Will AI Replace Programmers? What 200-Year-Old Weavers Teach Us

Will AI Replace Programmers? 200 Years Ago, Handloom Weavers Already Had the Answer

Around the year 1800, an English handloom weaver sat at home. His foot pressed the treadle, his hand pushed the shuttle back and forth, again and again. The machine was neither beautiful nor advanced — but it was his livelihood. His value lived in those two hands.

Then one day a power loom appeared in the factory. It did not require decades of craft from a single person. It only needed power, plus a handful of people who could keep it running. And the weaver discovered that the "skill" he had always been proud of was suddenly worth much less.

For programmers today, exactly the same thing is happening in the AI era.

Before talking about AI, go back 200 years

Before the Industrial Revolution, most British textiles came out of cottage industry: an order arrived, raw material was delivered to the home, the worker wove cloth there, the merchant collected it and resold it. In that model a skilled weaver mattered enormously, because the machines were weak and output depended heavily on human expertise.

Here is the single most important sentence: the Industrial Revolution did not make "people" worthless overnight. It gradually reduced the scarcity of one very specific skill.

The cruelest part: machines do not take jobs first — they cut prices first

Suppose a skilled weaver could produce 10 metres of cloth a day. After ten years of training, almost nobody could replace him, so the market price of his labour was high. Then the machine arrived and let an ordinary worker finish what only a master used to be able to do. The weaver's advantage did not disappear — it simply stopped being worth as much.

This is the part of technological revolutions that people most often miss. Everyone asks "how many jobs are lost?" Almost nobody asks "how much is one person's skill still worth on the market?" That second question is exactly what every programmer should be watching today.

The first wave of shock that programmers are already facing

Building a website used to require a product manager, a UI designer, a frontend developer, a backend developer, QA and an operations engineer. A capable AI Coding Agent can already take part in most of those stages. Tell it: "Build me a SaaS with user registration, login, payments and an admin panel." It may not succeed in one shot, but it already writes code, runs tests, fixes bugs, generates database schemas, implements APIs and drafts documentation.

If a feature that once took 10 hours now takes 2, where did the other 8 hours go? The market will start redefining the role around them.

What history actually teaches us is not "programmers will lose their jobs"

There is a very common misunderstanding about the Industrial Revolution: machines arrived, workers became unemployed, so workers smashed the machines. History is more complicated than that. The Luddite movement really did happen in early-19th-century Britain — around 1811 to 1816 workers destroyed textile machinery to protest falling wages, changing production systems and threatened employment. But they were not simply "anti-technology". What was being destroyed by machines was the old contract between labour, income and social order.

Industrialisation also did not make every textile worker vanish at once. Mechanisation unfolded over many years, and the machines themselves kept improving. What really reshapes a job market is rarely the day a technology first appears — it is the following years, when it keeps getting better.

Replace the loom with AI and the story becomes instantly familiar

Both eras placed enormous value on a skill that could be standardised, copied and automated. The weaver's core skill was "I can turn thread into cloth." The programmer's core skill was "I can turn requirements into code." AI is making the second sentence cheaper and cheaper.

The big reversal that ends the panic

If you are already anxious — "am I about to change careers?" — slow down. After the Industrial Revolution, humanity produced more goods, not fewer, and entirely new industries appeared. The underlying law is simple: when a way of producing becomes cheaper, people do not stop producing; they start producing all the things they previously could not afford to make.

If a website used to cost one programmer two weeks and AI now delivers it in a day, the world will not need fewer websites because they got cheaper. On the contrary, a million small websites that nobody was ever willing to build may suddenly get built. That is the most underestimated opportunity of the AI era.

AI will probably not eliminate programmers — it will first eliminate parts of a programmer's job

Those two statements look identical and are completely different. Break a programmer's work apart:

The real danger is not "the profession of programmer disappears." The real danger is that 60 of the 100 skills a programmer was expected to master suddenly stop being scarce.

At that moment the most exposed people are not juniors — they are "the ones who only had those 60 skills"

Imagine two programmers.

Programmer A spends every day on writing prettier Java, micro-optimising React, tuning SQL, and reading a framework's source code. He works harder than most people. But his entire edge is: "I write code faster and make fewer mistakes than you."

Programmer B is not as fast at code, but he deeply understands one industry: why users pay, how the company's real processes work, which requirements are worth building, which should never be built, and how to use AI to turn an idea into a shipped product. His edge is: "I know what should be built, and I can make AI build it."

In the AI era, A is arguably the more vulnerable one, because what AI collapses fastest is the cost of execution. The expensive thing becomes judgement.

So the upgrade programmers need is not "writing more code"

The AI-era programmer starts to look like a product manager + engineer + project lead combined. Previously the job was: the product manager tells me what to build, and I build it. Increasingly it becomes: I understand the user's problem → I design the solution → I have AI generate the code → I verify the result → I ship → I iterate from data.

Notice that the code is now only a short segment in the middle. The complete capability chain is: find the problem → define the problem → design the solution → orchestrate AI → verify the result → own the outcome.

One person used to write a few hundred or a few thousand lines a day. Now one person can review and govern tens of thousands of AI-generated lines. Your bottleneck becomes: "Can I tell whether what AI wrote is actually correct?" And here the weaver returns: after the machines arrived, not everyone connected to textiles disappeared. The industry re-divided the work — some operated machines, some designed products, some sold, some invented new business models. Old roles shrank while new ones appeared.

The brutal part is this: new jobs do not automatically go to the people who used to hold the old ones. The speed at which an old skill becomes obsolete is often faster than the speed at which you learn a new one.

What programmers should genuinely be afraid of

Not AI, and not even "the AI programmer". The thing worth fearing is that your capability is becoming easier to purchase. A company once paid 200k a year for one programmer because that person could reliably convert requirements into software. If one person plus AI does the work of three, why would the company hire the old way? That is not a moral question, it is an economic one — it is the inevitable consequence of higher productivity. It happened with the Industrial Revolution, it happened with the internet, and it is very likely happening with AI.

The counterintuitive conclusion: the stronger AI gets, the more valuable great programmers become

An ordinary person with AI builds a small demo. An excellent engineer with AI builds a complex system. An engineer who understands the business builds a product that actually makes money. So AI does not flatten everyone to the same level — it amplifies the gap between people.

Give 100 weavers the same super-powered loom and they will not all earn the same amount. Some will optimise throughput and fabric quality, some will go find customers and open a factory, some will design an entirely new product. The machine itself stops being the advantage. The winner is whoever uses the machine to create more value.

This is how an odd kind of programmer will emerge: not the fastest typist in the company, but someone who simultaneously manages several AI agents — frontend, backend, QA, documentation — while he defines goals, decomposes problems, reviews output and makes the critical calls.

Previously: 1 programmer = 1 unit of production. Going forward: 1 programmer + AI = a small software production team.

If you are between 20 and 35, get clear on these five layers

Do not go and master 100 AI tools because you got scared. Do not switch careers because you got anxious. And do not spend all your time guessing which framework comes next. Upgrade these five layers instead.

The higher the layer, the harder it is to replace with plain code generation.

Layer 1: stop treating "I can code" as the destination

The trap for newcomers is: "If I go deep enough technically, AI cannot replace me." That sentence no longer holds. If your entire competitive edge is writing good Java, good React, good SQL and good Kubernetes, you are racing an ever stronger tool. The better formula is technical depth + business depth + AI leverage. You are not competing with AI on typing speed; you are learning to make AI your productivity.

Layer 2: move from executor to problem definer

An ordinary person asks AI: "Write me a login page." An expert asks: "Why does this user need an account at all? When should we not require login? Why is our login failure rate high? How does this flow affect conversion?" The first is a request; the second is problem solving. AI is superb at problems that have already been defined clearly, so the ability to define problems becomes more and more valuable.

Layer 3: actually understand one industry

"Pure programmer" gets harder. The programmer who understands finance, healthcare, e-commerce or education becomes more valuable, because domain knowledge is not something a single prompt can replace. Real complex businesses are full of unwritten rules, competing interests, user habits and risk — and that is the hardest part of the physical world.

Layer 4: shift from code assets to judgement assets

The most valuable thing a programmer used to own might have been code on GitHub. Increasingly it will be the record of correct decisions he made. AI can generate 1000 lines in ten seconds, but if the direction is wrong, those 1000 lines are simply garbage produced faster. The strongest people start to resemble film directors — yet somebody still has to answer "what film are we making?"

Layer 5: learn to take responsibility for the result

This is the most important layer. AI helps you write code, but no boss, client or user will hold AI accountable. In the end a human must answer: why did we not make money, why is the data wrong, why did the system crash, why does nobody use the product? People who own outcomes are hard to replace, because accountability is itself a form of value.

Back to that weaver in 1800

What he truly lost was not only a job. He lost the fact that his craft was scarce on the market. That was the deepest change of the Industrial Revolution. The machine never said "you have no value". It told the market: "This person's skill is no longer scarce."

AI is doing the same thing to programmers. It is not saying "programmers have no value." It is telling the whole market that the skill of writing code is no longer scarce. That is the biggest watershed in the career of a programmer.

Two kinds of programmers will emerge over the next few years

The first treats AI as a faster typewriter. They still write the code themselves and occasionally ask AI a question. The second treats AI as a production system: they decompose tasks, define goals, manage agents, verify results, and wire AI into the company's whole workflow. Today both groups are still called "programmers". In a few years they may be two completely different professions.

History does not repeat, but it often rhymes

The Industrial Revolution did not stop humans from working. The internet did not erase every traditional industry. Automation did not empty every factory. Each technological revolution has done something structurally similar: it lowered the cost of one way of producing. Then the market changed. Some people were pushed out or forced to reinvent themselves. Some seized the chance and created new industries. In the end the whole society's way of working was redefined. AI is doing exactly the same.

The question you should actually be asking

If you are a programmer between 20 and 35, ask yourself: "If writing code becomes as cheap as typing, what value can I still provide?" AI may not be your ending; it may be your lever.

The weavers of 1800 could not choose the machine. They saw "the machine weaves cloth" and concluded "then what are we for?" But history shows that machines replace part of human work and then redefine what a valuable person looks like. Programmers stand at a similar crossroads: "being able to write code" will very probably be worth less and less. That is not the same as "programmers being worth less and less."

People who understand technology can start solving problems they never had the capacity to solve before. One person can build a product that once needed ten. A small team can ship software that only large companies could once afford. An ordinary programmer can now own production capacity that used to belong only to startup founders. That is what makes AI genuinely frightening — and genuinely exciting.

So stop asking "will AI replace me?" and change the question: "If AI works for me, what am I going to have it do?" Two hundred years ago weavers had no choice about the machine. Today's programmers can choose where to stand. Whoever stands in front of the machine gets changed by it. Whoever learns to operate it uses it to change everything else.

FAQ

Will AI replace programmers?

AI is unlikely to remove the profession, but it is already removing specific tasks — CRUD, simple APIs, tests, common bug fixes — and reducing the market price of those skills. Programmers who move up the stack toward problem definition, architecture, business understanding and outcome ownership become more valuable, not less.

Should I learn more AI tools to stay safe?

Tool collection is not the moat. Focus on five layers: prompting and delegating to AI, verifying AI output, judging which requirements are worth building, orchestrating AI across a production pipeline, and owning the business result.

Which skills keep their value as AI coding gets stronger?

Domain expertise, system architecture, problem framing, technical judgement, and accountability for production outcomes. These are the areas where scarcity — and therefore price — has not collapsed.