Who is the capital?

Every age builds an image of humanity consistent with its own productive system.

Agricultural civilisation saw the human being as guardian of the land and of the natural cycle. Industrial civilisation turned them into labour force, a measurable element of a larger machine. Consumer society described them as a desiring individual. The digital era converted them into user, profile, datum, source of attention.

Artificial intelligence now forces us to a more radical question: what will remain of the human being when a growing part of their capabilities can be reproduced, accelerated or simulated?

For centuries material capital was relatively easy to recognise: land, buildings, machinery, money, infrastructure. Today it also includes semiconductors, computing power, energy, networks, platforms and immense quantities of data. It is a capital ever less visible and, for that very reason, ever more pervasive.

Human capital, by contrast, seems more fragile. Knowledge, experience, responsibility, creativity, memory, reputation and the capacity for judgement belong to people, but they can be extracted, codified and embedded in technological systems.

When an experienced worker unknowingly trains an algorithm through their daily activities, what is happening? Are they using a tool, or transferring part of their knowledge to the capital that employs them?

The question has no simple answer.

Every tool contains human knowledge. A lathe, a formula, a piece of software or a protocol are forms of crystallised experience. Humanity progresses precisely because one generation embeds part of its knowledge in the tools it leaves to the next. Artificial intelligence continues this process, but it takes it to a new threshold: the tool no longer preserves merely a gesture; it seems to take part in the production of new answers.

The machine no longer needs to be programmed in every single behaviour. It can learn structures, imitate languages, generate images, propose strategies and modify its output based on context. It does not necessarily possess consciousness, intention or understanding in the human sense. And yet it produces effects that, until recently, we associated exclusively with intelligence.

From this the dilemma arises.

A society centred on material capital will use artificial intelligence above all to compress costs, replace labour, monitor performance and concentrate power. The worker will be regarded as a residual element: necessary only where the machine is not yet economical or reliable enough.

A society centred on human capital will instead use automation to increase people's capacity, reduce degrading activities, distribute knowledge and free up time for more complex responsibilities. Technology will not eliminate work, but it may make it more human.

The difference will not be determined by the machine. It will be determined by ownership relations, by institutions and by the criteria with which we define value.

The greatest risk is not that artificial intelligence becomes human. It is that the human being agrees to become similar to the reduced image the algorithm has of them: a sequence of predictable behaviours, manipulable preferences and measurable performances.

When every activity is turned into an indicator, what is not measurable tends to disappear. Prudence may look like slowness. Care may look like inefficiency. Tacit experience may look like resistance to change. Doubt may be interpreted as an inability to decide.

And yet it is precisely these imperfect qualities that make judgement human.

An algorithm can optimise a decision relative to an objective. It cannot autonomously establish whether that objective is worth pursuing. It can indicate the most efficient means, but the choice of the end still belongs to politics, ethics and conscience.

Material capital answers the question: "How much can we produce?"

Human capital must answer the question: "Why should we produce it?"

There is no need to set human being and technology romantically against each other. The human being without material capital remains poor, fragile and dependent on toil. Material capital without human capital becomes, instead, a power without direction. We need both, but not on the same level.

Material capital must remain a tool. Human capital must remain the subject that chooses the objectives, assigns responsibility and sets the limits.

To succeed we will have to rethink work. We will not be able to keep tying dignity, income and identity exclusively to the number of hours during which a person performs economically measurable tasks. If machines produce a growing part of wealth, we will have to decide who owns that productivity and how to distribute its benefits.

The question of artificial intelligence is therefore also a question of ownership.

Who owns the model? Who owns the data it was trained on? Who receives the value generated by the collective knowledge embedded in the machine? Who answers when the system errs? Who has the right to switch it off?

These are economic questions, but first of all metaphysical ones. They concern what we consider a person, an action, a responsibility and an end.

The future will not necessarily see the victory of the machine over the human being. It might see something more ambiguous: people who own machines and people owned by systems built by others.

The central conflict will not be between natural and artificial.

It will be between those who can choose and those who can only adapt.

In the end, the most important question will not be whether artificial intelligence manages to think like us.

It will be: will we still be free to decide what to think for?