artists against art
the tools changed. the desire to decide who may create did not.
I keep seeing the same reply whenever somebody makes an image with AI:
It is rude, but the larger problem is that it makes no sense. Drawing is one way to make an image. It is not the admission exam for having an imagination. If somebody wants to make a game, a film, or a picture, why must they first master the instrument preferred by a stranger on the internet?
Most people making this demand have a camera in their pocket. Do they know how to use it? I don't mean whether they can press the button. Have they learned what aperture and shutter speed do, why lens choice matters, or how to work with light instead of merely recording whatever happens to be in front of them? Have they studied Annie Leibovitz, Gordon Parks, Fan Ho, or any other photographer closely enough to understand why their pictures hold attention?
The phone provides access to photography. It does not provide an eye.
This seems obvious with every tool except the one currently threatening somebody's income. We can look up nutrition and training, but information does not make everyone healthy. Investing has never been easier, but no sane person explains poverty by telling somebody to pick up a brokerage account. Programming tools, music software, and university lectures are freely available. We have not all become engineers, composers, and physicists.
Access is only the beginning. Learning to do something well takes part of a life, and no one has enough life to master everything. Most of us build on tools and knowledge made by people who specialized somewhere else. Nearly all human work depends on that division.
“Pick up a pencil” is also needlessly cruel. Some people cannot hold one steadily or spend years training fine motor control. Others lack the time, money, or education demanded by the person issuing the order. If a new interface allows more people to express an idea, then human expression has gained something. Requiring everyone to use the same physical process before their work can count as art gains nothing.
I have spent twenty years making video games. Hundreds were never released, and many were never intended to be sold. I made them because I enjoyed learning how, because I wanted to explore an idea, or because building the thing was satisfying on its own. They earned no money. They still count.
If they do not count, art has been reduced to customer service.
Photography has occupied a similar place in my life. I have shot professionally on sets, in studios, and for marketing work, but those jobs make up perhaps two percent of all the photographs I have taken. The other ninety-eight percent came from interest, practice, memory, and the simple desire to make a good photograph. Nobody asked for most of them. Nobody had to.
I was criticized for shooting only digital. According to the familiar argument, film required more care and therefore produced a more legitimate photographer. I ignored it and kept taking pictures. Film remained beautiful, digital photography kept developing, and the people guarding that particular border eventually moved on to another one.
Getting paid did not turn me into a photographer. Shipping a game did not make the unreleased work real after the fact. Money can confirm that somebody purchased labor; it cannot decide whether the labor was artistic.
That distinction matters because much of the anti-AI argument stops being about art as soon as money appears. There is an honest claim that says, “I need to be paid for my work.” There is a very different claim that says, “You must hire me before you are allowed to make yours.” The first deserves respect. The second is a demand for control over somebody else's ability to create.
When the legitimacy of another person's art ends exactly where your commission is threatened, money is the only value your definition still measures. You may possess extraordinary technical skill. You may have made great art in the past. In that moment you are acting as a member of a guild protecting its market, not as an artist protecting art.
Those are the artists against art.
There are serious arguments to have about how AI is built and who benefits from it. The claim that AI cannot be art is not one of them. If that is the basis of your opposition, you are dead wrong. You have mistaken your preferred method for art itself.
I don't expect every artist to enjoy every medium. I do expect an artist to feel some curiosity when a new one appears. Most people live their entire lives among forms whose basic grammar was settled before they were born. We have the rare opportunity to watch one develop in real time, before its conventions settle and while its best uses are still waiting to be found.
Instead, some artists have decided that the important question is whether every part of an image was touched by an approved hand. I care more about what the person tried to make and whether the result is any good.
No traditional medium has been taken away. The pencil is still there. So are paint, clay, film, wood, and every other material people already love. Sculptors continue working in a world with 3D printers. Painters continued after photography, and photographers continued after digital imaging. New tools give us additional choices; they do not travel backward through history and damage the old ones.
An artist can keep working entirely by hand. A team can make the same choice for a project and enforce it strictly. My own studio works with commissioned and in-house artists and does not use generative AI for final game art. That is how I want those games made.
That isn't hypocrisy. A team rule describes how that team has agreed to work; it does not tell the rest of the world what art is. Anyone who cannot distinguish those two claims has a problem with basic logic, not with AI.
The same distinction governs disclosure. If a client pays for hand-drawn work and you quietly generate it, you cheated the client. If a team bans AI and you hide its use, you broke the agreement. Defending AI as a medium does not excuse lying about how something was made. The problem is the lie, not the existence of the tool.
A few years ago I wrote about the digital shepherd, a programmer who directs AI while remaining responsible for the program. I was thinking too narrowly. The same idea applies to other kinds of creation.
A painter does not make the paint, the brush, or the canvas. A photographer does not design the sensor and grind the lens. Digital artists use programs they did not write. I make games without first building an operating system, compiler, engine, modeling package, and audio workstation. If I had to create every tool behind a game before the game could count as mine, I would never get past the compiler.
Art has never required that kind of self-sufficiency. It requires direction and judgment. The artist decides what belongs, notices what is wrong, changes course, and accepts responsibility for the finished work.
AI gives the tool more freedom than a paintbrush has, but this is a difference of degree. Photographers depend on events they do not control. Directors depend on actors. Game developers build systems because they want unexpected things to happen inside them. Dada artists used chance and manufactured objects more than a century ago, and MoMA describes that surrender of control as a challenge to ordinary ideas of artistic agency. Art survived.
Using AI does not automatically make somebody an artist. A person can type one lazy prompt, accept the first result, and produce something empty. Most AI output is bad. Most creative work made by every other method is bad too. Access has never guaranteed quality.
“AI slop” is a perfectly good name for bad AI work. It is a poor name for the entire medium. If the work is generic, ugly, dishonest, or boring, criticize those failures. Saying that a machine was involved tells me how part of it was made. It tells me nothing about whether it was worth making.
Defending the medium does not require anyone to praise the result. Easier production will give us more disposable work, and criticism should become more demanding. Anyone claiming authorship has to own the failures along with the successes; you cannot take credit when the model surprises you in a good way and blame the machine when it produces something stupid. AI artists do not need protection from judgment. They need to make better art.
The usual technical description is wrong as well. A diffusion model is not generally searching a database for pieces of stolen pictures and pasting them together. Training changes numerical parameters as the model learns relationships in the data. Generation usually begins with noise and develops an image under whatever conditions the creator supplies. Those conditions may include a prompt, sketch, mask, pose, or depth map, followed by inpainting and manual editing. The latent-diffusion and ControlNet papers explain the process better than the stolen-collage story.
Models can memorize training material and sometimes reproduce it too closely. Researchers study this problem because it actually happens. Examine the model and its output. Repeating an inaccurate story about every generated image does not protect anyone.
I have no interest in defending a company that acquires data badly, or a user who makes a cheap imitation of a living artist. Copyright cases will continue sorting through uses that differ in important details. Bad conduct should carry consequences. It still cannot make ethical AI art impossible.
Anna Ridler took and labeled thousands of her own tulip photographs to create Mosaic Virus. Other artists can use licensed, public-domain, or self-created material. Even if a particular commercial model were found unlawful from beginning to end, that judgment would apply to the model and its data. It would not prove that no generative system can be used artistically. The Copyright Office's reports are more careful about these differences than most online arguments.
I also don't need to pretend AI is free to run. It consumes energy and water. We should measure that cost, reduce it, and decide which uses are worth it. The IEA reports rising data-center demand alongside improvements in efficiency. An electricity bill can expose a wasteful practice; it cannot tell us whether an image is art.
Jobs are where the argument becomes serious. Artists are already losing work, and many others expect to. Studies of freelance markets have found falling demand in exposed fields. The 2026 GDC survey found visual artists among the people most hostile to generative AI, while Harvard Business School has summarized evidence of displacement. Their fear is reasonable.
A person can be right about that harm and wrong about the nature of art. When a company uses AI to cut wages, imitate an employee, eliminate jobs, and keep every productivity gain, the company is the proper target. Workers should organize, negotiate stronger contracts, and protect their identities and work. Declaring the output incapable of being art will not recover a single lost dollar.
Now put the tool in an individual creator's hands. A filmmaker can attempt something that once required a funded studio. A game developer can build more of a world before asking a publisher for money. People can complete projects that would otherwise end as explanations of the projects they wished they could afford.
If an idea requires forty people before anyone can see it, then it requires somebody able to pay forty people. That person usually receives ownership, influence, and the power to stop the project. Expensive creation gives capital more authority over which ideas are allowed to exist.
This is where the politics turn backward. Labor has spent generations demanding ownership of what it creates. AI can allow an individual worker to perform more of the production process and retain more of the result. Demanding that creation remain expensive may protect a particular paid task today, but it also protects the system that forces creators to seek permission from investors tomorrow. That is not resistance to capital. It is protection of capital's advantage, even when expressed in the language of solidarity.
The companies building these models are not charities. Large models require money and computing power, and a closed service can change its prices, rules, or availability whenever its owner chooses. The FTC has warned about concentration around data, cloud infrastructure, compute, and foundation models. We should prefer open models, local tools, portability, and real competition. Refusing the technology does not weaken the companies controlling it. It leaves them alone with it.
The artisan is still here. Original paintings have value in a world full of prints. Handmade furniture exists beside mass production, and people attend live performances while carrying flawless recordings of the same music. People care about how things are made. Research has even found a “handmade effect”, where knowledge of the human process adds value to an object.
Artisanal work can compete on whatever quality machines fail to reproduce. When its value comes from the process, artists can document that process and make provenance part of the work. A buyer who sees no difference and cares nothing about the method was purchasing the result. An artist may dislike that fact, but disliking it does not create a right to control every way the result can be made.
Selling skilled labor is honorable, but it does not entitle anyone to charge a mandatory fee for somebody else's expression. Craft does not need compulsory patronage to be valuable.
People are free to avoid AI. They may produce better work without it. A painter who loves painting should continue painting, even if another process becomes faster, because speed is a poor reason to abandon something meaningful. A project built around artisanal work may become more valuable as automated work becomes common.
The choice still has a cost. Generative systems will become ordinary parts of cameras, editors, game engines, animation software, audio tools, and programming environments. People learning them now are developing judgment about when they help and when they make the work worse. Refusing to learn does not stop that process. It delays the day you begin catching up.
An artist can say, “I will not use this.” The artists against art say, “You cannot use this.” That is the whole difference.
Keep the pencil, the brush, the camera, and the clay. Make work by hand when the process matters to you. Set strict rules for your own projects and honor them. Then allow other people the same freedom. The future is beautiful, and it has no reason to punish anyone who refuses it. It will simply be made by the people who did not.