“No one escapes the permanent underclass” is the kind of sentence the internet was built to reward.
It has the whole package: totalizing dread, class anxiety, technological panic, and just enough plausibility to feel intelligent. It sounds biblical. It sounds final. It sounds like the sort of thing you should repeat to prove you’re no longer fooled by optimism.
That’s exactly why it works.
The permanent-underclass idea usually means a mass of people locked out of the basic promises of adult life: steady work, housing, family formation, savings, status, and upward motion. The darker version goes further. It imagines an economy that permanently stops needing a large share of people at all. A world where millions remain present as consumers, renters, debtors, patients, and data points, while losing any real claim to productive dignity.
It’s easy to understand why the idea spreads.
Younger generations can look around and see every classic milestone turned into a toll booth. A starter home now feels like something that requires two professional incomes, family help, and lucky timing. Rent often gets first claim on the paycheck. Healthcare, childcare, tuition, insurance, and groceries have trained people to feel like the essentials keep sprinting ahead of wages. Even when someone is employed, adulthood can still feel delayed.
The job market adds another layer of humiliation. People apply to hundreds of jobs and hear nothing back. AI writes the resume. AI adjusts the cover letter. AI human-resources systems filter the application. The whole process can feel like robots talking to robots while the human being waits outside the room. The old advice of “just apply yourself” starts to sound like a joke when the application process itself has become an arms race.
Then AI arrives as the final accelerant. It touches writing, coding, design, research, analysis, customer service, legal work, marketing, administration, and nearly every form of symbolic labor that educated people were told would keep them safe. That’s why the fear has become so emotionally powerful. It attacks both income and identity.
The feeling deserves respect. The prophecy doesn’t.
The permanent-underclass story makes one mistake over and over again. It treats the awkward middle of a transition as the final shape of history. It assumes today’s broken hiring process, today’s housing bottleneck, today’s job titles, today’s credentials, and today’s AI tools are the architecture of tomorrow.
They aren’t.
Civilization rarely announces its next structure in advance. The new arrangement usually arrives as weirdness, then convenience, then normal life. By the time everyone agrees it’s real, the people who said it was impossible have already moved on to warning about the next thing.
A person in 1975 could understand a secretary, a factory worker, a bank teller, a professor, a mechanic, a lawyer, and a doctor. They probably wouldn’t have predicted app developers, cloud architects, YouTubers, UX designers, SEO strategists, data privacy officers, podcast producers, cybersecurity analysts, telehealth coordinators, remote fractional CFOs, Shopify store owners, or the strange economy of people making a living by explaining software to other people on the internet.
The future enters as a joke before it becomes a job title.
That’s the part doom misses. We keep trying to imagine the next economy using the categories of the current one. Then the next economy arrives with new roles, new signals, new institutions, new scams, new ladders, new status games, and new ways to be useful.
Technology destroys tasks. It reorganizes roles. Then it creates demand the previous world could barely imagine.
Jevons paradox is one way to see this. When a technology makes something more efficient, society often uses more of it. More efficient steam engines helped push coal consumption higher because cheaper power made power useful in more places. Cheaper computing pushed computation into phones, cars, watches, thermostats, hospitals, banks, factories, and toys. Cheaper cameras turned images into a global language. Cheaper publishing created an ocean of text.
AI follows that same pattern. If AI makes basic cognition cheaper, we should expect more cognition running through every system. More proposals. More simulations. More research summaries. More forecasts. More lessons. More software experiments. More audits. More legal drafts. More medical triage. More financial models. More personalized tutoring. More business plans. More synthetic garbage too.
Cheap output creates abundance. Abundance creates demand for selection.
That’s where human value moves. When the world gets flooded with answers, the scarce thing becomes judgment. Which answer is true? Which one is useful? Which one is safe? Which one fits the client, patient, student, buyer, team, market, law, or moment? AI can generate options at scale. Someone still has to decide what matters.
We’ve seen this before. Old office work had typing pools, filing cabinets, dictation, physical mailrooms, paper calendars, and manual document preparation. Computers destroyed huge parts of that world. Yet the office survived. Secretarial work mutated into administrative assistance, executive support, operations coordination, project management, HR operations, medical administration, legal administration, customer success, and business support.
The worker who only typed letters lost power. The worker who understood the organization, protected the executive’s time, coordinated people, handled context, managed systems, and solved ambiguous problems became more valuable.
Spreadsheets offer another example. They absorbed calculations that once required armies of clerks. Finance grew faster, more analytical, more data-driven, and more central to corporate life. Computers killed repetitive calculation. They also expanded the number of decisions worth modeling.
The job title can die while the human function survives.
AI will destroy certain tasks. It’ll reduce the value of generic writing, generic coding, generic research, generic analysis, and generic design. That shouldn’t scare us into fatalism. It should clarify the pattern. Generic work loses value when machines can do it. Human judgment gains value when machines can produce infinite drafts, answers, and options.
The human bottlenecks will shift toward trust, taste, context, accountability, execution, and relationship.
Trust matters because someone has to stand behind the output. A model can draft a contract, suggest a diagnosis, summarize a meeting, generate code, or analyze a dataset. When consequences arrive, a responsible person or institution has to own the decision.
Taste matters because infinite options are useless without selection. The average AI answer will be good enough for many things. Excellence will still require a sense of timing, audience, aesthetics, strategy, restraint, and fit.
Context matters because real organizations are full of politics, incentives, grudges, habits, regulations, history, and unstated rules. A clean answer can fail inside a messy institution. Humans who understand the room will remain valuable.
Execution matters because most life still touches atoms. Houses still need framing. Wires still need running. Pipes still need fixing. Patients still need care. Children still need adults. Elderly people still need help. Data centers need land, power, cooling, electricians, technicians, security, and operations. Robots need maintenance. Infrastructure needs crews. Energy needs permitting and construction. The future gets more digital, while the physical world keeps collecting the bill.
Relationship matters because humans still prefer humans when stakes are high. A person may accept an AI-generated answer for trivia, shopping, or a first draft. They want a trusted human when money, health, children, reputation, grief, risk, or status enters the picture.
This is why the underclass meme is too simple. It sees task destruction and assumes human uselessness. A better reading is that human value migrates, usually before society has good words for where it went.
The middle-class story is also more complicated than the collapse narrative suggests. The American middle-income share has fallen substantially since the early 1970s. That sounds like pure decline until you look at where some of the movement went. The lower-income share grew, which is the painful part. The upper-income share also grew a lot.
The collapse narrative misses that upward movement. The middle class is shrinking partly because some people are falling behind and partly because more people have moved into the upper-income tier.
You can see this in daily life. A large part of the upper-middle class now lives with advantages older generations could hardly imagine. Remote and hybrid work let many professionals avoid brutal commutes, eat lunch with their kids, attend school events, live farther from expensive office cores, and regain small pieces of family life that used to be sacrificed to the highway. Digital tools make learning, investing, building, publishing, networking, and starting a small business easier than before. A motivated person can access courses, markets, audiences, software, mentors, and collaborators from a laptop.
That’s a real improvement in the texture of life.
A lot of people today have more flexibility, more information, more tools, more entertainment, more market access, more learning access, and more creative leverage than wealthy people had in previous generations. They may still be angry about housing, and often for good reason. But it’s strange to call this a march toward permanent uselessness when so much of daily life has become more capable, more flexible, and more abundant.
The benefits are uneven. A nurse can’t do a hospital shift from home. A teacher can’t fully automate classroom presence. An electrician, plumber, firefighter, mechanic, police officer, warehouse worker, server, or home health aide can’t simply become remote. But this unevenness points toward churn, not closure. The future looks less like one giant underclass and more like a constant rearrangement of advantage around tools, location, skills, timing, ownership, and leverage.
The mistake is thinking we have to know the final arrangement in advance.
We don’t.
New structures will emerge because people hate friction. They hate wasting time. They hate being poor. They hate broken hiring systems. They hate expensive housing. They hate bad schools. They hate useless credentials. They hate bottlenecks. They hate feeling powerless. So they build around the problem, route around the problem, profit from solving the problem, vote against the problem, move away from the problem, automate the problem, or create a new language that makes the old problem look ridiculous in hindsight.
That’s how systems work themselves out. Not magically. Not morally. Because billions of people keep responding to incentives, copying what works, abandoning what fails, and creating arrangements that nobody centrally designed.
Nobody had to fully imagine the internet economy in advance for it to arrive. Nobody had to predict every remote-work norm before millions of people discovered they could do their jobs from home. Nobody had to design the entire creator economy before people began building careers on platforms that originally looked unserious. Nobody had to write the final map of software employment before computing quietly entered everything.
The same thing will happen with AI.
At first, we’ll describe it with bad language. We’ll say jobs are disappearing. We’ll say people are being replaced. We’ll say nobody knows what young people should do. We’ll say the ladder is broken. Some of that will feel true because the old ladder really will be broken in places.
Then new ladders will appear.
Some will be obvious: AI operations, model auditing, workflow design, robotics maintenance, data-center infrastructure, biotech automation, elder-care coordination, cybersecurity, energy systems, education tools, healthcare navigation, and small-business automation.
Others will sound fake until suddenly they’re normal.
That’s usually how the future enters. First as a joke, then as a side hustle, then as a weird job title, then as an industry, then as something parents tell their kids to consider.
So the right response to the permanent-underclass panic isn’t to appoint yourself administrator of civilization. You don’t have to personally solve housing, redesign education, reform hiring, regulate AI, create the next credential system, and invent the new economy before breakfast.
People will do those things because people always do those things.
Some will do them for money. Some will do them for status. Some will do them because they’re annoyed. Some will do them because they’re ambitious. Some will do them because the current system failed them. Some will do them by accident. A lot of the best structures in life are created by people trying to solve their own immediate problem, then discovering that millions of other people had the same problem.
That’s why doom is such a bad bargain.
If you spend the next ten years grieving a permanent underclass that never fully forms, you still paid the emotional cost in full. You still lost the mornings. You still lost the ambition. You still lost the ordinary pleasures. You still trained your attention on collapse when you could’ve been living, learning, building, loving people, taking walks, having dinner, raising kids, reading books, going outside, and enjoying the strange abundance already here.
Doom feels serious because it borrows the posture of intelligence. It says, “I see through the illusion.” But sometimes the deeper illusion is despair itself. Despair pretends to be wisdom while quietly stealing the only years you actually get to experience.
Housing can still be too expensive. Hiring can still be dehumanizing. AI can still disrupt work. Institutions can still be slow, stupid, and unfair.
None of that proves permanence.
The permanent-underclass meme is powerful because it captures a real humiliation: the feeling that the world became expensive, impersonal, automated, and rigged all at once. It captures the weirdness of applying to a machine, being judged by a machine, paying rent through a portal, losing to asset inflation, and wondering whether the next machine will do your job before you ever get good at it.
But memes flatten time. They turn turbulence into destiny.
The future will be stranger than the doomers think. It’ll have new roles, new ladders, new markets, new status games, new institutions, new scams, new opportunities, new forms of comfort, and new ways to be useful. Some of the best parts haven’t been named yet.
That’s the reason for optimism. Civilization is adaptive at a scale no individual mind can model from inside the panic of the present.
AI will lower the price of capability. Lower prices will expand demand. Expanded demand will create new bottlenecks. New bottlenecks will create new work. Around that work, new structures will form.
The trap isn’t that humans become useless. The trap is believing the meme so deeply that you waste your life rehearsing grief for a future that never arrives.
The permanent underclass is a clickbait myth because permanence is exactly what makes the story clickable. It gives anxiety a shape. It gives resentment a villain. It gives uncertainty an ending.
But the future doesn’t have an ending yet.
It’s still being built by people responding to pressure, annoyance, opportunity, desire, boredom, ambition, love, laziness, necessity, and the simple human refusal to accept friction forever.
Permanence is the myth.