio to the Impressionists for their show. The red facade with the 15-meter sign was a Paris landmark.
He wasn’t selling photographs. He was selling the identity of having been photographed by Nadar. From 1860 onward, most of the actual photography in his studio was done by his staff. Unless the sitter was particularly famous, Nadar didn’t bother showing up. He had become a brand under which his staff delivered.
In 1889, the year Disdéri died in the asylum, Nadar was still alive, living comfortably in his Paris apartment. He lived to 1910. Eighty-nine years old.
Don’t build an app on top of OpenAI. Build a distribution layer on top of your own customers. Model providers are commoditizing. Token costs will keep falling.
The long-term winners are platforms with customers, trust, and distribution. Not the companies with the strongest underlying technology. The App Store didn’t invent the iPhone. Salesforce doesn’t write AI models. WeChat didn’t invent conversational technology. All of them placed someone else’s technology inside their own distribution channel and collected rent on the channel.
Watch Out for the Bouguereau in Your Organization
Bouguereau was the biggest market winner of his era. He perfected Academic painting to its highest finish. The king appointed him. The Academie decorated him. His paintings were bought before he finished them.
He was buried for seventy years after his death.
He had perfected the accurate rendering of classical mythology at exactly the moment the photographic age was draining the market for that function. His success was his blind spot. He didn't know that what he was perfecting was a disappearing function.
The irony is that he used photography himself. Just to paint better. Not to rethink what painting was for in the photographic age. He used new tools to do old things, and the better he did them, the further out of step he became.
The most successful people of the AI Agent era are the most exposed. The most skilled prompt engineers. The designers most fluent in ChatGPT workflows. The executives who restructured their teams around AI earliest. Their KPIs this year look excellent.
The more they've invested in the AI Agent paradigm, the more exposed they are five years from now. When autonomous AI becomes the norm — and Anthropic's own Economic Index already shows enterprise API usage at 77% automation — their entire skill set, “human plus AI collaborate on tasks,” becomes redundant, because AI does the tasks alone.
As a manager, identify the people on your team whose stardom rests on the AI Agent paradigm. The more brilliant their current performance, the more they are perfecting a disappearing function. The point isn't to discard them. The point is not to bet the future of the organization on them.
Protect the Apprenticeship, Even When It Is Economically Irrational
The miniature trade of the early nineteenth century reproduced itself through apprenticeship. Joseph Wood apprenticed in John Wesley Jarvis's studio in 1803, opened his own studio in 1809, and later took on Nathaniel Rogers as his apprentice. Rogers took on others. That was how the profession reproduced itself, master to apprentice, generation after generation.
After 1839 no twenty-two-year-old wanted to apprentice as a miniaturist anymore. They opened photography studios instead.
This is how the miniature trade actually died. Not by being replaced. By no one new joining.
When Goodridge died in 1853, Boston still had a few miniature studios. But there were no new entrants. Within ten years those studios closed too. The intergenerational pipeline broke, and the profession went with it.
AI Agents make it possible to skip hiring junior employees, because a new hire plus AI is good enough. But no juniors hired means no mid-levels in five years means no seniors in ten. Yale's 2025 Big Freeze research has already documented this in motion. Firms aren't laying people off. They've just stopped hiring. They use AI to do what new hires would have done. Headcount looks stable. The pipeline is quietly breaking.
The decision a manager has to make explicitly is the economically irrational one. Keep hiring twenty-two-year-olds. Give them experience gradients. Even when AI can do their work more cheaply in the short term.
Education systems, medical systems, and the military all run on this principle. They accept short-term inefficiency because they have to protect intergenerational transfer. If every manager in an industry chooses the short-term optimum, the industry hemorrhages from the bottom. No single company can reverse the trend. But the trend is made of specific choices, and yours is one of them.
The Pose is Yours
Return to the Parisian of the 1860s.
He didn't know he was posing for the camera. So his body was silently, irreversibly reshaped by the new technology. Chin up, shoulders relaxed, gaze off-axis. He thought that was just how a person stood.
We are posing for AI too. We ask ChatGPT first. We write in AI-friendly structures. We optimize our resumes for the algorithm. Sparrow's experiment, Gerlich's correlation, the MIT EEG data, they all tell us this is happening below conscious awareness. The Parisian of 1860 didn't know. We can at least know that we're doing it.
This is what reading the script is actually worth. It can't stop the replacement. It can't turn you into Monet. It can't even guarantee you won't become Goodridge. But it can let you walk through the script in a conscious pose, instead of being shaped without noticing.
The script can't be rewritten. The pose is yours.