Why AIs Won’t Take Your Job (And Why That’s a Much Bigger Problem)
The real driver of the artificial revolution isn’t machine intelligence. It’s a brutal demographic mismatch. And that changes all the rules of the game.
By Claudio Bresciano
We’ve been reading alarmist headlines for years. “AI will replace lawyers by 2030.” “Radiologists will be obsolete within five years.” “Your job will disappear before you retire.”
The fear is understandable. But it’s based on a false premise.
Artificial intelligences aren’t replacing humans because they’re smarter, faster, or cheaper. They’re occupying those spaces because, in many parts of the world, there simply aren’t enough humans available to do the work anymore.
And that’s not good news. It’s the beginning of the greatest structural crisis of the 21st century.
The Great Demographic Mismatch: Exploding Demand, Collapsing Supply
Look at these figures. They’re not apocalyptic predictions. They’re official 2024 data:
- South Korea: fertility rate of 0.72 children per woman. The lowest on the planet.
- Japan: 1.2 children per woman.
- China: 1.0 children per woman. The most populous country in history is losing 200 million young people by 2050.
- Italy, Spain, and Germany: hovering between 1.2 and 1.4 children per woman.
The problem isn’t just a general shortage of labor. It’s a brutal structural mismatch: we will have an exploding and growing demand for services for an aging population that lives longer, and a scarce, almost non-existent supply of young workers to provide those services.
By 2040, approximately 400 million young workers will be missing globally. It’s basic arithmetic.
When rural hospitals in Spain and Italy close because there are no young doctors willing to live there, it’s not because an AI fired them. It’s because there are no doctors. When Japan automates 70% of its food logistics, it’s not for business efficiency. It’s because there are no truck drivers. When South Korea deploys caregiver robots in nursing homes, it’s not for technological innovation. It’s because there are no caregivers.
AI isn’t conquering the labor market. It’s plugging demographic holes.
Why This Changes Everything
For decades, the discussion about AI focused on the wrong question: “When will machines be more intelligent than us?”
The correct and urgent question is different: “What happens when an entire civilization depends on systems that act, but cannot be held responsible for their actions?”.
And this is where philosophy stops being an academic exercise and becomes a survival tool.
The Problem of Responsibility Without Agents
Imagine this scenario, which is already happening:
An autonomous vehicle runs over a pedestrian. Who goes to jail? The algorithm? The programmer who no longer works at the company? The company that went bankrupt three years ago?
A financial AI causes a stock market crash that wipes out billions in savings. Who pays the compensation?
An automated medical diagnosis system fails to detect cancer in an elderly patient. Who answers to the family?
The traditional answer — “the human who was supervising” — no longer works when there is no human supervising, precisely because of that demographic mismatch. There are no humans available to supervise.
This isn’t a technical problem. It’s a legal and ethical vacuum of historical proportions. And contemporary philosophy has no tools to resolve it.
Or at least, it didn’t.
The “Bresciano Test”: A New Way to Think About Responsibility
Within the framework of the Theory of Axiomatic Necessity (TNA), I’ve developed a distinction that proves crucial for this debate. It’s not about whether the AI “thinks” or “feels.” It’s about something more precise and more useful:
An AI is a purely operational system. It can process information, generate coherent responses, optimize routes, diagnose diseases, and write poetry. But it cannot instantiate experience. It has no domain of admissibility of its own. It cannot be a subject of moral responsibility because it cannot be a subject of experience.
In simple terms: an AI acts, but cannot answer for its actions.
Get Cbresciano’s stories in your inbox
Join Medium for free to get updates from this writer.
This isn’t a philosophical opinion. It’s a mathematical consequence of the architecture of any computational system. And it has immediate legal implications.
If an AI cannot be a subject of responsibility, then responsibility must fall on someone who can be: the human supervisor, the manufacturing company, or the State that authorized its deployment.
This distinction — between operational competence and legal capacity — is what I call the “Bresciano Test” in my papers. And it’s exactly the kind of tool that European, Japanese, and Korean courts will need in the next decade.
The Law That’s Coming
By 2038, it’s very likely we’ll see the first “Mandatory Human Supervision Law” in the European Union, Japan, and South Korea. Not out of fear of AI. Out of demographic and legal necessity.
The law will say something like this: “Every autonomous system must have a human validation node in its decision loop, or the system is structurally irresponsible and cannot be deployed in critical contexts.”
This law won’t be born from philosophy. It will be born from courts, from accidents, from lawsuits, and from social pressure. But when legislators look for a theoretical framework to justify it, they’ll find that TNA has already formalized it mathematically.
The Functional but Empty Society
But the deepest problem isn’t legal. It’s existential.
Imagine the year 2050. Trains arrive on time. Hospitals diagnose with precision. Agriculture produces enough. Logistics work. Civilization, in operational terms, functions better than ever.
But 60% of your daily interactions are with AIs. Your personal assistant is an AI. Your primary care physician is an AI. Your lawyer, your therapist, the restaurant waiter, the bus driver.
Everything works. But something has been lost.
Because information — genuine meaning, real connection — isn’t a property of objects. It’s a relational property. It emerges when two systems with experience share a code. And an AI, no matter how sophisticated, has no experience with which to share anything.
A society where the majority of relationships are human-AI is operationally stable but relationally empty. It functions. But it means nothing.
The “Human Zones”
It’s no coincidence that cultural movements are already emerging that anticipate this. By 2050, it’s likely there will be “Human Zones” in major cities: spaces where the use of AIs is prohibited to preserve human relational capacity.
Restaurants attended by people. Doctors who look you in the eye. Teachers who remember your name. Not out of nostalgia. Out of ontological hygiene.
Because if we delegate all functional relationships to machines to cover the demographic mismatch, we’re left with the relationships that machines cannot cover: those that require shared experience. And if we don’t practice those relationships, we lose them.
The Demographic Winter Has No Return
No birth rate policy has ever managed to reverse a fertility rate below 1.5 in any developed society in history. It’s a threshold you don’t come back from.
This means that dependence on AIs isn’t a technological fad. It’s a structural necessity. We’re not going to “disconnect” AIs because we’ve changed our minds. We’re going to need them more and more, because each year there are fewer humans available to do the work that civilization requires.
The question is no longer “should we use AI?” The question is: “how do we use AI without losing what makes us human?”.
The Answer Isn’t Technological. It’s Structural.
TNA offers a framework for thinking about this without falling into panic or naivety:
- AIs are not people. They cannot be, structurally. They have no domain of experience. This isn’t a temporary limitation that will be solved with more parameters. It’s a topological property of any computational system.
- Responsibility must remain with humans. Always. Even when there are no humans available to exercise it directly. If there are no humans, the system shouldn’t be deployed in critical contexts.
- Genuine relationships require shared experience. Delegating all functional relationships to machines impoverishes us relationally. We need to preserve spaces of non-delegable human interaction.
- Information isn’t an object. It’s a bond. And bonds require two ends with experience. An AI can simulate the bond. But it cannot be one of its ends.
Conclusion: The True Challenge of the 21st Century
AIs aren’t going to take your job. Simply, there’s no one to do it.
And that’s a much bigger problem than technological unemployment. It’s the problem of a civilization that must keep functioning with half its workers, delegating more and more functions to systems that act without experience, decide without responsibility, and respond without understanding.
The challenge of the 21st century isn’t building more intelligent AIs. It’s designing a civilization that remains human in a world where humans are scarce.
Philosophy, ethics, and law have to catch up. Because demography has already decided for us.
And demography, unlike technology, doesn’t negotiate.
Claudio Bresciano is a researcher and author of the Theory of Axiomatic Necessity (TNA), a framework that explores the structural limits of logic, physics, biology, and artificial intelligence. His work on the non-derivability of instantiation in computational systems offers formal tools for the debate on responsibility, consciousness, and the limits of AI. If you enjoyed this article, follow me for more explorations into the hidden architecture of reality.
