Opening: Two Mornings
Chicago, a Tuesday in January 2025
Sarah Kowalski woke to the sound of her alarm.
6:47 a.m. She no longer needed to get up at six. There was nowhere to go. She pulled a levothyroxine pill from the nightstand drawer and swallowed it with water. Hypothyroidism. Without the medication, chronic fatigue, weight gain, and depressive symptoms would set in. The drug cost $180 a month without insurance.
A one-bedroom apartment near Lincoln Park, Chicago. Rent: $1,700 a month. Through the window, Lake Michigan in January. Gray lake, gray sky.
She brewed coffee and opened the mail. Three envelopes.
First, a COBRA health insurance notice. $1,200 a month. The national average for individual COBRA coverage is $584 per month (KFF 2024), but a PPO plan for chronic disease management costs more. Switching to a cheaper plan would mean losing her primary care physician. Second, a federal student loan repayment notice. $541 a month. Balance: $47,000. Third, a 401(k) quarterly statement. Balance: $28,000. Early withdrawal would mean a 10 percent penalty plus income tax. She could not touch it. Not yet.
Two weeks earlier, the midsize Chicago law firm where she had worked for twelve years let her go.
The litigation support team shrank from twelve to five. Harvey AI had taken over case law research; Relativity was handling document review for discovery. Sarah's annual salary: $62,000. A Harvey AI subscription: $3,000 to $6,000 per year. For the firm, the math was simple.
On the day of the layoff, the senior partner told Sarah she had been an excellent paralegal. This was not a performance issue, he said.
Not a performance issue. That was what made it crueler.
Shenzhen, the Same Day, 8:00 a.m.
Wang Lei sent his child off to school and sat down on the living room sofa.
A 70-square-meter apartment in Longgang District. He had bought it in 2020 for 4.55 million yuan. Both sets of parents pooled together a down payment of 1.37 million yuan. The remaining 3.18 million yuan was a thirty-year mortgage. Monthly payment: 16,900 yuan.
He opened the real estate app Anjuke on his smartphone. Listings for the same complex, same floor plan. 3.5 million yuan. What he had bought for 4.55 million five years ago was now worth 3.5 million. Outstanding mortgage: 3.18 million yuan. Subtract the loan from the market value and 320,000 yuan remains. But selling would cost close to 300,000 yuan in agent fees and taxes. In practice, he would walk away with almost nothing. He was effectively underwater.
The day before, the cross-border e-commerce company where he had worked for ten years delivered the news.
His position as CS manager was eliminated. The thirty-person customer service team had been cut to six. Sobot's AI customer service system now handled 95 percent of inquiries. The remaining six agents dealt with complex claims that the AI could not resolve. When there is no team to manage, there is no need for a manager. The company offered N+1 severance of 132,000 yuan. The price tag for ten years of service.
132,000 yuan. Divided by the monthly mortgage payment of 16,900 yuan, that was less than eight months.
Wang Lei closed the real estate app. He opened WeChat. A message from Zhang Hao, a former colleague, was waiting. A registration link for Meituan delivery riders. A single line appended:
"Former manager, future rider."
Wang Lei read the message and turned off the screen.
12,000 Kilometers
The distance between Chicago and Shenzhen is 12,000 kilometers.
Sarah Kowalski and Wang Lei do not know of each other's existence. They do not know that on the same morning, they received the same kind of notice. The force that pushed them out has the same name: artificial intelligence.
The same technology is pushing two lives in the same direction. But the shape of the fall is entirely different.
In America, the drop is fast. A six-month cliff awaits. In China, the descent is slow. A two-year slide toward the bottom. The speed differs, but the direction is the same. Down.
Section A: Two Worlds
A-1. Sarah's World — The Fifteen-Minute Demo
An afternoon in March 2024.
Senior partner Mark Hollander gathered the team in the conference room. The Harvey AI interface was on his laptop screen. Founded by former OpenAI researchers in 2022, the company had raised $100 million in a Series C round at a $1.5 billion valuation. Sequoia Capital led the investment. After Allen & Overy became the first major law firm to adopt it, the rest of the industry began to move.
Hollander typed a query into the search bar — a request to find Illinois commercial litigation case law on confidentiality clause violations. For Sarah, this task would take six hours. Log into Westlaw, construct keyword combinations, read through relevant cases one by one, identify the applicable ones, and write summaries. Work she had repeated thousands of times over twelve years.
Harvey AI delivered the results in fifteen minutes.
Twelve cases, a relevance analysis for each, a summary of applicable legal principles, and strategic implications for both plaintiff and defense. A full day's work for Sarah, done in fifteen minutes. The cost was $1 to $5 per hour. Sarah's fully loaded hourly rate: $50 to $75.
Sarah remembered thinking: Impressive. Using this tool, I can focus on more complex analysis.
She did not know she was looking at her own termination letter.
One number circulated through the legal industry in 2024: 69 percent. That was the share of paralegal work hours that could be automated by AI, according to the 2024 Legal Trends Report. Seventy-seven percent of legal professionals said AI would have a "significant impact" on their work, a Thomson Reuters figure from April 2024. But "significant impact" was ambiguous. Assist or replace. The distance between those two words was exactly one termination letter.
The United States employs 345,800 paralegals. Median annual salary: $61,010. BLS data as of May 2024. Women make up 85 to 87 percent of the workforce. Many enter the profession through two-year community college programs. Some 40 percent of paralegals hold an associate's degree.
Sarah was the archetype of that pathway. South Side Chicago. Polish-American family. She completed her studies at Harold Washington College, earned an ABA-approved paralegal certificate in 2013. Two years of education, twelve years at this firm. That expertise was being replaced by software costing $250 a month.
A-2. Wang Lei's World — The Implementer Gets Implemented
Wang Lei first encountered AI in early 2023.
He was the one who introduced it.
The company demanded cost reductions. Wang Lei surveyed the market and selected Sobot's AI customer service solution. The basic package cost 2,000 yuan per month. Compared to the monthly labor cost of 240,000 yuan for a thirty-person CS team, the AI system cost 10,000 to 30,000 yuan. An eight-to-seventeen-fold difference.
Wang Lei explained it to his team. The system would handle simple, repetitive inquiries, freeing them to focus on higher-value work. Whether he genuinely believed this or was merely echoing management's language is no longer distinguishable.
The AI customer service transformation in Chinese e-commerce was already a massive wave. Taobao's AI system Ali Xiaomi had been operational since 2017. As of 2024, AI handled 97 percent of first-contact customer inquiries on Taobao and Tmall — a figure from Alibaba's investor report. JD.com's JIMI processed over 90 percent of inquiries automatically. Pinduoduo went further still, designing its system to make it deliberately difficult to reach a human agent. Pinduoduo's headcount is roughly 13,000 — one-seventeenth of Alibaba's. Radical efficiency.
In mid-2023, the integration of Tongyi Qianwen changed the equation. A large language model replaced the legacy rule-based chatbot. What had previously handled only simple FAQs now managed complex claims, emotional complaints, and multi-turn conversations. The scope of work that "required a human" narrowed sharply.
Wang Lei's team went from thirty to twelve by the end of 2023. From twelve to six in the second half of 2024. In January 2025, the manager position was eliminated. When AI replaces most of the agents, there is no one left to "manage." The manager loses the object of management. This was a new phase: AI dismantling not just frontline labor but supervisory labor as well.
The irony was perfect. The man who introduced the AI was displaced by the AI.
Wang Lei is from Changsha, Hunan Province. A second-tier city. He studied e-commerce at a vocational college and graduated in 2008. He moved to Shenzhen in 2012. Over ten years in customer service, he was promoted to manager in 2017. A manager overseeing a team of thirty. That was Wang Lei's identity.
A CS manager's salary in Shenzhen runs 10,000 to 12,000 yuan per month (BOSS Zhipin 2024 data). As a ten-year veteran manager, Wang Lei earned 12,000 yuan. Converted to an hourly rate, 55 yuan. The AI system's hourly cost is in the single digits.
Against that gap, management's decision was already made.
Section B: Anatomy of the Fall
B-1. Sarah's Fall — Three Clocks
After the layoff, Sarah's finances looked like this.
Income: $2,537 per month. Illinois unemployment insurance pays a maximum of $586 per week; converted to a monthly figure, that is the number.
Expenses: COBRA health insurance, $1,200. Student loan, $541. Rent, $1,700. Living costs (food, transportation, utilities, phone), $450. Total: $3,891.
Monthly deficit: $1,354.
Savings balance: $9,200. At this rate, the money runs out in seven months.
Three clocks are ticking simultaneously. Clock one: unemployment benefits, twenty-six weeks. Illinois pays a maximum of twenty-six weeks. Six and a half months. Clock two: savings depletion, seven months. Almost exactly overlapping with the end of benefits. Clock three: COBRA, eighteen months. She can maintain her employer health plan for up to eighteen months after termination, but the full cost falls on her, including the employer's share.
The point where the three clocks converge is the cliff.
Six to seven months out, unemployment benefits end, savings are gone, and COBRA still has eleven to twelve months remaining, but there is no money to pay for it. If she drops COBRA, she must buy thyroid medication without insurance. Switching to the ACA marketplace lowers premiums but changes her physician and narrows coverage.
Losing a job in America does not mean losing only income. It means losing health insurance. Among OECD nations, the United States is the only country where employment and health coverage are this tightly fused.
Sarah searched Indeed for paralegal openings. Within fifty miles of Chicago.
Two years ago, a single page would have returned more than twenty listings. Now there were seven. According to Indeed Hiring Lab analysis, paralegal job postings fell 18 percent between 2022 and 2024. Simultaneously, postings for "Legal Technology Specialist" rose 156 percent.
Three of the seven listings contained this line:
"AI tool proficiency required: Harvey AI, CoCounsel, or equivalent experience."
A structure that fires people because of AI, then demands AI proficiency from the fired. The word "irony" comes to mind, but for Sarah, it is not irony. It is reality. According to Robert Half's 2024 legal hiring guide, the share of job postings requiring "AI/legal technology proficiency" surged from 12 percent in 2023 to 34 percent in 2024.
That afternoon, Sarah searched a community college website for a "Legal Technology Specialist" certificate program. Twelve months. Cost: $12,000.
Twelve thousand dollars. With $47,000 in student loans still outstanding. Another $12,000 to borrow on top of that. Twelve months with no income. And even with the certificate, reemployment is not guaranteed. Age: forty-two.
According to AARP's 2023 survey, 78 percent of job seekers over forty reported experiencing or witnessing age discrimination. The Age Discrimination in Employment Act protects workers over forty, but proving a claim is difficult and litigation costs are prohibitive. Sullivan and von Wachter's 2009 study (American Economic Review) showed that displaced high-earning workers who found new jobs still experienced a long-term income decline of 15 to 20 percent on average.
Sarah closed her laptop.
Through the window, Lake Michigan. In January, the lake does not freeze completely. The surface partially ices over while water flows beneath. You cannot walk on the ice, and you cannot jump into the water. Sarah's situation was the same. Not fully collapsed, not fine either. Standing on a lake with a cracking surface.
B-2. Wang Lei's Stagnation — The Invisible Expiration Date
Wang Lei's financial situation involves larger numbers than Sarah's.
Income: 2,360 yuan per month. Shenzhen's unemployment insurance benefit — roughly 90 percent of the city's minimum wage. Even this is uncertain. Of China's 240 million unemployment insurance enrollees, only 2.3 million receive benefits, a take-up rate of 1 percent. The figure comes from the Ministry of Human Resources and Social Security's 2023 statistical bulletin. Complex application procedures, the difficulty of proving involuntary separation, and benefit levels too low to justify the effort all suppress claims.
Expenses: mortgage, 16,900 yuan. Living costs (food, utilities, transportation), 2,000 yuan. Total: 18,900 yuan.
Monthly deficit: 16,540 yuan.
Even adding his wife's retail income of 8,000 yuan, the shortfall is still 8,540 yuan.
N+1 severance: 132,000 yuan. Divided by a deficit of 8,540 yuan, that is fifteen months. Factor in 2,000 yuan per month for the child's education, and the runway drops to about twelve months. One year.
He must find a new job within one year. But Wang Lei is thirty-eight.
The age-35 crisis. In China's labor market, hiring opportunities drop sharply after thirty-five. Legally it is prohibited; in practice, it is industry standard. The age cap for the National Civil Service Examination is thirty-five. Some local governments raised the limit to forty starting in 2024, but most private enterprises still enforce the thirty-five threshold.
Wang Lei searched BOSS Zhipin for "customer service management" positions. The default filter was set to "under 35." When he entered his actual age of thirty-eight, more than half the results vanished. Changing the filter to "under 40" reduced the results even further — meaning even companies willing to consider candidates under forty were rare.
BOSS Zhipin's 2024 data analysis shows that over 80 percent of internet and IT job postings apply explicit or implicit age restrictions. Average employee age at ByteDance: twenty-seven. At Tencent: twenty-nine. At Pinduoduo: twenty-six to twenty-seven. A thirty-eight-year-old applicant is unlikely to receive even an interview invitation.
Zhaopin's 2024 report: the probability that a job seeker over thirty-five receives an interview invitation is 60 percent lower than for those aged twenty-five to thirty. Ministry of Human Resources and Social Security 2024 survey: the six-month reemployment rate for displaced workers aged thirty-five to forty-five is 42 percent, compared with 68 percent for those aged twenty-five to thirty-four.
For Wang Lei, the age-35 crisis had already begun three years before. Even before the AI layoff, his career had an invisible expiration date stamped on it. AI merely moved the date forward.
And then there is the apartment.
In 2020, Wang Lei believed he had put down roots in Shenzhen. In 2019, he had obtained a Shenzhen hukou through the points-based settlement system. Ten years of social insurance contributions, a stable address, a vocational college degree — all these scores added up to barely pass the threshold. The Shenzhen hukou is directly tied to a child's eligibility for public school, the scope of medical insurance coverage, and the right to purchase property. On the day the hukou was approved, he and his wife went out for a celebratory dinner. Now we belong to this city.
A year later they bought the apartment. Shenzhen housing prices peaked in 2021 at 87,000 yuan per square meter, the highest in the country. Longgang District, being on the outskirts, was 65,000 yuan per square meter. Seventy square meters. 4.55 million yuan. Still more than fifteen times the couple's combined annual income.
By the end of 2024, average Shenzhen housing prices had fallen 30 to 35 percent from the peak (China Index Academy). Longgang District fell further. Under 50,000 yuan per square meter. Wang Lei's apartment was now worth approximately 3.5 million yuan.
Underwater. The market value of the home below the outstanding mortgage balance. In Wang Lei's case: mortgage balance 3.18 million yuan, market value 3.5 million yuan. Sell and, after agent fees and taxes, he walks away with virtually nothing. Yet the loan obligation of 16,900 yuan per month for thirty years remains. He cannot sell, and he cannot hold on.
Consumption among underwater households contracts sharply. Even when income rises, spending fails to keep pace — household income grew 5.1 percent in 2025 while consumption growth reached only 4.6 percent (NBS). Every yuan locked into the mortgage is a yuan that never reaches the shops below. When housing prices fall, spending contracts; when spending contracts, surrounding businesses collapse; when businesses collapse, jobs disappear; when jobs disappear, housing prices fall further. A chain reaction.
Wang Lei's unemployment benefit of 2,360 yuan is just 14 percent of his 16,900-yuan mortgage payment. A structure in which survival is mathematically impossible.
So what are the options?
First, become a delivery rider. A Meituan rider in Shenzhen, working ten to twelve hours a day, six to seven days a week, can earn 6,000 to 10,000 yuan per month. Per-delivery fee: 4 to 8 yuan. Thirty to fifty deliveries a day. No social insurance. No workers' compensation. If a traffic accident happens, the rider bears the cost alone.
But even a rider's income of 10,000 yuan cannot cover the 16,900-yuan mortgage. Add his wife's income of 8,000 yuan and the total reaches 18,000, just enough to cover the mortgage and minimal living expenses. The child's education budget vanishes. Savings are impossible.
Second, return to Changsha. But that means giving up the Shenzhen hukou. The child's school would have to change. And even if he finds a CS manager position in Changsha, the salary would be 5,000 to 8,000 yuan per month — nowhere near enough to service the Shenzhen mortgage. Sell the apartment and the loss is locked in. Keep it and the deficit compounds every month.
A dead end. Wang Lei scrolls through WeChat Moments. Updates from former colleagues. One has started driving for Didi. One is trying live-stream commerce. One has posted nothing at all. No news is the most worrying news.
11:00 p.m. The child is asleep. His wife is preparing for work tomorrow. Wang Lei steps onto the balcony. Longgang District at night. In the apartment complex across the way, lights are out in patches. There was a time when most windows were still lit at eleven. Now more than half are dark. Empty units are multiplying. People who could not keep up with their mortgages and left. People who have not left but turn off the lights.
Wang Lei does not count the dark windows. If he counts, he will start calculating when his turn comes.
Section C: Cross-Comparison — Same Formula, Different Pain
Two Lives in the Mirror
The same AI passing through two systems produces different forms of suffering.
| Dimension | Sarah (US, age 42) | Wang Lei (China, age 38) |
|---|---|---|
| AI replacement cost ratio | 1:10–20 (salary vs. AI subscription) | 1:8–17 (CS team payroll vs. AI) |
| Monthly deficit after displacement | $1,354 (~9,800 yuan) | 16,540 yuan (~$2,286) |
| Unemployment benefit replacement rate | 40–50% (of prior income) | 16–20% (take-up rate ~1%) |
| Core debt | Student loans: $47,000 | Mortgage: 3.18 million yuan (~$440,000) |
| Age barrier | 40 (ADEA protection, weak enforcement) | 35 (legal protection virtually absent) |
| Income decline in gig transition | 40–60% | 60–75% |
| Political expression | Voting, social media, union attempts | tang ping (lying flat), bai lan (letting it rot) |
What does this table say?
In absolute terms, Wang Lei's monthly deficit is nearly twice Sarah's. The scale of core debt is overwhelmingly larger on Wang Lei's side. Between $47,000 in student loans and roughly $440,000 in mortgage debt, the ratio is nearly ten to one.
But Sarah's deficit includes an item Wang Lei does not face: health insurance at $1,200 per month. Lose a job in America and you lose your health coverage. In China, a displaced worker can still enroll in the urban resident medical insurance plan for 300 to 500 yuan per year ($42 to $70). Coverage is narrower than under an employer plan, but it still covers 50 to 60 percent of hospitalization costs. For Sarah, dropping COBRA means abandoning chronic disease management.
Wang Lei's debt has a feature Sarah's does not: inescapability. In America, student loan debt is nearly impossible to discharge in bankruptcy, but switching to an Income-Driven Repayment plan can lower the monthly payment to $300 to $400. Interest accrues, but immediate cash flow improves. Wang Lei's mortgage offers no such flexibility. Pay 16,900 yuan each month or default. Shenzhen introduced a pilot personal bankruptcy system in 2021, but its scope remains extremely limited.
Speed and Duration
Displacement in America is fast. It is a cliff.
Layoff → six months → unemployment benefits end + savings depleted → the cliff. The fall from that cliff is swift. But after hitting the bottom, legal and social pathways to recovery exist. The ADEA offers protection, weak but present. The ACA provides health insurance, inadequate but available. Bankruptcy protection (student loans excepted) opens the possibility of a fresh start. The existence of a possibility and its realization are different things, but the door is there.
Displacement in China is slow. It is subsidence.
N+1 severance, twelve to fifteen months, severance exhausted, gig conversion or return to hometown. The process is not sudden. Wang Lei will not be on the street tomorrow. But as the deficit accumulates month after month, options disappear one by one. A two-year slide toward the bottom. And at the bottom, the pathways to recovery are structurally narrow. The age-35 crisis shuts the door on reemployment. The underwater mortgage locks down mobility. A 1 percent unemployment insurance take-up rate reveals the absence of a safety net.
"Free fall in freedom" versus "stagnation under control."
Sarah in America is free. Free to fall. Every safety net depends on individual action. Apply for COBRA, enroll in the ACA, pursue retraining. Sarah must do all of it herself. The burden that goes by the name of freedom.
Wang Lei in China is inside the system of control. The state attempts to absorb. Unemployment insurance, public medical insurance, retraining programs. They exist on paper. But a 1 percent take-up rate reveals the distance between form and substance. The void that goes by the name of control.
Two Night Scenes
Chicago, 10:00 p.m. Sarah has Netflix running on the television and her laptop open. The screen plays a drama; her eyes are on the federal student loan website. She is filling out paperwork to switch to Income-Driven Repayment. IDR would lower her monthly payment from $541 to $300 or $400. But the repayment period stretches, and interest piles up in the interim. Pay less now, pay more later. Not solving the problem but relocating it into the future.
Her hand stopped at the "Current Employment Status" field. Checking "Unemployed." It is an administrative step, nothing more. And yet her finger will not move.
In another browser tab, Westlaw's "AI-Assisted Legal Research" online course is open. Tuition: $299. One-sixth of a single month's COBRA premium. Sarah has already added it to her cart. She tells herself that learning AI-assisted legal research will give her a competitive edge at her next job. But the skill this course teaches is precisely the skill that replaced her. Paying money to learn the tool that displaced her. She sees the contradiction and still cannot close the tab. For twelve years Sarah was "the one who finds things." Forgetting how to find is more frightening than losing the job itself.
Shenzhen, 11:00 p.m. Wang Lei has moved from WeChat Moments to a real estate forum on Douban. A post titled "How to Escape Negative Equity" has hundreds of comments. Most reach the same conclusion: you cannot escape. Surrender the apartment to auction and your credit is destroyed. Hold on and you bleed every month. Someone comments: "Hang in there. In three years, prices might recover." A reply below: "In three years, my mortgage interest alone adds another 600,000 yuan."
Wang Lei closed the forum. Then he did what he does every night. He opened the Meituan rider registration page. "Requirements: 18 or older, in good health, smartphone owner." Three lines. His resume lists ten years of CS team management, twelve consecutive quarters of meeting customer satisfaction KPIs, and experience evaluating a staff of thirty. Between the places that resume can take him and the places where three lines of qualifications suffice, Wang Lei freezes every night. Tomorrow he will tell his wife he is going to the talent market. Looking for a CS manager position. They both know that applicants over thirty-five do not even get their resumes accepted. He will say it anyway. One more week.
Wang Lei closed the Meituan page and got up to check the light in the child's room.
Both of them, at night, in front of a screen, wrestling with numbers. One on a student loan website. One on a real estate forum. The numbers are different in kind, but the helplessness before them is the same.
Section D: Meta-Argument — "The Illusion of Coexistence" and the Return to Volume 1
Four Illusions
Every productivity explosion has been accompanied by an "illusion of coexistence."
A period when the old order and the new temporarily coexist. During this window, people say: "This time is different. The new technology will complement existing jobs. They will coexist." And every time, the moment the technology crosses the threshold of "good enough," the illusion shatters.
The illusion of coexistence has occurred twice before.
The first. Rome's smallholders — the proletarii. In the early-to-mid second century BC, small farmers believed that when they returned from war, they could go back to farming. But after the Second Punic War, large slave-worked estates — the latifundia — expanded rapidly on the back of war spoils and captive labor. Smallholdings were absorbed into plantations. While the farmers fought abroad, their livelihood vanished at home.
The second. Lancashire's handloom weavers. In the 1790s through the 1810s, the early Industrial Revolution's demand explosion gave weavers more work than they could handle. Factory spinning machines mass-produced yarn, and that yarn needed weavers to turn it into cloth. "Machines are actually increasing our work." But after the Napoleonic Wars ended in 1815 and the power loom went into mass production, the handloom weaver's weekly wage fell from 25 shillings to 4.5 shillings. An 84 percent collapse. Technology that had multiplied their work proceeded to eliminate it.
In Volume 2, the third and fourth illusions are shattering now.
The third. Sarah's illusion. 2023–2024. "AI is just a tool; judgment remains the human's domain." Law firms introduced AI as a paralegal's assistant. "Finish a six-hour research task in fifteen minutes and spend the rest of your time on higher-value work." But when GPT-4-class models scored in the top 10 percent on the U.S. bar exam, the boundary between "assist" and "replace" collapsed. What was the paralegal's "higher-value work"? Whatever AI could not yet handle. That scope narrowed every month.
The fourth. Wang Lei's illusion. 2020–2022. "AI only handles simple FAQs. The complex stuff requires a human." During the era of rule-based chatbots with clear limitations, this was true. But in 2023, when large language models were integrated into customer service platforms, the definition of "complex" itself changed. Emotional complaints, multi-condition refund requests, cross-referenced inquiries. Every time AI's processing scope expanded, the human agent's role contracted. The territory Wang Lei had described as "things AI can't handle" became AI's territory within a year.
The four illusions share a common structure.
| Era | Coexistence period | Content of illusion | Tipping point |
|---|---|---|---|
| Roman smallholders | ~100 years | "We can farm again after the war" | Post-Punic War latifundia expansion |
| Handloom weavers | ~40 years | "Rising demand means weavers are still needed" | End of Napoleonic Wars + power loom diffusion |
| Sarah | ~2 years | "AI only assists; judgment is the human's domain" | GPT-4 scores top 10% on the bar exam |
| Wang Lei | ~2 years | "AI handles simple FAQs; complex cases need humans" | LLM integration, complex inquiry processing |
The coexistence window is shrinking. For Rome's smallholders, it was a hundred years. For the handloom weavers, forty. For Sarah and Wang Lei, two. Technology is reaching the "good enough" threshold faster. The time to adapt is disappearing.
Engels' Pause, Twenty-First-Century Edition
There is a concept called "Engels' Pause."
Between 1780 and 1840, British labor productivity rose 46 percent, but real wages increased by only 12 percent. Robert Allen's calculation. More than two-thirds of the productivity gains did not reach workers. A period in which the wealth generated by technology fails to flow sufficiently to labor.
The handloom weaver's weekly wage was the embodiment of this pause. From 25 shillings to 4.5. While factory productivity soared, the weaver's income collapsed. The wealth that technology created went to the factory owner; what remained for the weaver was poverty.
Sarah and Wang Lei exist inside a twenty-first-century Engels' Pause.
AI is raising the legal industry's productivity. Allen & Overy cut research time by 40 percent. More work done in the same hours. But the fruits of that productivity gain go to law firm partners and Big Tech. What came back to Sarah was a termination letter.
AI is raising Chinese e-commerce productivity. Taobao's AI handles 97 percent of customer inquiries. The work of thirty people, done by AI. Costs reduced by a factor of eight to seventeen. But those savings become platform profits. What came back to Wang Lei was N+1 severance of 132,000 yuan. Eight months of mortgage payments.
Productivity rises; workers' income falls. Two hundred and sixty-five years separate 1760 Lancashire from 2025 Chicago and Shenzhen, but the structure is identical.
But One Thing Is Different This Time
What replaced Rome's smallholders was physical labor. Plowing, sowing, harvesting.
What replaced the handloom weavers was also physical labor. Spinning yarn, weaving cloth. Hands and feet.
What replaced Sarah is cognitive labor. Reading case law, analyzing it, summarizing it. What replaced Wang Lei is also cognitive labor. Understanding a customer's words, identifying the problem, proposing a solution. The mind.
For the first time since the Industrial Revolution, "people who earn a living with their minds" are direct targets of displacement.
This is the point where Volume 1's pattern expands in Volume 2. The displaced of Volume 1 were pushed out of the execution layer. The smallholder's plowing, the weaver's loom, the Rust Belt worker's welding and assembly. Physical execution replaced by machines.
The displaced of Volume 2 are being pushed out of the cognitive layer. Sarah's legal research, Wang Lei's customer management judgment. Reading information, finding patterns, making decisions. This domain was long considered "uniquely human." Machines replaced the hand two hundred years ago. Machines are beginning to replace the mind now.
In Volume 1's framework, Sarah's structural position is the modern equivalent of Rome's smallholder. Just as the farmer returned from military service to find his land absorbed by a plantation, Sarah worked diligently for twelve years and discovered her job had been absorbed by software. "While I was working, my work disappeared."
Wang Lei's structural position is the modern equivalent of the Lancashire handloom weaver. The reason weavers could not abandon the loom even at devastatingly low wages was that weaving was both a skill and an identity. "I am a weaver." The reason Wang Lei resists becoming a delivery rider is the same. "I am a manager." The fact that he once led thirty people is the foundation of his self-worth. Starting deliveries is not a question of income before it is a question of identity. A question of mianzi —face.
The weavers eventually became factory workers. Wang Lei may eventually become a delivery rider. But the weavers had decades of transition. Wang Lei has only the twelve to fifteen months before his N+1 severance runs out.
Volume 1 Connection Point: The Contrast Pair Expands
In Volume 1, the contrast pair consisted of two figures within a single society — the handloom weaver and Arkwright. One was displaced; the other rose. The divergence was not the technology itself but the ability to organize technology into a system.
The contrast pair in Volume 2 operates along a different axis. Sarah and Wang Lei are both being displaced. The difference is not "who was displaced" but "within which system they were displaced." If Volume 1's question was "Where does the difference between the displaced and the discerning originate?" then Volume 2 adds this question: same technology, different systems, different forms of suffering. Technology is universal. The shape of suffering is determined by institutions.
Transition: What Has Not Yet Arrived
Final Scene
Chicago, an evening in January 2025.
Sarah Kowalski is standing by her apartment window. Lake Michigan is visible. At dusk the lake is the color of lead. The January surface where ice and water coexist. She is thinking about whether to call the community college or reopen Indeed. She will probably do both. Probably neither will produce results. She has to do it anyway. There are no other options.
At the same moment — 7:00 p.m. in Chicago is 9:00 a.m. in Shenzhen — Longgang District, Shenzhen.
Wang Lei is looking down at the apartment complex from his balcony. Morning sunlight slants between the buildings. He calculates when the child will be back from school. 4:00 p.m. Open BOSS Zhipin and the age-35 filter will block him. Open the Meituan app and the rider registration screen will appear. He stands on the balcony without opening either.
Neither of them knows. That on the other side of the planet, someone shares their predicament. But the reader knows. That the same technology pushed both of them out. That the AI developed competitively by two empires is simultaneously displacing ordinary citizens of both.
Volume 1's formula is at work.
Technological innovation → capital concentration → social instability.
We have reached the third stage. Sarah's monthly deficit of $1,354 is social instability. Wang Lei's 1.05 million yuan of negative equity is social instability. The instability exists in individual nights, in the form of numbers.
The fourth stage — institutional redesign — has not yet arrived.
It took sixty-four years from the Industrial Revolution to an effective Factory Act. From Arkwright's first factory in 1769 to the Factory Act of 1833. During those sixty-four years, the handloom weaver's weekly wage fell from 25 shillings to 4.5, and the average age of death for a Manchester worker was seventeen. Institutions did not arrive in the interim.
Where is the AI era's "Factory Act"?
In the United States, not a single comprehensive federal AI regulation has been enacted. More than one hundred AI-related bills have been introduced, but only one — the TAKE IT DOWN Act — has been signed into law. There are 3,570 AI lobbyists, accounting for 26 percent of the total. Seventy-nine percent of the public supports AI regulation. Yet the laws do not pass.
China enacted provisional measures on generative AI nine months after ChatGPT's launch — the fastest response in the world. But the regulation targets companies that use AI, not the people displaced by it. A 1 percent unemployment insurance take-up rate reveals the distance between the speed of regulation and its substance.
In both countries, institutional redesign for Sarah and Wang Lei has not yet begun.
The historical band stretches from fourteen to sixty-four years. Can the AI era compress that interval? Or will decades of suffering precede the arrival of institutions?
Sarah will turn from the window and open her laptop. Wang Lei will come in from the balcony and prepare the child's lunch.
Both of them must build their tomorrow before institutions arrive.
That is the condition of the displaced. It has never been otherwise.
Investor Lens: For Readers of Part 4
What Part 4 reveals is not numbers but structure. From the chip war (Ch. 13) and the contrast pair (Ch. 14), an investor should read three signals.
First, decoupling is a cost. ASML's China revenue share shrank from 36.1 percent (2024) to roughly 33 percent (2025), reflecting the effect of export controls and the beginning of supply-chain restructuring costs. Companies forced to build dual supply chains (in semiconductor equipment, materials, and packaging) will see short-term margin compression, but over the long run they secure the position of supplying infrastructure to both markets. Mid-stream firms in South Korea, Japan, and Germany fall into this category.
Second, the scale of the displaced determines the political tipping point. The stories of Sarah (US) and Wang Lei (China) are not individual tragedies but samples of a structural trend. If AI-related layoffs spread into white-collar America, regulatory pressure will compress valuation multiples across the tech sector. If youth unemployment and underwater mortgages compound in China, the timeline for domestic consumption recovery is delayed. Tracking the scale of the displaced in both countries is a leading indicator for macro positioning.
Third, "free fall in freedom" versus "stagnation under control" differentiates asset-class behavior. For U.S.-type risk (high volatility, sharp corrections, rapid recovery), options strategies are effective. For China-type risk (low volatility, slow erosion, opaque inflection points), avoiding the value trap is paramount. Even under the same label of "AI beneficiary," the shape of risk differs.
These three signals connect directly to the three scenarios in Chapter 15. If decoupling accelerates, Scenario C (a fractured world) gains strength. If the scale of the displaced crosses the political tipping point, the conditions for Scenarios A and B weaken in both countries — because regulatory pressure constrains innovation while delayed consumption recovery stalls growth. The asymmetry of risk — America's plunge versus China's erosion — means that even the same "US-China allocation" carries different volatility structures. Chapter 16's ten key signals and scenario-based asset allocation are built on this architecture.
The question shifts. From "what is happening" to "what could happen." A scenario in which institutional redesign arrives, a scenario in which it does not, and a scenario in which it arrives too late. How Sarah's and Wang Lei's lives diverge across these three paths is the next inquiry.
Sarah Kowalski and Wang Lei are composite characters. Each was constructed from real cases, public reporting, and labor market data from the United States and China, respectively. Specific details — workplace, family composition, financial circumstances — reflect typical cases but do not reproduce any particular individual. All figures in this chapter follow verified values registered in facts_registry.md.