The AI and the Era of the New Enlightened — Part V
AIfication: a reflection on knowledge, work, and freedom, and on who decides how the benefits of Artificial Intelligence are shared.
AIfication
Did you write this, or did an Artificial Intelligence write it?
Until very recently, asking that question would have seemed strange. Today it has become almost an automatic reaction to a text that is too polished, an illustration that is too perfect, a presentation that is too orderly, or even an idea expressed with a clarity we are beginning to find suspicious.
The question, however, is starting to feel insufficient to me.
Because the real problem does not begin when an Artificial Intelligence writes for us. It begins when we repeat what it wrote without having traveled the path necessary to understand it and, even so, end up convinced that those ideas belong to us.
We let ourselves be captivated by eloquence and the subtlety of words. We ask something and receive a perfectly articulated explanation, delivered with a confidence few human conversations possess. The arguments seem to unfold naturally, contradictions disappear, and the conclusion lands before us with such clarity that we immediately think: exactly, that is what I wanted to say.
Perhaps it is.
But perhaps it is not.
Perhaps that was not exactly what we thought, but what we had just learned to think.
I do not mean that an Artificial Intelligence thinks like us. If there is something I have tried to defend throughout these essays, it is precisely the opposite. A language model does not experience words the way we do. It does not remember the smell of its childhood home when it writes about nostalgia, does not fear the possibility of losing its job, and does not contemplate a painting and recognize something that reminds it of someone who is no longer here.
But we do not think like an Artificial Intelligence either.
The problem arises elsewhere: in how easily we confuse immediate access to an explanation with the acquisition of knowledge.
Today someone can ask an AI who Karl Marx was and, a few minutes later, discuss surplus value; they can request an explanation of Arthur C. Danto and speak with absolute certainty about the end of art; they can ask for an interpretation of a historical event, a philosophical position, or an economic theory without ever having read the author they are citing, without having encountered that author’s contradictions, and without even having had the sometimes uncomfortable experience of discovering that the author was not actually saying what they had expected.
The possibility of appropriating a result without going through the intellectual process that produced it is new in its scale, though not in its nature.
There is not necessarily any bad intention. In a sense, those answers would not exist in exactly that form without us either. We asked the question, provided context, chose what to keep, rejected other answers, and finally decided to incorporate certain ideas into our view of the world. There is real human participation.
But participating in the generation of an answer does not necessarily mean understanding it.
Having an idea in front of us does not mean we have thought it through, just as owning a book does not mean we have read it.
This tension between what we see and what actually happened behind the object inevitably reminds me of a discussion that apparently has nothing to do with Artificial Intelligence.
In 1964, Arthur C. Danto attended an Andy Warhol exhibition at the Stable Gallery in New York and encountered the now famous Brillo Boxes. Warhol had produced objects that could appear virtually indistinguishable from the commercial Brillo boxes anyone might find outside the gallery. For Danto, this posed an extraordinary philosophical problem: if two objects can share practically all their perceptible characteristics, and yet one can be considered a work of art while the other remains a simple commodity, then something fundamental had changed in the way we recognize art.
Looking was no longer enough.
The object’s visible properties were no longer sufficient to determine what was art. To understand why one box could belong in a supermarket and another in the history of art, we needed to know its context, its interpretation, its intention, and the cultural and historical framework within which it appeared. Out of problems like this, Danto developed the notion of the artworld: a world of theories, history, and interpretation that allows certain objects to be understood as art (Danto, 1964).
Years later, he took this idea even further in his thesis on the “end of art.” Danto was not claiming that artists would stop painting, sculpting, or creating. Nor was he announcing the death of art. What he considered finished was a particular historical narrative: one in which the history of art could be understood as a succession of movements advancing toward some identifiable destination. Once practically any object could become art depending on its meaning, context, and interpretation, no visual property remained that could tell us what the next step necessarily had to be. Art entered, in Danto’s account, a posthistorical condition in which many possibilities could coexist without one having to historically defeat its predecessor (Danto, 1997, 1998).
Artificial Intelligence is beginning to confront us with a strangely similar problem, but this time the subject is not art. It is thought.
Two texts can be formally indistinguishable. Both can be well written, use correct references, construct interesting arguments, and reach apparently profound conclusions. Without knowing their history, we might be unable to determine whether one represents months of reading, doubts, experiences, contradictions, and personal reflection, while the other appeared after thirty seconds and a good prompt.
The properties of the object are no longer enough.
But there is an unsettling difference from Danto. In art, that rupture opened up enormous creative freedom. In knowledge, we may be entering an era in which the intellectual product survives while the intellectual process that originally gave it value slowly begins to disappear.
We can produce essays without ever really testing an idea, research without researching, arguments without working through the contradictions necessary to construct them, and even programs that work without fully understanding why they work. Production can become increasingly detached from the intellectual experience we once assumed lay behind it.
Then the question “did you write it, or did an AI?” takes on a different depth. Perhaps what matters is no longer who pressed the keys, but who went through the intellectual process that the result represents.
Because if that process disappears, we are not merely automating writing. We are beginning to automate part of the path through which we develop judgment.
For lack of a better word, I have begun calling this transformation AIfication.
I do not simply mean using Artificial Intelligence. I use it constantly myself, and I believe we are facing one of the most extraordinary intellectual tools we have ever built. The AIfication that concerns me is something much broader: the progressive reorganization of how we learn, work, produce, consume, and even desire around systems whose economic objectives, owners, and incentives we rarely question.
To understand why this matters, we have to leave Silicon Valley for a moment and go back almost a hundred and sixty years, to the industrial factory Karl Marx observed in the nineteenth century.
The machine that was supposed to free us
There is a reason I am particularly interested in the chapter of Capital devoted to machinery and large-scale industry. Depending on the edition, its numbering may differ—Chapter XIII in some editions and XV in others—but what matters is not the number. It is the question Marx asks there.
What actually happens when a machine dramatically increases human productivity?
At first glance, the answer seems obvious. If a machine allows us to manufacture in four hours what previously required eight, humanity has just won four hours of time.
That is the technological interpretation.
Marx introduces another question.
For whom?
His argument is not simply that machines destroy jobs. That would be a poor reading of Marx and, moreover, historically insufficient. Marx recognizes machinery’s extraordinary ability to increase productivity. His concern lies much deeper: the machine does not enter an abstract society, but a particular set of relations of production.
Within capitalist production, machinery is not introduced primarily to ease the worker’s effort. It is introduced because it reduces the necessary labor embodied in commodities, increases productivity, and expands the production of surplus value. Marx describes machinery as a means of producing relative surplus value and analyzes how its introduction can lengthen the working day, intensify labor, and displace workers precisely because technology is being used within a relationship whose objective is the valorization of capital (Marx, 1867/1990).
This difference is fundamental.
A technology contains possibilities.
Whoever controls the technology decides which of those possibilities are economically attractive.
Imagine that yesterday someone needed eight hours to complete a particular task, and today, thanks to an Artificial Intelligence, they need only four. From a technical standpoint, we have just produced four hours of potential freedom.
But we still do not know who owns those four hours.
They could belong to the worker. We could preserve their salary and shorten their working day. They could spend that time with their family, studying, resting, walking, painting, reading, or simply producing nothing at all.
They could also belong to the owner of the company.
In that case, those four hours would not appear as free time. They would appear as available productive capacity. The worker would finish the first task and receive a second. Then a third. Targets would rise because we now know that technically they can. Yesterday’s extraordinary achievement would become tomorrow’s minimum expectation.
The same technology allows both futures.
The machine does not choose.
Someone chooses.
And this is where I think many contemporary discussions of Artificial Intelligence begin to become deliberately naive. We speak of capitalism as though it were an atmospheric phenomenon. We say “the market demands,” “the company needs,” “efficiency requires,” or “competition forces,” and by repeating these expressions so often, we end up erasing the people who actually make decisions.
Incentives exist. Competition exists. Companies that do not generate profits can disappear. None of that means every decision is inevitable.
There are owners, boards of directors, shareholders, and executives who choose how to distribute the benefits, which investments to retain, what margins they consider sufficient, how much to pay their executives, how much to return to shareholders, which offices to keep, and also how many people to lay off.
Laying off workers is not a law of physics.
It is a business decision.
And when a decision is repeated often enough, we stop recognizing it as a decision and begin calling it efficiency.
Marx observed this contradiction during industrialization. What could technically shorten the working day could end up intensifying it. The machine that could physically relieve the worker could turn them into a component subordinated to the rhythm of the machinery. He even described the inversion through which it no longer seemed that the worker was using the instrument of labor, but that the instrument was using the worker.
We do not have to stretch the analogy very far to recognize something similar in our own time.
Today we use assistants capable of programming, designing, summarizing, writing, analyzing, and automating tasks that once took hours. At first, we feel that productivity belongs to us. We finish earlier. We are better.
Until the organization discovers our new pace.
Then that pace becomes the benchmark.
And we begin competing against our own enhanced productivity.
The billionaires’ utopia
Bill Gates has spoken publicly of a future in which the productivity gains brought about by Artificial Intelligence could reduce the working week to just two days. Elon Musk has gone further still: he imagines a future of abundance in which AI and robotics make work optional and eventually even diminish the importance of money.
I confess I would love them both to be right.
A civilization capable of producing everything it needs using a fraction of human labor would, in principle, be one of the greatest achievements in our history. If five days of work can become two without reducing what we produce, it would be absurd to keep working five simply because we have been doing so for generations.
But there is something profoundly hypocritical about hearing this promise from some of the very men who best embody contemporary technological capitalism.
Musk can publicly imagine a world where working is optional and, at the same time, as a chief executive, order massive staff reductions when he considers it beneficial to his companies. In 2024, Tesla eliminated more than 10% of its global workforce in a restructuring Musk justified as necessary to reduce costs and prepare the company for its next phase of growth.
Bill Gates no longer runs Microsoft, and it would be wrong to hold him personally responsible for the company’s current employment decisions. But Microsoft, the company from which much of his wealth emerged and of which he remains an inseparable historical figure, perfectly represents the contradiction between that utopia and the decisions of actual technological capitalism. In 2025, it was reported that the company had achieved more than $500 million in savings through AI applications in its call centers alone, while cutting thousands of jobs and investing tens of billions of dollars in Artificial Intelligence infrastructure. In July 2026, it also announced another 4,800 layoffs as part of a new reorganization. Not all those jobs can be attributed directly to replacement by AI, and doing so would be dishonest. But the facts show something more important: increasing productivity, saving costs, and investing aggressively in automation does not automatically lead to less working time for those who remain at the company.
We do not even need to limit ourselves to Microsoft. During 2026, Meta attempted a much more explicit internal transformation: its so-called Project OT sought to turn parts of the organization into smaller teams extensively supported by AI systems, even contemplating very deep reductions in certain areas. The project ended up encountering problems with productivity, reliability, and internal resistance, and the company itself had to scale it back. The episode is especially revealing because it shows how some executives are not passively waiting to discover what AI can do: they are actively trying to reorganize the relationship between technology and workers around it.
This is the point that interests me.
Business owners are not innocent passengers within capitalism.
They lead it, make decisions within it, and receive an extraordinarily large share of the benefits when those decisions work.
When a company needs to improve its margins, it can cut workers, but it could also reduce dividends, share buybacks, executive compensation, facilities, certain projects, extraordinarily expensive offices, or simply accept lower profitability.
I am not saying that all these alternatives always have the same financial impact. That would be absurd. I am saying something much simpler: which cost is considered expendable is also a political and moral decision within the company.
And it is revealing that the worker’s physical presence, their salary, or even their existence within the organization can quickly be subjected to the logic of efficiency while many other forms of accumulation are treated as untouchable.
When Marx analyzes machinery, he is not blaming the gear.
He is looking at who owns it.
That is why the fundamental question about Artificial Intelligence cannot be limited to how much work it will be able to perform. The truly political question is what those who own that productivity will do once they finally have it.
We do not need Gates or Musk to explain what will be technologically possible.
We need to know what they are willing to give up when that possibility arrives.
If someone can produce in two days what they previously produced in five, will they preserve that person’s salary and give the other three days back? Will they accept lower margins? Will they share the additional productivity with workers? Or will they discover that one person capable of doing the work of three is, above all, an excellent reason to hire only one?
That is where the utopia begins to crack.
Because perhaps Gates is technically right and Marx remains politically right.
Perhaps two days will become enough to produce five days’ work.
That does not mean the other three days will belong to us.
Having the tool does not mean being free
It is precisely here that another economist becomes fundamental to this discussion.
In Development as Freedom, Amartya Sen proposes moving away from an overly limited idea of development. It is not enough to ask how many resources a society possesses or how much money a person has; we must ask what effective freedoms they actually have to turn those resources into a life they have reason to value.
This is the foundation of his capability approach.
Possessing a resource does not automatically mean being able to use it in the same way as someone else. Economic, social, political, physical, and cultural conditions determine how we can turn what we have into real opportunities. Sen therefore distinguishes possession of resources from the substantive freedom to achieve certain functionings and ways of life (Sen, 1999).
I find this distinction extraordinarily useful for thinking about Artificial Intelligence.
Let us imagine a programmer and the company they work for once again. Both have access to the same model. Both can use it to produce more code, analyze problems, write documentation, and automate tasks.
On the surface, technology has democratized a capability.
But the freedoms each possesses around that capability are completely different.
The programmer can use AI to finish their work earlier.
The company decides what “earlier” means.
It can turn it into an afternoon off.
It can turn it into another task.
It can turn it into higher targets.
It can discover that it now needs fewer programmers.
It can economically appropriate practically all the additional productivity without the person who generated it receiving more time, higher pay, or greater security.
The tool is the same.
The freedom is not.
This makes the idea that democratizing access to AI automatically means democratizing its benefits look naive.
Tomorrow, we could give the world’s most powerful model to every human being free of charge and still live within economic structures that prevent millions from turning that capability into a freer life.
We can have a machine capable of teaching practically any discipline and people who do not have time to study.
We can automate vast amounts of work and maintain a society where losing a job means losing the ability to pay for housing, food, or medical care.
We can produce food with extraordinary efficiency and still have hunger.
The existence of the resource does not, on its own, resolve the distribution of capability.
Hunger was precisely one of the fundamental fields of Sen’s economic research: a society can have food available and yet some people may lack the effective ability to access it. His analysis of freedoms and capabilities forces us to distinguish material existence from real access.
That distinction should haunt us every time someone promises that Artificial Intelligence will produce abundance.
Abundance for whom?
Freedom for whom?
Free time for whom?
Are you using AI, or is AI using you?
So far, AIfication might seem to be a problem concerned only with work.
I do not think it is.
There is a perhaps even deeper transformation taking place in our relationship with these machines, because we are not just workers who use Artificial Intelligence.
We are consumers too.
And perhaps we are becoming something else.
For decades, the digital economy has tried to learn what we want. Google learned what we search for. Amazon learned what we buy. Netflix learned what we watch and what we abandon. Facebook, Instagram, TikTok, and the rest of the social platforms learned what keeps our eyes on a screen.
We turned an enormous part of our behavior into economically valuable information.
But conversational AI introduces something different.
It does not just observe what we do.
It listens to us explain why we want to do it.
We tell it how much money we have and what we want to buy. We explain our professional doubts. We ask for advice about a relationship. We tell it what worries us. We show it how we write, which arguments we reject, and which ones end up convincing us. We present our plans, our problems, and increasingly our decisions before making them.
Traditional advertising always wanted to discover what we desired.
Now we are building systems to which we voluntarily explain why we desire something.
This does not mean that every answer an AI produces today is a secret advertising operation. It is not, and turning this discussion into a conspiracy theory would destroy precisely the question I consider important.
My concern lies in the future of the economic model.
Today, many AI companies earn revenue through subscriptions, enterprise contracts, APIs, cloud services, and different levels of access. But it would be extraordinarily naive to think that these platforms’ business models will remain frozen forever.
The internet has already taught us otherwise.
So we must ask what will happen when hundreds of millions of people consult an AI first before buying a car, purchasing insurance, choosing a university, investing money, accepting a job, choosing a vacation, or deciding which product they need.
What will happen when these systems stop being merely extraordinary search engines and become our advisers?
Until now, advertising needed to interrupt us.
The ad appeared between us and what we wanted to do.
An adviser occupies a completely different position.
Imagine telling an Artificial Intelligence for weeks about our financial problems, our family’s needs, our journeys, and what we are looking for in a car. Finally, we ask what we should buy, and it replies: “Based on what you have told me, I think this is the option that best suits you.”
Psychologically, that sentence does not arrive as an advertisement.
It arrives as judgment.
And then a question emerges that we will have to learn to ask before it is too late:
When will we know that AI is advising us, and when will it start selling us something?
The question is not simply about imagining advertising inserted into ChatGPT, Claude, or any particular product. The problem is much broader: we are creating a new interface between ourselves and the world, and whoever controls that interface will have an extraordinary ability to decide what information appears, which options are presented first, which alternatives seem reasonable, and which arguments accompany each recommendation.
For decades, the battle was over our attention.
The next battle could be over our trust.
And trust is worth much more.
Here the question takes a form I deliberately want to leave uncomfortable:
Are you using Artificial Intelligence, or is Artificial Intelligence using you?
Because the economic relationship is becoming difficult to describe.
We are workers because we use these systems to produce.
We are consumers because we pay directly or indirectly to access them.
We are users because we build our digital lives around their services.
And our interactions can, depending on the product, policies, permissions, and particular business model, also become useful information for improving systems, understanding behavior, or building new products.
We work with the machine, consume through the machine, and can end up economically feeding the ecosystem that builds the machine.
At what point do we stop being exclusively customers?
At what point do we also start being the product, the worker, or the raw material?
Technology is not neutral
I often hear that technology is neutral and everything depends on how we choose to use it.
The phrase is reassuring because it separates the tool from the world that produces it.
I also find it profoundly naive.
A technology of this scale does not spontaneously appear on a table, waiting for humanity to decide what to do with it. It needs extraordinary amounts of capital, data centers, electricity, minerals, chips, intellectual property, telecommunications infrastructure, governments capable of regulating it, companies capable of commercializing it, and workers capable of building it.
Even before we write our first prompt, there is already an economic architecture around the machine.
Technology is not born outside society.
It is born within relations of power.
That does not mean a language model has a secret ideology or an algorithm possesses political will. It means its design, access, financing, ownership, and application are shaped by human decisions.
That is why simply asking whether AI “is good” or “is bad” seems almost childish.
We have to ask who owns it, who decides its objectives, who can modify it, who receives the economic benefits it produces, who absorbs its costs, and what it needs from us to remain profitable.
A hammer does not recommend which nail to buy. A steam engine does not spend hours talking with us about our problems. A calculator does not learn what kind of explanation ultimately convinces us.
A conversational Artificial Intelligence can occupy a much more intimate place because it begins to mediate our relationship with the world.
It mediates our access to knowledge.
It mediates our work.
It can mediate our consumption.
It can mediate our decisions.
And eventually, it could participate in the construction of our desires.
This last point seems especially important to me because we have learned to use the word freedom in an extraordinarily impoverished way.
We consider someone free if they can choose.
As long as there are enough products, enough platforms, enough brands, and enough possibilities, we assume there is freedom.
But it is foolish to think human needs cannot be instrumentalized to generate consumption. Hunger can become a market, loneliness can become a market, insecurity can become a market, fear can become a market, and our need for recognition can become a market.
It is also naive to think corporate monopolies offer us freedom simply because their catalogs are varied.
Having a hundred options does not necessarily mean controlling the conditions under which we choose.
And here I want to go even a little beyond Sen.
His approach helps us understand that possessing a resource does not mean having the effective freedom to turn it into what we value. But another question remains, one I find especially urgent in the face of systems designed to know us better and better:
Where does what we value come from?
Human beings have never constructed their desires in absolute independence. We are shaped by our families, our friends, the books we read, the cities we inhabit, our experiences, our culture, our economic conditions, and also the advertising we consume.
There is no perfectly pure “self” hidden beneath all those influences.
But not all forms of influence are equal.
There is a difference between growing up within a culture and interacting daily with a commercial system capable of learning from millions of interactions which argument works best, which tone generates trust, and what information we need before making a decision.
That is why freedom is not only about pursuing what we want.
Freedom also lies in discovering how much of that desire belongs to us and how much was imposed on us.
And suddenly we have returned to the beginning of this essay.
To an answer we read and, after it has been expressed with sufficient clarity, begin to recognize as our own.
“That was exactly what I was thinking.”
Perhaps.
Or perhaps we have just witnessed the precise instant in which we began to think it.
AIfication
That is what I am trying to describe when I speak of AIfication.
It is not using ChatGPT. It is not asking Claude to review code. It is not generating an image, automating a tedious task, or using a model to research more quickly. Reducing the problem to that would turn a social transformation into a discussion about tools.
AIfication appears when we begin to reorganize our values around what the machine makes possible.
It appears when a worker seems slow to us because they still need to think before producing.
When we confuse twenty generated documents with twenty understood documents.
When we begin to measure a person’s worth exclusively by how much production they can extract from a tool.
When knowledge begins to be evaluated by the answer rather than by the ability to construct the question.
When a company receives an extraordinary increase in productivity and its owners’ first question is not how much time they can give back to people, but how many people they can eliminate.
When an executive uses Artificial Intelligence to justify a reduction in workers and then promises us that, at some indefinite point in the future, that same technology will finally allow us to work less.
When the worker uses a tool to become more productive until that enhanced productivity becomes the new minimum expected of them.
In the nineteenth century, Marx observed how the worker could end up subordinated to the industrial machine.
Our version may be much subtler.
We will not need to be physically chained to a factory.
It will be enough for us to accept permanent competition against the speed of what we ourselves built.
And perhaps that is where one of the deepest ironies of our time lies.
For centuries, we dreamed of building machines capable of working for us.
Now that they are beginning to do so, we are afraid they can work without us.
Not because there is anything inevitably evil within technology, but because we built a society in which working became the necessary condition for accessing almost everything we consider freedom.
We work to have a home.
We work to eat.
We work to access healthcare.
We work to support our families.
We even work so that we can rest.
Then we introduce a machine capable of freeing us from an enormous share of work into a system that requires us to work for the right to live.
It should not surprise us that what technically looks like emancipation produces fear socially.
The contradiction is not inside the language model.
It is inside the society that decided what to do with it.
The last oracle
I began this series by talking about the Oracle of Delphi.
It took me nearly three months to write these five texts. During that time, the ideas came and went again and again as I tried to distill them into five relatively simple ideas. While I hesitated over a sentence, discarded a reference, or tried to understand where I wanted to take this series, more than a dozen new models presented as smarter, faster, or more capable appeared before us. Business decisions of enormous consequence were also announced, making some of the richest men on the planet even richer, while more than a hundred thousand jobs have been publicly linked to cuts associated with Artificial Intelligence, automation, efficiency, or the restructurings accompanying this new technological race.
The speed of all of it contrasts strangely with the time it took me to try to understand it.
Across these texts, I wrote about oracles, about temples reused by new religions, about the printing press and the machines that transformed our relationship with knowledge, about the Industrial Revolution, and about workers who discovered that a machine capable of making their work easier could also be used to increase their exploitation. I ended up going further back still, to the end of the Bronze Age and those extraordinarily complex societies that seemed permanent until they were not.
Perhaps from the beginning I had some idea of where I wanted to go when I decided to call this series The Era of the New Enlightened.
Because after these months, one thing is especially clear to me: the world changes.
It always has.
Societies change, economies change, technologies change, and so do the ideas we use to justify what we are building. There is no guarantee that this change will move in the fairest, most ethical, or even most humane direction. History is full of extraordinarily productive transformations whose benefits were captured by a few while their costs were distributed among many.
But it changes.
And we are here.
We are workers, consumers, developers, artists, business owners, students, parents, children, and citizens observing a transformation we do not yet know how to name completely.
We are spectators of it.
Although perhaps one of the great mistakes of our time is believing that we are only spectators.
I was interested in that deeply human need to hear an answer, find meaning in it, and attribute to the oracle an understanding perhaps greater than it actually possessed.
Five essays later, we are still standing before an oracle.
Only this one no longer speaks through priestesses and ambiguous phrases.
It answers immediately.
It speaks practically all our languages. It can explain physics, program an application, produce a painting, summarize Marx, discuss Danto, teach us economics, and talk with us at three in the morning when no one else is available.
It is extraordinary.
Precisely for that reason, we must resist the temptation to turn it into something it is not.
It is not a new god.
It is not a digital human mind.
But neither is it simply a neutral tool waiting for our orders in some space separate from economics, politics, and society.
It is a technology built within our world.
And perhaps the reason I was so interested in writing these five essays is that Artificial Intelligence ended up revealing far less about machines than I expected and far more about us.
Danto showed us that a point comes when observing an object is no longer enough, and we need to understand the context that gives it meaning.
Marx taught us that technically understanding a machine is not enough either if we ignore who owns it and what economic relations determine its use.
Sen reminded us that having a capability or a resource does not mean possessing the real freedom to turn it into a life we have reason to value.
Artificial Intelligence forces us to confront all three questions at once.
We can have access to an extraordinary share of human knowledge and understand less and less of what we repeat.
We can produce five days’ work in two and keep working five.
We can generate material abundance without ensuring that those who produce it have access to it.
We can choose among millions of possibilities and slowly lose the ability to recognize where some of our desires came from.
We can build the most powerful intellectual tool in our history and discover that possessing it does not automatically make us freer.
That is why perhaps the great question about Artificial Intelligence should never have been whether it would someday become human.
Perhaps we should ask what kind of human beings we are willing to become around it.
Are you using AI, or is AI using you?
I do not think the answer has been written yet.
And that is precisely why the question is worth continuing to ask.
The true promise of Artificial Intelligence should not consist of producing more presentations, more documents, more code, more images, or higher margins with fewer workers. If we really are facing a technology capable of multiplying our productivity as never before, then it should be capable of giving something back to us.
Time.
Knowledge.
Capability.
Freedom.
Technology is already beginning to show that it can produce those possibilities. What no machine can decide for us is who will have the right to claim them.
If we fail, perhaps Bill Gates will be right from a technical standpoint, and one day two days will be enough to produce what once required a week.
And perhaps Marx will remain right politically and economically: the fact that a machine can save us labor does not mean that time will be handed to the worker.
Perhaps Musk is right and one day we will be capable of producing such abundance that working is no longer materially necessary.
But that does not mean those who control that abundance will spontaneously relinquish the power that controlling it gives them.
We will be able to ask a machine anything.
It will be able to advise us.
It will be able to write with us.
It will be able to help us build ideas we would never have built alone.
It will even be able to help us imagine a different society.
But no Artificial Intelligence can assume, on our behalf, the responsibility of deciding who owns the machines, who receives what they produce, what our time is worth, what we consider knowledge, and what kind of freedom we are willing to demand.
At the end of the era of the new enlightened, perhaps we will discover that we never needed to fear machines learning to think like us.
Perhaps the real danger was that we would stop asking who was thinking, who was deciding, and who was winning while they did so.
And those are still human questions.
Precisely for that reason, they are the ones we should delegate least.
References
Danto, A. C. (1964). The artworld. The Journal of Philosophy, 61(19), 571–584.
Danto, A. C. (1997). After the end of art: Contemporary art and the pale of history. Princeton University Press.
Danto, A. C. (1998). The end of art: A philosophical defense. History and Theory, 37(4), 127–143. https://doi.org/10.1111/0018-2656.721998072
Fore, P. (2025, March 27). Bill Gates says a 2-day work week is coming in just 10 years, thanks to AI replacing humans “for most things.” Fortune.
Marx, K. (1990). Capital: A critique of political economy: Volume I (B. Fowkes, Trans.). Penguin Books. (Original work published 1867).
Reuters. (2024, April 15). Tesla laying off more than 10% of staff globally as sales fall. Reuters.
Reuters. (2025, July 9). Microsoft racks up over $500 million in AI savings while slashing jobs, Bloomberg News reports. Reuters.
Reuters. (2026, July 6). Microsoft to cut 4,800 jobs, overhaul Xbox unit. Reuters.
Reuters. (2026, August 26). How Meta’s AI workforce transformation plans went kaput. Reuters.
Rogelberg, S. (2026, January 19). Elon Musk says that in 10 to 20 years, work will be optional and money will be irrelevant thanks to AI and robotics. Fortune.
Sen, A. (1999). Development as freedom. Alfred A. Knopf.