Where Does the Art Live?

Games are a useful medium for thinking about this because the relationship between design space and authorship is so clear.

A game is, in a fundamental sense, a set of rules. The rules define what's possible, what's rewarding, what's frustrating, what's fun. If you're designing a game and you don't know the rules, you didn't design the game. Someone else did. Or no one did, and you just have a collection of mechanics that happen to exist together.

For a puzzle game like mine, the rules are codified in the programming. The logic is the game. When a player feels that moment of clarity where a mechanic suddenly clicks, that moment was written into the code. It exists because of specific decisions about how systems interact. If I outsource that programming, I'm not outsourcing a task. I'm outsourcing the creation of the game itself.

But consider a walking simulator. Same medium, completely different design space. A walking simulator lives on screen. The story, the atmosphere, the visuals, the audio. The game is what you see and hear and feel as you move through the world. The code is infrastructure. It makes the experience possible, but it's not where the design decisions live.

I would argue you're not designing a puzzle game if you outsource the programming. And you are designing a walking simulator if you outsource the programming. The medium is the same. The authorship question is completely different. Because the design space is different.

This is what I kept coming back to in my first essay: the design space drives how AI can practically be involved in the process. The medium and the design space directly correlate to what it means to be the author of the thing.

Here's a distinction I've been thinking about.

A trade is when you use a tool. A craft is when you use a tool and your mind. Art is when you use a tool, your mind, and your intent.

Each layer builds on the last. A trade is mechanical. You learn the tool, you apply the tool. A craft adds understanding. You know why the tool works, when to use it, how to adapt it. Art adds intent. You're not just making something. You're making something for a reason. You have a purpose. You have something at stake.

AI operates comfortably at the trade level. Give it a tool, it will use the tool. It arguably operates at the craft level in some domains. It can understand patterns, adapt to context, make decisions that reflect something like judgment.

But intent is different. Intent requires a why. And a why requires having something at stake. AI doesn't have stakes. It has outputs.

When I write code by hand for my puzzle game, the practical thing I'm protecting is understanding. The design space of this game is the logic. The rules. The way systems interact. If I let AI generate most of that code, I might get something that works. But I lose the ability to reason about it. I lose the ability to look at a mechanic and know, intuitively, what happens if I change one variable. I lose the ability to expand on what I've built because I no longer fully understand what I've built.

For a game where the entire point is the elegance of the logic, that's not a tradeoff I can make. The code isn't infrastructure. The code is the product. Understanding the code is understanding the game.

This is the practical argument for doing things by hand: when the design space of what you're making lives in the artifact itself, you can't outsource the making without outsourcing the understanding. And if you outsource the understanding, you no longer own the design space. You're not the author anymore.

But there's another layer to this. The intent layer.

In the first essay, I drew a distinction: AI operates in the design space of what. Humans operate in the design space of why. AI can generate outputs. It can generate infinite outputs. But it can't supply the reason the outputs exist.

When I write code by hand, I'm not just protecting my understanding. I'm maintaining the connection between what I'm making and why I'm making it. The code is an expression of my intent, not a translation of it. Every decision in the system reflects something I was trying to do. That connection is what makes the work mine.

If I let AI drive, I might still get a working game. But the artifact becomes detached from its source. It's a what without a why. It might function. It might even be good. But it's not mine in the same way.

This is what I mean by art. Art isn't a category of work. It's a mode of working. You can be an artist in any domain if you're bringing all three layers: the tool, the understanding, and the intent. The intent is what makes it yours.

The walking simulator designer who outsources the programming but pours themselves into the world, the story, the atmosphere? They're doing art. Their intent lives in the design space they occupy. The puzzle game designer who outsources the programming? They've handed over the part where the art lives. They're not doing art anymore. They're commissioning it.

Here's the bigger question, and it's one I don't think we've fully reckoned with yet. What happens when AI becomes competent in every design space?

Right now, my argument rests on the idea that there are design spaces where human understanding still matters. Where the work requires reasoning that AI can't reliably do. But that line is moving fast.

If AI becomes competent everywhere, and we let it drive everything, something strange happens. Outputs become infinite. Anything can be generated. A story, a game, a song, an image, a system. The what becomes abundant to the point of meaninglessness.

When that happens, the value of the output collapses. Not because the outputs are bad. They might be great. But because infinite supply destroys scarcity. And when the what is no longer scarce, the only thing that differentiates is the why.

Who made it. Why they made it. What they were trying to say. What they had at stake. A story that makes you cry hits different when you know a person wrote it from lived experience. The tears might be the same if an AI wrote it, but the meaning isn't. Meaning lives in the connection between the artifact and the intent behind it. Sever that connection, and you have output without art.

This is why I still write code by hand. Not because AI can't do it. In many cases, it can. Not because I'm protecting my job—I use AI everyday at work. I do it because the design space of what I'm making demands it. Understanding the system is the system. If I don't know the rules, I didn't design the game. And I do it because I want the work to be mine. I want the artifact to carry my intent, not just reflect my prompts.

Doing things by hand is a choice about what kind of work you want to do and what you want that work to mean.

The question isn't whether AI creates something of quality. The question is whether you're making something, or whether something is being made and you're just nearby.

Where does the art live? It lives in the intent. And intent is something you have to bring yourself.