The way people are starting to look at art these days is wrong, and dehumanizing. As someone who's been an active experimental artist for over 30 years, I find it repugnant. And I have to get this out.
I've been thinking about this quite a bit lately, especially with the latest controversy in the games industry, where nineteen jurors for the Independent Games Festival announced that they wouldn't vote for games developed using generative AI, regardless of how it was used.
The situation with Big Walk, the latest game from House House, the people who made Untitled Goose Game, is particularly striking.
During development, the team ran into an obscure problem with Unity's rendering pipeline that caused the game to lose roughly half its performance during sunrise and sunset. They spent time investigating it, examined the engine code, and even consulted an engineer at Unity, who couldn't come up with a solution either.
Eventually, someone had the perfectly reasonable idea of asking a chatbot.
The first suggestions didn't work, but eventually it proposed an approach that the developers adapted into a working fix. About ten lines of code, in a project representing years of creative and engineering work. The studio estimates that less than 0.1% of the game's code was affected by AI assistance, and that no generative AI was otherwise used in producing the game.
They disclosed this to the festival, because the festival asked, and now they're apologizing for having done it.
I'm truly fucking gobsmacked by the whole thing.
The thing that gets me is that we've spent decades establishing professional best practices for software engineering, and today those practices include automated QA pipelines, continuous integration, static analysis, and increasingly, agents that help keep repositories manageable. You can decide to write every line of code yourself, or have agents write most of it, but either way the engineering processes that protect your customers from broken software are largely the same.
And I would argue that refusing to take advantage of those processes, when they're available and appropriate, isn't some principled artistic position. It's an actual lapse in professionalism and engineering ethics, particularly if what you claim to care about is delivering the best possible product to your customers.
I shouldn't be surprised by how little some of the people arguing about this understand even the basics of engineering, but I am.
I think maybe I'm too optimistic about other human beings sometimes.
But the truth is…
Nobody Cares About Your Abstract Syntax Trees
Well, almost nobody.
There are a handful of people in the world who take enormous pleasure in examining byte patterns, deciphering ancient encryption schemes, and figuring out precisely how software from forty years ago actually worked.
I follow a few of them.
People like Nathan Baggs are wonderfully rare. Nerdy beacons of light in an industry where almost nobody wants to think about what the compiler did.
Most consumers don't care about those things. They care whether the game is fun, whether it works, and whether they feel something when they play it.
Am I indicting Joe Sixpack for not understanding the nuances of balancing your Assembly trays? Nah.
But it creates a problem when people begin declaring that one method of producing software is artistically legitimate and another isn't.
Because I've always taken issue with the assertion that code isn't art. It clearly is.
But where does that art actually live?
Is it in the abstract syntax tree? The logical operators? The exact sequence of bytes that eventually gets executed by a processor?
Or is it in the decisions made along the way?
Indeed, is there any coherent reasoning happening here at all?
Two Photoshops Walk Into a Bar
I've been doing quite a bit of observational contract reversal lately. It’s been hugely educational, and actually some of the most challenging work of my engineering career.
The idea is that you study how a piece of software behaves, how its interface works, how its tools interact, and what happens when users perform various operations.
Then you build something that reproduces those observable behaviors, without the luxury of looking at the source code. In fact, it’s better, and more satisfying if you do it in an entirely different language.
And there’s art there. In some ways, it feels like forgery more than it feels like plagiarism. It makes my little internal anarchist happy. So the original algorithms are off the table. The decisions embedded in the code are gone. And all you’re left with is the artistic magic of the thing you’re recreating. Imagine it like making a painting from a reference, or restaging a photograph a hundred years later.
It’s even better when you know that the task is so intimidating, that nobody’s ever tried to do it before. Or better, that all who have tried have failed.
Anyone who says you can be an artist or an engineer without hubris is lying to you.
But modern AI agents make this process extraordinarily fast when the operator is skilled and the task is known.
Yesterday, I used a fleet of agents to undertake an observational reconstruction of Vector Craft. Vector Craft is an observational port of Adobe Illustrator, made with a similar method to the one I’m describing. In total, I completed the entire project, reaching full parity with the "original" in 17 hours.
It’s hard to explain the feeling of amazement, and existential dread that rolled over me. I think we need more complicated words in English to describe feelings like that. Shock and Awe is the closest thing I can think of, but it doesn’t quite capture that feeling of impending doom you feel as someone who’s dedicated 30 years to the craft.
I also think it’s interesting that even people who embrace this kind of technology are still grappling with the implications.
Imagine this thought experiment.
You now have two applications.
One is Adobe Photoshop, a product that represents decades of engineering, design, invention, refinement, and accumulated institutional knowledge, often based on real customer feedback over four decades.
The other is a hypothetical observational recreation called Photocraft.
For the purposes of this thought experiment, imagine they're identical.
It’s important to concede that they’re not, in real life. But they’re close enough that for most people they could be. So for the purposes of this mind-bending experiment… imagine they are. That Photocraft is a 1:1 reconstruction, that’s a dead ringer. A perfect counterfeit, and all things are equal in every way.
Every tool works. Every interaction feels the same. Every pixel ends up where it's supposed to be.
You can put screenshots of the two applications side by side, and the most significant difference is the name in the title bar.
One took decades to develop.
The other was created in seventeen hours by some random person with an unusually capable fleet of agents. The people and these systems do exist, today.
Now, here's the question.
Are they the same artistic expression?
The first application has an enormously meaningful history. Its designers solved problems, invented workflows, stole like crazy from their competitors, and made creative decisions that influenced generations of artists.
The second application reproduces those decisions through observation. In fact, it’s even possible that its creator never even understood the original implementation at all.
Does that matter?
Does the history of the first application make it spiritually more meaningful than the second?
And if the experience is identical, where exactly does that additional meaning reside?
In the pixels? The behavior? In what the user can accomplish during the work day?
Are we talking about the aura of art again or the perception of one?
What If the Art Doesn't Change?
There are perfectly reasonable answers to this question.
You could argue that artistic meaning depends on provenance, intention, historical context, or the relationship between the creator and the audience.
I wouldn't dismiss any of those arguments.
But I do think they expose something interesting.
We're no longer discussing the experience itself. We're discussing the story surrounding the experience.
Those are meaningfully different things.
Suppose you watch a ballet performance so emotionally overwhelming that it makes you cry. Ballet has this quality to it, if you’re sitting quietly, and paying attention. Every movement is beautiful. Every gesture feels deliberate. The performance communicates something profound about being alive.
Then imagine your shock, when somebody tells you it was generated by a machine.
Were you wrong to cry?
Okay, now, imagine that another video surfaced. This time, it’s an identical performance, only this time, it’s performed by a human dancer.
There’s a feeling of betrayal there. A kind of heartbeat, and feeling let down, knowing the performance you loved was a machine replica of this beautiful human performance you also loved. The moral calculus is pretty simple here, obviously AI is destroying art, and you’re offended. Not because of the work, but the principle of the thing, and people need to know how you feel!
But when you log in to your social media website of choice to discuss it, you learn that the human dancer saw the beauty of the AI-based performance… and painstakingly copied it, beat per beat.
How do you reconcile that?
If you valued the human performance because it was original, you've got a problem.
If you valued it because a human body performed those movements, that's a different argument, and potentially a perfectly valid one.
If that dance, and those movements, in that order can’t be found anywhere else in the history of ballet, which wouldn’t be surprising the way models like Seedance generalize… what part of the performance is valuable to you?
And if you’ve tried to think out this far, do you remember what drew you here?
The Difficulty of Creation Isn't the Definition of Art
There's something deeply anachronistic about treating the amount of labor required to produce something as a measure of its artistic legitimacy. Like we’ve thrown out the 20th century and lessons learned entirely. Like we no longer care about inclusion, and helping people express themselves.
What we’re seeing feels less like defense of art than it does a corporate consolidation of sanctioned methods of creativity.
And they call me the anti-artist.
I just think it’s funny that we live in a world where a photograph can be made in a fraction of a second, but a painting might take months. You think. Some of these things are non-intuitive when you’re talking about the obsession of it. A carefully orchestrated photograph very well might also take months, while a painting could be completed in seconds. Half the paintings you see are a matter of dumb luck, but nobody will tell. That’s trade secret territory.
There’s never been a universal conversion formula between hours of effort and artistic significance. If there was, Man Ray would have figured it out, while Dolly would have disputed it in the most verbose way possible and Warhol would have said fuck it, and traced it on the wall from a simple projection to prove them both wrong.
I imagine Warhol with an impish laugh.
Maybe that’s it. Maybe all art ever was, was a Campbell’s soup can.
Chances are good that Yayoi Kusama would not have cared.
But when AI enters the conversation, suddenly everyone wants to police the process.
How much did the human do?
Does it matter?
Could they tell the difference if they wanted to?
It’s all so dehumanizing.
Years of work down the toilet for ten lines of code.
So Who Gets to Make Things, Anyway?
I don't think the most important implication of AI-assisted development is that people can make mediocre images or games more quickly.
But rather, I think it's that people can create things they previously lacked the resources to create at all. I think AI critics are wrong, actually very wrong, about who is actually making and consuming this thing we call “AI art.”
To hear them talk about it, it’s all these no-talent fools that have never created anything a day in their lives, and now they want to, falsely, declare themselves “artists!”
They’re half right. There are people like that. Prompt-only wizards with no other skills. Some of which can do absolutely mind-boggling stylized work that has never existed, in some cases, could not have existed anywhere else, at any other time. But it’s been my experience that the community is usually the same experimental crowd that’s been there for decades.
Assuming they were right though, what’s the argument there? Competition from nonconsumption?
Spare me.
Software, animation, games, films, publishing tools. Indeed the entire renaissance of tools for the very traditional artists that speak of creative AI in the same terms as ICE raids.
Shockingly accessible, if you know how to use the things.
I don’t have a finished opinion here, but I lean towards the side of utilitarianism, the Commons, and the greater good. The ability to participate in these industries has historically been constrained by access to money, technology, specialized labor, and institutional support.
I think the art is unquestionably human, and that much is obvious.
Question is: at what point do we stop telling ourselves not to believe our own eyes and ears?