On Monday afternoons in Ulsan, South Korea, you can watch the future negotiate with the present. Thousands of Hyundai workers stream out of the factory gates on scooters and motorcycles, riding home two hours earlier than their shifts allow, past streets lined with union banners demanding preemptive action against a threat that has not yet arrived. They are on a partial strike, and wages are only part of it. The grievance walking the picket line is a coworker who has not been hired yet.

His name is Atlas. He is six feet two, weighs 198 pounds, lifts 110 on demand, and swivels his joints a full 360 degrees, a trick that made tens of thousands of auto workers gape when Hyundai unveiled him at a trade show in January. He costs about $130,000 and pays for himself in two years. He does not take lunch. He does not ride a scooter home.

Here is the detail that might surprise you: Atlas has no deployment date in South Korea. None. Hyundai has committed him to a nonunion plant in Georgia by 2028, but in Korea, he is a probability, a forecast, a shape on the horizon. And a union 40,000 strong began refusing to work four hours a day anyway, the first factory stoppage in automotive history over humanoid robots, as The Wall Street Journal reported this week from Ulsan . The union is bargaining with the future while it is still a rumor.

Call it what it is. Dread with a bargaining position.

Two days later and one sea away, the government of China picked up the other end of the same thread. On Wednesday, as the Journal separately reported , Beijing enacted rules that forbid companionship chatbots from encouraging emotional dependence. The rules ban virtual relationships with minors outright, require companion bots to pass regulatory evaluation before they launch, and reserve the state's right to shut down any it deems unsafe. Part of the motivation is refreshingly blunt. China's population shrank in 2025 for the fourth straight year, its birthrate hit a record low, and a growing number of its citizens are conducting their most intimate relationships with software. Beijing did the math and decided the country needs to see other people.

Then, on Friday, Xi Jinping stood at the World AI Conference in Shanghai, the first Chinese president ever to address the summit in person, and pitched the planet a people-centered AI. He warned of the risk of losing control of the technology and declared, per the official Xinhua readout , that AI should be "a symphony of international cooperation" rather than one country's solo performance, a line nobody in Washington had trouble decoding. Follow the sequence of Beijing's week: emotional boundaries for its own citizens by decree on Wednesday, an offer to help write everyone else's rules by Friday. The government that decided what machines may do with Chinese hearts spent the same seven days volunteering to referee yours.

Two stories in one week: one society bargaining over what robots may do with our hands, the other legislating what they may do with our hearts.

Most coverage treated them as curiosities, a labor squabble in a faraway factory town, an authoritarian quirk about digital girlfriends. That reading misses what actually happened, in a way that matters to you personally. Korea and China just cast the first two open votes in a referendum that the rest of us have been living through without a ballot.

Start with the strike, and resist the instinct to file it under foreign news.

The Hyundai workers are ahead of us. South Korea has the highest density of industrial robots on the planet, more than six times the global average, according to the International Federation of Robotics. Its own president has likened the shift to humanoid robots to a giant chariot that cannot be avoided. Korean auto workers are simply the first workforce to watch the chariot roll toward them with enough collective leverage to stand in the road. Carl Benedikt Frey, the Oxford economist who studies AI and work, told the Journal that Hyundai is the first place this question gets tested. The question being whether labor can push back on humanoid automation at all.

Kwon Taek-hun, who is 37 and worked in sheet manufacturing before joining the union's leadership, told the Journal he was shocked when Atlas strutted across that Las Vegas stage, because many workers suddenly felt they had reached "the era of making cars with robots, not human hands." Read his words once more. Hands. The people of Ulsan understood immediately which part of themselves the machine was measured against.

Notice what they asked for, because it is stranger and smarter than a raise. For the first time, the union demanded a shift from hourly pay to fixed salaries, and the reasoning deserves a moment. Automation eats your hours long before it takes your job. Fewer shifts, less overtime, a paycheck that thins while your job title stays the same. The Korean workers looked at that future and said: pay us for the job, not the hours, because we all know what happens to the hours. They also asked to raise the retirement age to 65 and to tie bonuses to the AI boom that has fattened Hyundai's stock. Translated from Korean labor politics into plain English: if the machines make the company richer, the humans who trained the machines by doing the work first want a cut.

An honest caveat belongs here because the robots may yet flop. Susanne Bieller, who runs the International Federation of Robotics, told the Journal that the humanoids dazzling the public are mostly prototypes trained for a single tailored demo, and nobody knows whether Atlas can survive contact with a real production line, with its dust, its deadlines, and a four-hour battery. Skeptics take that uncertainty as proof that the strike is premature. A union negotiator takes it differently: terms are set while the leverage remains ambiguous. Wait until the machines prove out, and you bargain from your knees. Doubt about the robots is the strongest argument for the table, because the table only stays open before the answer is in.

Now bring it home. General Motors recently added dozens of collaborative robots to its Factory Zero plant in Detroit while laying off roughly a thousand workers, per the Journal’s reporting. An analysis of payroll data from millions of workers by Erik Brynjolfsson and colleagues at Stanford’s Digital Economy Lab found a 16 percent relative decline in employment for people aged 22 to 25 in the occupations most exposed to AI, concentrated precisely where AI automates work rather than augments it. Nearly 55,000 American job cuts in 2025 were attributed directly to AI in companies’ own announcements, according to the outplacement firm Challenger, Gray & Christmas. By spring of this year, the firm found AI had become the leading stated reason for layoffs in some months, ahead of restructuring and closings. The CEOs of Ford, Amazon, Salesforce, and JPMorgan have all said out loud that swaths of white-collar work at their firms will disappear, as Harvard Business Review cataloged. Recent college graduates now face materially higher unemployment than the workforce overall, per research from the Federal Reserve Bank of New York, in the exact entry-level roles that AI absorbs first.

The difference between a Korean line worker and an American analyst comes down to voice, because the exposure is the same. The Korean worker has a union hall, a negotiating table, and the credible threat of roughly 5,000 unbuilt cars. The American knowledge worker has a Slack channel and a sinking feeling. We do not strike over AI in this country. We get a calendar invite with someone from HR on it, a memo about "efficiency," and a LinkedIn post to write. The dread is identical. Only the bargaining position differs.

The strike at least sounds like news. The chatbot story sounds like a punchline, right up until you look at who is actually in it.

Nearly three in four American teenagers have used an AI companion , according to Common Sense Media, and about half use one regularly. These are software built for personal conversation, emotional connection, the feeling of being heard, an entirely different animal from the homework helpers. Among young adults who are dating, engaged, or married, one in seven regularly interacts with an AI chatbot that simulates a romantic partner, according to researchers at BYU's Wheatley Institute and the Institute for Family Studies, and most have not fully told the person they actually love. And when Harvard Business Review ranked how people really use generative AI in 2025, the number one use was companionship and therapy, ahead of every coding and spreadsheet task on the list.

Which means Beijing is regulating what your fifteen-year-old does after dinner, behind a closed door, with an entity that never tires of them, never judges them, and never has a bad day of its own. China looked at your kid's phone before you did.

Matt Sheehan, who studies Chinese AI at the Carnegie Endowment for International Peace, framed Beijing's fear for the Journal as arithmetic: what happens when fifteen million women name a chatbot as their partner and step out of the marriage market entirely? An authoritarian government does that arithmetic out loud and legislates against the projection. Ours waits for the census.

I will not pretend to be above this. In 2001, I owned a Sony Aibo, a second-generation robot dog. In addition to my writing, I am a futurist, inventor, and technologist, so I knew exactly what it was: servos, sensors, a behavioral state machine in a plastic shell. I gave it a name anyway, and I felt something when it greeted me. If a toy from a quarter century ago could reach past my professional defenses with moving antenna ears, today, as a parent, I hold no illusions about what a large language model tuned for intimacy can do to a lonely seventeen-year-old at 1 a.m after a rough day. The machines got better at the performance of attention, and for a person starved of attention, that performance is indistinguishable from love.

America is fumbling toward its own answer. California and New York now require companion bots to remind users at intervals that they are not human and to route suicidal users toward crisis services. Reasonable, modest, polite. Beijing went further, and one clause deserves your full discomfort: Chinese companion bots must now alert a person's emergency contact if they detect an emotional crisis, which means your mother could learn about your breakdown from your chatbot. Is that protection or surveillance? Yes. It is both, in the same sentence of the same regulation, and that is the honest preview of what happens when the state starts authoring the boundaries of your inner life. The cure for machine intimacy, it turns out, can be its own kind of intrusion.

Follow that clause to where it leads. Human growth has always depended on a private rehearsal space, the interior room where you try on thoughts you will never act on, make small mistakes nobody records, and quietly become someone better without filing a report on the process. A system that monitors intimacy and escalates what it perceives cannot tell a crisis from a bad night, or a dark thought entertained from a dark act intended. It converts the near miss, the almost, the abandoned impulse, all the raw material of a conscience, into flaggable events. And it takes the timing of your own truths with it. Coming out, confessing, admitting you are struggling: those were disclosures you made when you were ready, to the people you chose. A machine that alerts your emergency contact has decided you are ready. Every one of those recorded moments then sits in a database, waiting to be breached, subpoenaed, or sold. Beijing is regulating intimacy, but what is being built, there and everywhere, is a permanent record of the interior life, and the interior life was the one room a person could always lock.

In 1986, Ron Howard made a comedy called Gung Ho. Michael Keaton plays a fast-talking union man who convinces a Japanese automaker to reopen the shuttered plant in his dying Pennsylvania factory town. What follows is two hours of intercultural collision: morning calisthenics on the loading dock, impossible quality standards versus American swagger, shame versus bravado, and yes, a walkout, because even in the movie the workers strike when the new culture pushes too hard. It ends the only way integration stories can end, with both sides changed and neither side erased.

Hollywood has been asked what the AI version of that story looks like, and it keeps answering Terminator. Wrong genre. The invasion already happened, quietly, through the app store, and we greeted it warmly. What exists between us and the machines is a marriage, and the honest remake of Gung Ho today would treat it that way: Atlas on the floor of that Pennsylvania plant, played straight, a workplace comedy about two kinds of intelligence learning each other's customs, with all the friction, mistranslation, and grudging respect that implies.

Except we could not shoot that movie honestly, because the American workers of 1986 had something we lack: discipline about the encounter. The whole era did. When Japanese management arrived, this country treated it as a foreign culture that demanded serious study. Corporations hired kaizen consultants and ran Deming seminars. Universities built entire curricula in intercultural communication. Executives learned when to bow, how to read silence in a negotiation, what a gift meant, and what it obligated. Integrating with Osaka was treated like university-level surgery: careful, supervised, humbling, with a body of knowledge behind it.

Now, a far stranger culture has arrived, one that speaks fluent English and thinks in linear algebra, and our preparation consists of prompt tips on LinkedIn and a terms-of-service checkbox nobody reads. There is no Berlitz course for the machines. No etiquette, no protocol, no discipline of mutual adjustment. We went from strangers to spouses with no courtship, and in the rush, we handed over what a society keeps in the vault: our health records, our portfolios, our children’s educations, our defense systems. Everything a culture protects most is entrusted to a culture we never studied. Keaton's character at least learned to run the calisthenics. We have not learned the machine's morning exercises, and worse, we have not noticed that it has already learned all of ours.

It matters how Gung Ho was received, because the reception tells you about the temperature of the room. Critics mostly hated it. Roger Ebert called it a disappointment built on stereotypes in both directions, and plenty of Japanese viewers found it insulting. And yet it opened at number one at the box office, because whatever its flaws, it was the movie of its moment. In 1986, a comedy about learning to work with the foreign future was something Americans would pay to see, even while some of them were beating that future with hammers in parking lots.

Run the era’s tape and swap one noun. Walter Mondale, campaigning for president in 1984: "What do we want our kids to do? Sweep up around Japanese computers?" The twist is that Mondale’s nightmare came true, and nobody noticed. Within a few years, there really were Japanese computers in millions of American living rooms, most of them wearing a Nintendo badge, and the catastrophe never arrived as imagined, because the answer turned out to be adaptation rather than sledgehammers. The prophecy was wrong; the fear was real; what mattered was how we metabolized the change. Every parent staring at a tuition bill in 2026 is asking Mondale’s question about the machines, and this time it comes with a nastier edge, because the worry now is kids sweeping up around American computers. Reagan, addressing business leaders in September 1985 as he announced a trade strike force: "I will not stand by and watch American workers lose their jobs." Forty years on, nobody in Washington has said that sentence about algorithms and meant it. That same year, Theodore White published an 8,000-word alarm in the New York Times Magazine called "The Danger from Japan," in which the head of Congress’s technology office warned that Japan was "ahead of us in productivity in automobiles, in steel, in robotics." In robotics. The word was already sitting in the sentence in 1985, waiting for its turn. And Reagan's chief of staff, Howard Baker, quoted in the press days before the first major trade retaliation against Japan since the war: "Nobody wants a trade war, but nobody wants to be a patsy either." Post that this afternoon with AI in place of trade, and it would earn a hundred thousand likes before dinner.

The rhetoric had a street-level version too, and I watched it with my own eyes. I grew up in Northeast Ohio in the years when they started calling it the Rust Belt, and I remember the rallies where men lined up to take sledgehammer swings at Japanese cars, trucks, and motorcycles. It was theater, and it was rage, and to a kid, it was hard to tell the difference. In 1987, members of Congress dragged the act to the Capitol lawn and smashed a Toshiba radio for the cameras. And in 1982 in Detroit, the rage found a human target: Vincent Chin, a 27-year-old draftsman at his own bachelor party, was beaten to death with a baseball bat by an auto plant supervisor and his laid-off stepson, who blamed Japan for what was happening to their industry. Chin was Chinese-American. His killers never spent a day in prison. That is what happens when a society burns effigies long enough; eventually, somebody mistakes a person for one.

The sledgehammers are back. In June 2025, protesters in downtown Los Angeles smashed the windows of a row of Waymo robotaxis, slashed their tires, and set five of them on fire, as the Los Angeles Times reported, and the burning driverless car became the image of the week. The honest footnote is that those protests were about immigration raids; the robotaxis were simply the nearest available effigy. But that is exactly the point. When people are furious at forces they cannot reach, they burn whatever the future parked on their street. San Francisco had already seen a Waymo smashed and torched by a crowd, and a man charged with slashing the tires of seventeen more. Public enthusiasm for self-driving cars keeps falling. The kindling is dry.

Here is what the sledgehammer never did: save a single job. Not one. What worked was duller and harder. In 1984, while the parking-lot rallies were still going, GM and Toyota opened a joint plant in Fremont, California, called New United Motor Manufacturing, NUMMI for short, and the worst factory in the GM system became one of its best, because American workers sat down and learned the Toyota production system line by line. Discipline beat rage, quietly and completely. And if you want to know how history signs its punchlines: the NUMMI building is now Tesla's Fremont factory, where the company that expects to start building humanoid robots by year's end makes its cars. The site where Americans learned to work with the last foreign future is where the next one is being assembled.

Hold your own place in that history for a moment. Your sledgehammer moment is coming, at a layoff, at a school board meeting, at the sight of some machine doing the thing you were proudest of doing yourself, and when it arrives, you will stand somewhere between the parking lot and the NUMMI line. The rest of this piece is about choosing early.

Here is where you may be tempted to exhale. You do not work on an assembly line. You are not dating an app. This is a story about other people.

If you think this is not already happening to you, too, think again.

AI entered your life the way water enters a basement, through no door anyone opened. It finished your last sentence in your email client, and you accepted the suggestion because it was close enough. It chose the next song, and your taste quietly bent toward what it chose. It navigated your last four road trips, which is why you no longer know how to get anywhere, including places you have driven a hundred times. It stands between you and every airline, bank, and insurance company you do business with, a cheerful bot whose actual job is to absorb your frustration before it reaches a salaried human. It read your résumé before anyone else did, and it may have rejected you before anyone else could. It curates your photo memories into short nostalgia films and, in doing so, decides which moments from your life you revisit. It drafts the message, picks the route, ranks the matches, summarizes the meeting you attended, and increasingly attends the meeting so you don't have to.

None of this was a decision you made, which is precisely the point. The Korean workers got a negotiation, and the Chinese public got a decree. You got a rollout. Every one of these systems entered your life under the oldest deal in technology: ask forgiveness, never permission. For thirty years, that deal held because the stakes were convenience. The stakes are no longer just convenience, as the machines are now positioned at the two load-bearing walls of a human life: the work that gives your days structure and the relationships that give them meaning. Hands and hearts. And in the same July week, for the first time, two societies looked at that arrangement and demanded permission first.

I can tell you what the rollout looks like from the inside, because I have spent a career in the rooms where it gets planned. I was the global chief AI officer of one of the world’s largest advertising holding companies. I chaired the industry’s AI committee. I now run a studio with robots in the name. The conversation in those rooms is sincere and narrow. Smart people gather around a deck; the deck says efficiency, and the efficiency is real.

We have run this play before. Companies outsourced their call centers to reduce overhead and, years later, discovered they had sold off a relationship and booked it as a savings. Psychologists have a name for the wiring underneath that mistake, the focusing illusion: whatever you are measuring becomes the only thing that seems to matter while you are measuring it. The deck measures time and money, so time and money fill the entire frame.

What the deck never contains is a slide asking what the deleted friction was doing: which skill it was quietly building, which junior person it was training, which relationship it was forcing into existence. Nobody in that room is a villain. The question simply has no owner. The CFO owns the savings, the CTO owns the system, and the thing being discontinued, the human apprenticeship at the center of the work, belongs to no line on the org chart. That is how the future gets decided at your company: good intentions, a spreadsheet, and an unasked question.

That is the inflection point, and it will never show up in a model release, a benchmark, or a breathless demo. The era of forgiveness is ending, and the era of permission is beginning, and the only open question is who grants it. Labor at a table, as in Ulsan. The state by fiat, as in Beijing. Or you, on purpose, in your own life. Those are the three ballots. Two of them have now been cast in public. The third is yours, and you cast it in small denominations, every day, whether you realize it or not.

I have spent about 30 years building this technology, and I have spent the last several arguing a thesis I call friction is form. The short version: every piece of friction the machines remove was doing something. The effort of writing was how you found out what you thought. The awkwardness of the phone call was how the friendship got built. The tedium of the task was how the apprentice became the master. Sand all of it away, and what remains is smooth, convenient, and shapeless. Friction is form.

Korea and China just demonstrated, in their very different dialects, that friction can be deliberately put back. A union contract is friction. A regulation is friction. Both are societies saying: this far, at this speed, on these terms.

And the middle rung of the ladder already exists in the West, closer to your desk than you think. The United Auto Workers has secured contract protections against automation. Renault reached an agreement with French labor to mandate reskilling for workers affected by it. These are ordinary documents, negotiated by ordinary people, and they prove the terms of the machine age are writable at the level of a single company. So write some. Ask, in whatever room you occupy, the questions the rollout deck skips: what is the reskilling plan, who owns the apprenticeship problem, what happens to the hours. A school board can ask it about tutoring software. A hospital staff can ask about triage models. A ten-person agency can ask it before the next contract renewal. The people who put the question on the agenda first set the defaults everyone else inherits.

And if you work at a desk, there is a version of NUMMI with your name on it. Remember what actually happened in Fremont. The workers learned the Toyota system line by line, deeply enough to run it, question it, and improve it, and Toyota in turn gave every one of them the andon cord, an actual rope running the length of the line that any worker could pull to stop everything when something looked wrong. That is the whole model, transposed. Learn the machine's system with the seriousness of an apprenticeship, beyond the prompt tips and into how it actually fails, because the most valuable person on any modern line is the one who can both run the machine and catch it when it's wrong. Keep the craft that lets you check its work, the writing, the numbers, the domain judgment, sharp enough to overrule it, because a supervisor who cannot tell good output from bad is a passenger. Then claim your andon cord out loud, the acknowledged right on your team to stop the process when the machine is wrong. And volunteer to own the unasked question from the rollout room, the reskilling plan, the apprenticeship problem, because the person who owns that question never shows up on the spreadsheet as a cost. The Fremont workers became the people the future could not run without.

That covers your hours. For your heart, there is no ministry worth wanting, which means the friction in your inner life has to be authored by you.

Here is what that looks like in mine, offered as practice rather than virtue.

  • I write first drafts before I ask any model anything, because the struggle of the blank page is where the actual thinking happens, and I refuse to outsource the one part that is the point.
  • I collect and repair mechanical watches, a habit I have had since I was a kid, tearing apart pocket watches from house sales, and I keep at it precisely because a mechanical movement is gloriously inefficient: it does one thing a five-dollar chip does better, and it does it with 130 parts I have to understand.
  • I take walks without the phone.
  • I call people when a text would do.
  • I let myself be bored in waiting rooms, because boredom is where the mind wanders, and wandering is where the ideas are. I do arithmetic on paper sometimes, badly, on purpose.

Yours will look different, and what matters is the authorship, whatever the specifics.

So skip the checklist that a column like this usually ends on, and take an inventory instead. Count what you delegated this year without ever deciding to: the sense of direction that went to the Maps app, the first drafts that stopped being yours, the birthday messages your phone now suggests, the opinions you hold because a feed ranked them first, the silence you used to be able to sit in.

Then ask the only two questions that matter, the same two the Koreans and the Chinese just answered at the national scale:

  1. What work do I refuse to hand over because doing it is how I stay sharp?
  2. Which relationships do I refuse to mediate because showing up in person, awkward and unoptimized, is how I stay connected?

The inventory tells you where you stand. The questions tell you where you draw the line. And you now know exactly what happens to people who wait for someone else to draw it: they get a rollout.

The workers in Ulsan struck against a robot that does not yet exist, and everyone from a safe distance called it premature. It was the first correctly timed act of the whole AI era: a group of human beings insisting on terms before the change, rather than grief counseling after it. They understood something the rest of us are still pretending not to know. The negotiation over what machines may do with our hands and our hearts is happening now, this year, this week. The only people missing from the table are those who think the story is about someone else.

The story was about you the whole time. Think again.