For a long time, creation was a rare activity, not because ideas were scarce, but because turning them into works required resources that were difficult to assemble. Writing a book took months or years. Composing and recording music required instruments, technical mastery and often a studio. Producing a film mobilized teams and considerable capital. Developing software meant learning languages whose syntax remained inaccessible to most people.

Creation was therefore surrounded by barriers. Some were economic, others technical, educational or institutional. Together, they limited the number of people capable of producing something that could realistically be published, distributed or commercialized.

Generative artificial intelligence is now lowering many of those barriers at the same time.

A few instructions can produce a text, an illustration, a voice, a melody, a video or lines of code. The results remain uneven, human intervention is often decisive, and professional expertise retains considerable value. But the essential change lies elsewhere: for the first time, the basic ability to produce intellectual or creative content is becoming accessible to a potentially immense population.

What is gradually disappearing, then, is not human creation.

It is its privilege.

And when production ceases to be scarce, the entire economics of creation begins to shift.

The Machine That Lowers the Cost of Entry

Technological revolutions have democratized creation before. The printing press reduced the cost of reproducing texts. Photography made it possible to produce images without knowing how to paint. Sound recording separated music from live performance. Personal computers gave individuals access to tools once reserved for organizations. The internet eliminated much of the cost of distribution.

Generative AI introduces another rupture: it reduces the cost of production itself.

This matters even more as models become cheaper and more accessible. The Stanford AI Index 2025 estimates that the inference cost of a model performing at roughly GPT-3.5 level on the MMLU benchmark fell by more than 280-fold between November 2022 and October 2024. Over the same period, open-weight models narrowed part of the performance gap separating them from the leading proprietary systems.

The consequences extend far beyond the artificial intelligence industry.

When an entrepreneur can develop a visual identity without an agency, when a musician can orchestrate a demo without a studio, when a researcher can accelerate the exploration of scientific literature, or when a developer can accomplish in hours what previously required days of programming, some of the costs that structured creative industries begin to disappear.

This does not merely transform professionals. It changes who can realistically become one.

Millions of people now have access to a creative infrastructure that would once have required several professions. A single individual can write, translate, illustrate, program, edit video, synthesize documents and distribute the result globally from the same computer.

The individual is acquiring capabilities that once belonged to the organization.

When Everyone Can Produce

This democratization carries an obvious promise. Skills once separated from ordinary users by years of training are becoming partially accessible through natural language. Someone who knows precisely what they want to achieve can cross certain technical barriers without fully mastering the tools behind the result.

But abundance immediately creates its own problem.

If ten times as many people can write, ten times as many texts can be published. If image production becomes almost free, the number of available images can become practically infinite. If generating music, video or software requires progressively fewer resources, supply can increase much faster than the human capacity to consume it.

The internet had already created an economy of information abundance. AI could turn it into an economy of permanent overproduction.

The central problem of creation then changes.

It is no longer simply a question of who can create.

It becomes a question of who will be noticed.

That distinction is fundamental. Throughout much of cultural history, the difficulty of producing something acted as a filter in itself. Writing, publishing, recording, software development and audiovisual production mechanically selected only a fraction of those who wanted to participate.

Those filters were neither perfectly fair nor necessarily desirable. They excluded enormous amounts of talent because of insufficient capital, connections, education or institutional access.

But they limited supply.

AI weakens those filters without solving the problem of demand.

A human day will still contain twenty-four hours, regardless of how much content a machine can produce during it.

Scarcity Moves Elsewhere

This may be the most important economic transformation.

When production becomes abundant, value migrates toward whatever remains scarce.

Attention is scarce. Trust is scarce. Reputation is scarce. Taste is scarce. The ability to select is scarce. A loyal audience is scarce. A brand capable of making a work instantly recognizable among thousands of alternatives becomes more valuable.

Creation does not disappear; its value chain is reorganized.

In a world saturated with texts, the ability to write competently may become less differentiating than knowing what deserves to be written. In a world saturated with technically flawless images, artistic direction may matter more than execution. When millions of songs can be generated, the value of an artist may increasingly reside in identity, history, performance and the community surrounding the work.

AI can automate the production of a proposition more easily than it can manufacture the reason someone should care about it.

That distinction is decisive.

Value may progressively migrate from execution toward intention.

The Paradox of the Augmented Creator

This helps explain why framing the transformation as a simple competition between humans and machines fails to capture what is happening.

The International Labour Organization estimates that roughly one in four workers worldwide is employed in an occupation with some degree of exposure to generative AI. Yet its analysis also concludes that transformation of occupations is, at this stage, more likely than outright replacement. Advances in text, image, voice and video generation have particularly increased the exposure of several occupations in media and web-related industries.

Within media and culture, the ILO identifies the same shift. Journalists, writers and translators are among the professions with significant exposure, while critical judgment, creativity and ethical reasoning may become more important as machines take on a greater share of execution.

The creator does not necessarily disappear behind the tool.

The creator can become its director.

A photographer is no longer defined solely by mastery of the camera. A writer may no longer be the person who formulates every first sentence unaided. A developer can spend less time writing elementary functions and more time designing the architecture of a system. A filmmaker can experiment with sequences that would once have required an inaccessible budget.

Competence gradually shifts from the ability to execute toward the ability to direct, select, correct and ultimately take responsibility for the result.

That does not mean technical expertise becomes irrelevant. The opposite may occur: as tools become more powerful, the gap between ordinary and expert use may become more visible.

Giving two people the same piano does not make them equally accomplished pianists.

Giving them the same generative model does not give them the same judgment either.

Originality After the Machine

A more uncomfortable question nevertheless emerges: what does it mean to be original when generation becomes instantaneous?

Cultural production has always been partly combinatorial. Writers inherit literary forms. Musicians work within harmonic traditions. Filmmakers borrow visual conventions. Scientists build upon the work of their predecessors.

AI industrializes this logic.

It can absorb structures, identify regularities and generate variations at a speed unavailable to human beings. This increasingly forces societies to clarify what they actually consider a work of authorship and what they intend to protect.

In January 2025, the U.S. Copyright Office reaffirmed that using AI does not in itself prevent a work from receiving copyright protection. However, AI-generated material is protectable only where sufficient human contribution determines the relevant expressive elements; merely providing instructions to a system does not necessarily meet that threshold.

Behind the legal issue lies a much broader cultural question.

For a long time, we have partially confused the value of creation with the difficulty required to produce it.

AI is gradually separating the two.

A beautiful image can be generated in seconds. An important investigative article may require six months of work. A technically polished song can be produced almost instantly. An imperfect work may embody twenty years of human experience.

Production cost can therefore no longer determine value.

Other criteria will have to emerge.

The Economy of Trust

This is where the transformation may become paradoxical.

Artificial intelligence democratizes production, yet that democratization could strengthen certain intermediaries.

When the volume of available content becomes enormous, nobody can navigate it directly. Filters become indispensable.

Search engines, social networks, video platforms, streaming services, app stores, marketplaces and AI assistants become the infrastructures responsible for organizing abundance.

Power no longer necessarily lies in the ability to produce.

It lies in the ability to distribute.

A world containing a billion creators is not automatically a world in which a billion creators receive equal visibility. On the contrary, as supply expands, the mechanisms organizing visibility become increasingly strategic.

AI could therefore produce two apparently contradictory movements at once: a spectacular decentralization of creation and a growing concentration of the power of selection.

Everyone may eventually be able to create.

A small number of systems may nevertheless decide what everyone sees.

New Professions, New Hierarchies

Economic history also suggests that technologies which destroy the value of certain skills can create value elsewhere.

In its Future of Jobs Report 2025, the World Economic Forum estimates that technological, demographic, economic and geopolitical transformations could affect 22% of existing jobs by 2030. Its survey of more than 1,000 employers projects 170 million new roles alongside 92 million displaced ones, for a theoretical net gain of 78 million jobs. These figures should be treated as estimates rather than predictions, but they illustrate the scale of the expected restructuring. The same report anticipates rapid growth in AI-related skills while continuing to identify creative thinking as an important human capability.

The boundary between creator and non-creator may itself become less meaningful.

A lawyer produces machine-assisted documents. An engineer generates simulations. A teacher creates educational materials. An entrepreneur designs interfaces. A scientist can automate part of the documentary research process. Professions that did not traditionally define themselves as creative are becoming large-scale producers of content.

AI therefore does not merely transform creative industries.

It extends the logic of creation across a growing share of the economy.

The Privilege That Remains

It would nevertheless be misleading to conclude that creativity has been democratized in any absolute sense.

The most advanced models require enormous infrastructure. The data, semiconductors, computing centers and capital required to develop them remain highly concentrated. In 2024, global private investment in generative AI reached $33.9 billion. The United States alone accounted for $109.1 billion in private AI investment more broadly, far ahead of China and the United Kingdom.

Creation is therefore becoming democratized at the level of the user while its infrastructure can simultaneously become concentrated at the industrial level.

This may be one of the central paradoxes of the AI revolution.

Never have so many individuals possessed such extensive creative capabilities.

And rarely have those capabilities depended upon such a small number of technological infrastructures.

After Abundance

For a long time, we lived in a world in which the act of creating was already a form of distinction.

Writing a book was remarkable because relatively few people could carry the undertaking to completion. Producing a film was exceptional because it required substantial resources. Building software demanded specialized knowledge. Recording an album required infrastructure.

Those obstacles will not disappear entirely. But they are declining enough to alter the economic meaning of creation itself.

In the world now emerging, producing something may become the easiest part.

The challenge will be producing something worth keeping.

The difference is immense.

Because if artificial intelligence gradually ends the privilege of creating, it does not end the privilege of having something to say.

It may make that privilege more valuable.

In a civilization capable of generating almost unlimited quantities of text, images, sounds and video, the ultimate scarcity may no longer be creation.

It may be meaning.

Main Sources

Stanford Institute for Human-Centered Artificial Intelligence, AI Index Report 2025 — AI performance, inference costs, investment and economic trends.

International Labour Organization, Generative AI and Jobs: A Refined Global Index of Occupational Exposure, 2025.

International Labour Organization, Generative AI and the Media and Culture Industry, 2025.

U.S. Copyright Office, Copyright and Artificial Intelligence — Part 2: Copyrightability, 2025.

World Economic Forum, Future of Jobs Report 2025.