The Future's Losses Have Names. Its Possibilities Don't.

Since 2020, I've met monthly with a Chatham House–style group of CEOs — hospitality, entertainment, restaurants, advertising, technology; publicly traded, private equity–owned, venture-backed. Six years of monthly conversation about creativity, exploration, and lately, almost every time, AI.

Here's what surfaces once the trust is high enough. These leaders are driving AI adoption across their companies — encouraging their people to explore, to build skills — while hearing, underneath the enthusiasm, a quiet fear: will my job go away?

I hear the same fear at keynotes. From senior teams I advise. From college students asking me what to study.

And I've noticed a pattern in how the fear works: the jobs people can imagine are the jobs that are disappearing.

What I keep observing underneath the anxiety is an imagination problem. People can picture, in detail, the thing that's going away. They cannot picture the thing that's on the horizon. So I've started giving worried rooms a thought experiment.

Stand on the corner

It's 1900. You're on a Manhattan street corner.

More than a hundred thousand horses work this city. You can hear them, and believe me, you can smell them — they produce a couple million pounds of manure a day, and shoveling it is somebody's whole job. Teamsters hauling freight. Stable hands. Blacksmiths. Sweepers. (You may have heard the famous prediction that London would be buried under nine feet of the stuff by 1950. That quote is folklore — nobody's ever found the original. The manure, however, was extremely real.)

Around the horse, an entire economy: carriage builders, harness makers, wheelwrights, hay farmers, feed merchants, veterinarians, coachmen. Feeding the nation's working horses consumed, by some estimates, well over a hundred million acres of farmland.

Now here comes the future, sputtering past at eight miles an hour.

Notice what we call it: a horseless carriage.

We can only describe the future by subtracting from the present. A carriage, minus the horse. And we're doing it again — most of our AI vocabulary is some version of work, minus the human.

When I run this experiment live, I ask the room to list the jobs this machine threatens. That part takes seconds. You just met them — the teamster, the stable hand, the sweeper, the blacksmith. They have names. They have faces. They have payrolls you could count.

Then I ask for the jobs the machine will create.

That's where the room goes quiet.

Somebody usually offers mechanic. Fair — that one's visible from the corner, right next to the blacksmith.

But nobody says gas station attendant, because gas stations don't exist. Nobody says traffic engineer, or motel operator, or long-haul trucker, because highways don't exist. Nobody predicts the drive-in, the suburb, or the shopping center designed around a parking lot, because the world that makes those things sensible hasn't been built yet. Forget predicting that your great-grandchild will blend smoothies at a Jamba Juice in a strip mall — you'd have to explain the strip mall, the mall's parking lot, the commute, and possibly the smoothie.

The automobile didn't invent restaurants, by the way. Restaurants are ancient. What the automobile changed was the radius of commerce — and a changed radius meant roadside diners, destination dining, and eventually national chains built on the assumption that customers arrive on four wheels.

Standing on that corner in 1900, the losses are a census. The gains are a blank page.

We confuse "I can't picture it" with "it won't happen" ‍

This asymmetry has a mechanism, and it's one of my favorite subjects.

Kahneman and Tversky showed that we judge how likely something is partly by how easily we can picture it. (If you haven't read The Undoing Project, Michael Lewis's book about their partnership, stop reading me and go read that. I mean it. Two men who thought together for decades and produced ideas neither could have produced alone — there's a lesson in there beyond the psychology.)

A threatened job is easy to simulate. You can run the movie in your head — the pink slip, the empty stable, the teamster looking for work. Concrete things with names generate vivid mental footage.

A job that doesn't exist yet generates nothing. No footage available. And our minds file "I can't simulate it" under "it probably won't happen."

The thing that's going away is easy to imagine because it's known — definite, finite, tangible. The thing that's coming is hard to imagine because it's yet to be invented.

This summer I was telling the street-corner story to the leadership team of one of the world's great automakers — there was something delicious about pitching the horseless carriage to the people who make the carriages — and a line came out of my mouth that I hadn't planned:

The fact that we can't imagine it doesn't make it any less real.

Here's what the blank page eventually said. Between 1910 and 1950, the automobile displaced roughly 600,000 American jobs — carriage makers, harness makers, most of the horse economy. Real losses, real people. Over the same stretch, it created an estimated 7.5 million jobs, more than ten times what it destroyed, and only about a tenth of those were in car manufacturing itself. The rest emerged in the world the car made possible. By 1929, dealerships employed 330,000 people. Service stations, 127,000. Garages, another 104,000. A few of those jobs were visible from the street corner. The ecosystem was not.

Zoom out further and it gets wilder: economist David Autor and his colleagues combed through eight decades of census records and found that roughly 60% of the work Americans did in 2018 was in job categories that didn't exist in 1940.

A net number is not a person‍ ‍

I need to say this part plainly, because I refuse to sell you cheap optimism.

"7.5 million new jobs" was no comfort to a fifty-five-year-old harness maker in 1915. New work appearing somewhere in the economy does not automatically rescue a particular person, in a particular town, at a particular salary. The automobile transition was genuinely painful for real families, and honest people should say so. History gives us evidence that new work emerges. It gives us no promise about who gets it.

Hope is not a transition plan.

But look closer at who actually made it across, because the aggregate numbers hide the part I find most instructive.

The first auto mechanics weren't issued from some retraining program. They were blacksmiths, machinists, and bicycle repairmen who walked toward the strange new machine instead of away from it. By the 1920s, more than 60,000 auto service shops had opened in America — many founded by people whose original trade was dying underneath them. C.R. Patterson & Sons of Ohio spent decades building horse-drawn carriages; when the horseless kind showed up, they started repairing them, learned engines, and became automobile manufacturers.

Nobody handed those people a job title.

They noticed what had become newly possible, and they commissioned themselves.

The question underneath the question

So when the room goes quiet and someone asks what happens to the jobs, I've started offering a trade.

The question in the air — what job will AI leave me? — puts you on the 1900 street corner, counting losses, waiting to learn your fate. The question I'd rather hand you: what problem just became solvable, and what role might I name by solving it first?

I've written before that the bug is the brief — that in the age of AI, the premium moves upstream from problem solving to problem finding. This is that idea with a fifty-year time horizon attached. Nobody posted "gas station owner" in 1905. Somebody noticed cars kept running out of fuel, and named the work by doing it.

Want to try? Take a problem off your bug list — something at work that everyone complains about and nobody owns — and give your AI teammate this:

"Here's a problem I keep noticing that nobody owns: [describe it]. Given what AI now makes cheap, draft the description of a role that should exist to solve it. Then help me do the first week of that job — starting with a small proof I could show someone by Friday."

Then go do the first week of a job that has no name.

In 1900, "auto mechanic" wasn't a job title yet. Somebody had to walk toward the machine first.

The future's possibilities are only unnamed. Naming them is work a human still has to do.

Might as well be you.

Related: The Bug Is the Brief
Related: Be the Prompt
Related: The Room Is the Prompt

Join over 54,147 creators and leaders who read Methods of the Masters each week.