The Last Hungry One
An essay on civilization at the edge of abundance · May 2026
OpenAI appears to be moving toward a device where the familiar screen of app icons is no longer the center of the interface. The user does not open apps — the agent understands the context, suggests the action, and performs the task.
It sounds like liberation.
And that is exactly why it should make us uneasy.
Civilization has always been built by people who faced resistance: scarcity, pain, competition, inner hunger, the impossibility of simply exiting the game. What happens to a person when there is always a way out — and almost everything can already be done on their behalf?
I. The Icon — How Desire Is Born
Imagine a device that knows what you want before you have managed to formulate it.
It books the table.
Replies to emails.
Buys the tickets.
Chooses the route.
Writes the message.
You do not need to open an app.
You do not need to choose.
You do not even need to know precisely what you want.
There is real value in this promise. Much of digital life is overloaded: dozens of apps, notifications, menus, and unnecessary actions. An agentic interface can remove the noise.
But along with the noise, it may remove something far more important — the moment in which desire is born.
The Instagram icon on your screen is not just a button. It is a visual impulse. It does not wait for you to develop an intention. It provokes one. You do not always decide to open the app — you see the icon, and your hand moves almost by itself.
The entire digital market is built on this mechanism. Marketing does not merely satisfy needs. It creates them. People were not dreaming of food delivery in fifteen minutes until they were shown that it was possible. Nobody missed AirPods before they became a symbol of convenience, status, and a new kind of behavior.
Remove the icons, and many users will not become freer. They will simply lose an external source of impulse.
The agent solves one problem and creates another.
It removes friction between desire and action. But if this process goes further, it may also remove the very moment when a person comes into contact with their own desire.
Advertising will not disappear. It will simply move deeper — from banners and icons into recommendations, suggestions, and “the best option.” An agent bought to work for the user will inevitably enter the field of those who pay to influence the user’s choice.
And this is more dangerous than ordinary advertising.
A banner is visible.
An icon is visible.
But an agent’s recommendation can look like care.
II. Hunger — Why Scarcity Built Civilizations
Not every form of hunger creates culture. More often, hunger creates poverty, violence, and humiliation. To romanticize deprivation is a mistake.
But to deny the other side would also be dishonest: almost every great civilization was built around resistance. People faced a limit they had to overcome. Land that was not enough. Resources that had to be protected. An enemy that had to be resisted. An inner pain that could not be silenced.
The Phoenicians became great traders not because trade was a beautiful idea, but because they had little fertile land and a sea in front of them. Japanese culture of quality did not emerge from an abstract love of perfection, but from life on limited islands where mistakes are expensive. Dostoevsky did not write because it was a convenient career path. He wrote because something inside him was unbearable.
Hunger — literal and metaphorical — was a civilizational engine. Not by itself, but as pressure: the force that makes a person search for form, discipline, mastery, and a way out.
When the washing machine freed women from part of manual domestic labor, it did not lead to mass apathy. A portion of that freed energy flowed into education, professions, science, art, and public life. Liberation worked because there was still a frontier ahead. The world was still hungry for new professions, roles, discoveries, and ideas.
Today the situation is different.
AI is freeing human beings faster than society can create new directions for their energy.
Illustrators, copywriters, translators, and junior developers already feel this not as an abstract fear of the future, but as the disappearance of real work. Yet the economic shock is only the first layer. Beneath it lies a deeper question: motivation.
Why spend ten years learning to draw if an agent can create a “good enough” image in seconds?
Why spend years learning to write if a text can be assembled through a prompt?
Why become a master if the market increasingly rewards not depth, but the speed of packaging?
The objective answer is unpleasant: for most tasks, mastery will become less necessary. It will not disappear entirely, but it will lose much of its mass economic value. Human labor will survive where taste, trust, personal responsibility, cultural status, physical presence, or a unique authorial position matter.
But the “average performer” will come under the strongest pressure.
Skills do not disappear instantly. They atrophy when they stop being necessary. GPS did not destroy the ability to navigate, but it weakened it in everyday life. Autocorrect did not destroy literacy, but it lowered the cost of making mistakes. An agent that takes over more and more cognitive operations may do the same to thinking: not forbid it, but make it optional.
And an optional skill always loses to convenience.
III. The Algorithm — Why Protecting Human Labor Will Be Harder Than Protecting a Market
If states protect local producers with tariffs, it is natural to ask: why not protect human labor from algorithmic replacement?
The question is fair.
The answer is uncomfortable.
A tariff works against a factory in another country. A factory can be localized, taxed, restricted, or blocked from import. An algorithm lives everywhere and nowhere. It cannot be stopped at a border. It can be banned in one jurisdiction, but it cannot be banned as a possibility.
China built a wall against the open internet — and received not the absence of the internet, but a parallel internet. With AI, the same thing will happen faster. Bans will not cancel the technology. They will merely redistribute access: large players, governments, and gray markets will keep it; ordinary people will have less of it.
Attempts at regulation already exist. The EU is building legal barriers. Hollywood unions have negotiated restrictions on AI use. Taxes on automation are being discussed. These efforts are not meaningless. But they are local measures, not an answer to the central question: what exactly does society want to protect?
Protecting “jobs” as bundles of operations is pointless. Operations will be automated. The history of technology has always moved in that direction.
What must be protected is not the old form of labor, but the human role in the chain of value: authorship, responsibility, visibility, and the right not to compete with machines in areas where society has consciously decided to preserve human presence.
For example, we can say honestly: yes, handcraft is less economically efficient. But we preserve it because it is part of culture.
Yes, a human teacher is more expensive than an automated tutor. But at certain ages, a person matters more than efficiency.
Yes, a doctor assisted by AI may be more accurate. But responsibility for the decision must remain human.
This is not an economic choice.
It is a civilizational one.
The problem is that the state itself is trapped in a conflict of interests. It wants to:
- lead the AI race;
- preserve employment;
- collect taxes and investment from corporations that automate labor.
These goals do not fit together well. As long as the geopolitical race dominates, the social consequences will be postponed. Not because politicians understand nothing, but because refusing AI looks like strategic defeat.
So the honest forecast is this: regulation will arrive late. It will soften the consequences, but it will not stop replacement.
Real protection of the human being is possible only where society decides in advance which human roles must not be reduced to efficiency.
So far, no such decision has been made.
IV. The Window — Why a Human With an Idea Is Still Stronger Than an Agent
Right now, we are living through a rare point in history.
AI is already powerful enough for a person without a large team to do what once required developers, designers, copywriters, analysts, and managers. The distance between idea and implementation has shrunk to the smallest point in the history of technology.
But AI still has no stake of its own.
It can generate variants of ideas.
It can suggest hypotheses.
It can write code, text, strategy, and presentations.
But it has no biography, no pain, no risk, no reputation, no intuition paid for with years of mistakes. It does not choose a direction for its life. It does not know why this specific idea must exist.
That is why a human being with real vision is still stronger than an agent without inner necessity.
This window is small. A year, two, perhaps three — not because AI will then “become human,” but because agents will become increasingly capable of formulating hypotheses, testing them, building products, and optimizing results with minimal human involvement.
Then the advantage will shift.
It will not be enough to simply have an idea. There will be too many ideas. Value will move from generation to position: audience, trust, distribution, reputation, access to real problems, and the ability to make decisions under risk.
Code can be copied.
An interface can be repeated.
Content can be generated.
Position cannot.
Humanity has created, for the first time, a tool smarter than itself in most practical tasks. And it still has not agreed on why the human being is needed inside that system.
Rome knew abundance. Bread and circuses held the population in a state of satisfied indifference. It was not the only reason for the empire’s decline, but it remains a powerful symbol: a society that loses the need to be strong gradually loses the ability to defend the complexity on which it stands.
Abundance by itself is not freedom.
Freedom requires form.
Form requires meaning.
Meaning requires inner hunger.
AI can remove routine from human life. That is a blessing. But if, together with routine, we lose the need to want, choose, fail, learn, and become masters, liberation will turn into soft atrophy.
The only way to avoid this is to preserve the one inside us who still knows what they want.
The one who can formulate the task alone.
The one who does not wait for the agent to guess.
The one who uses AI as an instrument, not as a replacement for will.
The window is open.
It is small.
And this may be the last moment when hunger is still an advantage.