The New Gatekeepers of Digital Art

Débora Kaz ·

I often see posts from people looking for artists to create covers, illustrations, or visual projects, accompanied by one condition:

“Artificial intelligence cannot be used.”

Anyone commissioning work has the right to choose a process, a language, or a form of production. They may want a painting, a drawing, a photograph, or an illustration made through specific means.

What catches my attention is the comments.

The conversation quickly stops being about a commission and turns into a moral investigation into who works, who deceives, who deserves the name of artist, and who should be sent into some conceptual exile not yet recognized by criticism.

There are real questions involving the use of works without consent, copyright, and the concentrated power of the companies that control these technologies. There is also a barely disguised pleasure in policing the legitimacy of other people’s work.

I follow these discussions with a particular kind of fatigue, because I have already lived through earlier versions of them.

I began with painting. I worked with objects, analog photography, video still recorded on tape, digital photography, and image-making software. Then came modeling, virtual installations, immersive environments, and navigable worlds.

Today, artificial intelligence enters some works, not all of them.

It enters when it helps construct an atmosphere, a body, a spatial condition, or a situation that the concept requires. At other times, it gets in the way. I have begun processes in which AI seemed to expand the work and ended up realizing that it had taken up too much space, while the image grew and the idea diminished.

I do not consider technology an obedient extension of my will. It interferes, suggests, diverts, and offers solutions that may seem much better than they really are.

That is precisely why it requires direction.

My research has never been organized around fidelity to a single medium. It moves between languages, although I am somewhat suspicious of the phrase “the work asked for it.”

Artists often say that the work asked for exactly what they happened to feel like experimenting with. Sometimes it really did. At other times, we find an elegant justification for our curiosity.

There is nothing wrong with that. Research is also born from desire, error, seduction, and accident. It becomes stranger when every choice is presented as though it came from some higher conceptual entity.

What matters is understanding what that decision does to the work and being willing to abandon it when it does not work. At least in theory. In practice, we know how easy it is to become attached to a solution simply because it was difficult, beautiful, or expensive.

A concrete boundary

Resistance to AI did not appear to me only in scattered comments or social-media arguments.

In one of the virtual galleries where I often exhibit, the curator had always made it clear that she opposed the use of artificial intelligence in art.

It was not an ambiguous position. She disliked the technology and did not consider it appropriate for the kind of work she wanted to present.

Even so, she made an exception for my work.

Not because she had come to agree with AI in general. She recognized the continuity of my research, the construction of the concepts, and the way I integrated different digital procedures.

The condition was clear: in the exhibition texts and project communication, I had to state that those works had been produced through a combination of digital art and artificial intelligence.

I accepted.

Transparency did not seem like a punishment to me. It was part of how the work could be read.

But the situation stayed with me.

On one side was a curator who rejected AI as an artistic language. On the other was a body of work she considered consistent enough not to be discarded because that technology was present.

The work entered, but it had to cross a boundary.

This experience does not prove that I was right and the curator was wrong. The episode interests me because a firm opposition became less stable when it had to deal with actual works.

It is relatively easy to reject “AI-made art” as an abstract category.

The classification becomes more complicated when the technology appears in combination with modeling, composition, editing, spatial construction, and an investigation that existed before it.

Nor do I want to turn this episode into a special absolution for good conceptual behavior. That reading would be far too comfortable.

Acceptance depended on trust, trajectory, and recognition. An artist whose research is already known may be given room to experiment with something a beginner might not even be able to present.

The art field has always operated through differences in credibility. AI did not create that mechanism. It merely entered it.

Concept can also function as a social passport. Those who master curatorial vocabulary, have an established trajectory, and circulate within certain circuits can defend choices that someone else may not yet know how to sustain publicly.

That does not make those choices false. It only makes the situation less clean than we would like.

Even so, the demand for transparency seemed fair to me.

The process mattered to the reading. Saying that there was a mixture of digital art and AI avoided both concealment and the reduction of the works to a single resource.

The technology had to be named without allowing it to swallow everything else.

The old accusation with a new interface

Before artificial intelligence, the friction already existed in virtual worlds.

An installation seemed less like an installation because it did not occupy a physical room.

Modeling and programming were treated as though they diminished the artistic presence.

A navigable experience seemed closer to a game than to an artwork.

The question appeared in different forms:

Where is the real work?

This doubt rarely presents itself as prejudice. It usually arrives dressed as criteria, and in part it is criteria.

Virtual environments can also produce empty experiences, technically sophisticated spectacles, and conceptually fragile works. I myself have been dazzled by spatial possibilities that, seen later, did not sustain what they promised.

The recurring pattern, however, was something else: the medium appeared before the work.

Instead of looking at what was being constructed, some people decided in advance how much art could exist in that space.

Digital artists spent years hearing that the computer did everything. Now some of them repeat something very similar about artificial intelligence.

The contradiction exists, although the situations are not the same.

Generative systems were built on enormous datasets, opaque relationships with preexisting works, and corporate structures that cannot be treated as neutral background.

But a work does not cease to exist because the technology participating in it carries problems.

Art has never worked only with innocent materials. It works with institutions, economic circuits, technologies, and forms of power. The presence of those structures does not resolve the work, but neither can they simply be left outside the room.

I do not know whether there is an entirely coherent way to use these technologies today. Perhaps there is not.

What remains is to name the conditions, make choices, revise some of them, and live with a degree of contradiction that does not fit neatly into declarations of purity.

Technique, identity, and the right to remain special

Learning a technique takes time.

Modeling, drawing, editing, lighting, programming, and painting require practice. That knowledge does not become irrelevant because a technology has appeared that can quickly perform part of the process.

There are AI enthusiasts who treat all previous expertise as outdated, as if years of experience were merely an inconvenient phase that innovation has finally eliminated.

I disagree.

Technique expands the capacity to choose. It makes it possible to notice errors, combine procedures, correct results, and recognize when the system is merely offering a seductive image.

The problem appears when technical mastery begins to function as artistic citizenship.

A person may draw extraordinarily well and produce images with no force. They may master modeling and build nothing but demonstrations of skill.

They may also work with very few resources and shift an entire discussion.

Contemporary art has never been a competition organized around the difficulty of execution. That does not diminish the value of technique, but it prevents technique from occupying the entire field.

Part of the resistance to AI seems to arise from the fear of losing a territory built through effort. Work, prestige, identity, and the feeling of possessing knowledge that few people mastered become entangled.

When a new technology lowers certain barriers, it does not distribute competence automatically, but it weakens some visible distinctions.

Someone who spent years reaching a certain result sees another person produce something similar in minutes. It is understandable that this provokes anger.

From there, the risk is turning that anger into a theory about what may or may not be art.

The image may look similar. The capacity to direct the process, the continuity of the research, and the understanding of the medium may not be.

Those differences require time, context, and attention, exactly the kind of things social media treats as an extravagance.

The label “made without AI” then appears as a moral guarantee.

It may represent a consistent ethical choice, inform the process, and matter to someone commissioning work. But it does not automatically transform the work.

A hand-drawn cover can be banal. An analog photograph can be empty. A painting can take months and produce no tension whatsoever.

On the other side, AI users receive an impressive result and conclude that the visual experience already constitutes a body of research.

It does not always.

Technology can provide matter, variation, chance, noise, and conflict. It does not automatically provide continuity, context, or thought.

But the boundary is not simple either.

How many interventions turn a generation into a work?

Does the editing need to be visible?

Should the concept be legible without text?

Does combining AI with other procedures guarantee elaboration?

There is no minimum amount of manual labor capable of settling these questions.

It is possible to combine several techniques and still arrive at a fragile work. The sophistication of the process can also function as a curtain.

Perhaps that is why classification by medium is so attractive. It simplifies something that is not simple.

Used AI.

Did not use AI.

Artist.

Not an artist.

Inside these categories, very little remains to be observed.

What interests me is how the technology participates in the research, what relationship it establishes with the other materials, and how much the work depends only on the impact produced by the system.

Does it produce reflection, or merely consume possibilities?

I cannot always answer quickly. Sometimes not even the artist can.

In the exhibition where I had to declare the combination of digital art and AI, information about the process did not resolve the work. Nor did it invalidate it. It became part of it.

Perhaps this is one possible path, although it is not a general solution.

Name the presence of the technology, assume its implications, and allow the work to continue carrying the contradictions we cannot close.

There are things the technical sheet clarifies.

Others it only begins to complicate.