Where Are You Still Making the Decisions?
Map how you work with AI — and see which decisions still shape the work.
AI can research, structure, draft, edit and generate ideas. But using AI isn't the same as giving AI creative authority.
Your decision map
Where AI enters — and where it leads
AI contributes:
AI leads:
Contribution isn't the same as authority.
What your map suggests
Three places to look more closely
What happens if AI gets one more consequential decision?
— Carsten Jan Weichelt
AI can do the work without making the decision.
AI can research a topic, suggest an outline, generate a paragraph, rewrite a sentence or propose ten different directions for a story. None of that tells us, by itself, who is directing the work. That distinction matters to me. I use AI in different parts of my professional and creative work, but I don't think the amount of AI-generated material is a particularly useful measure of authorship. A process can be heavily AI-assisted and still remain human-directed. Another can contain relatively little generated material while handing an important creative decision to the machine. The Human Decision Map by Carsten Jan Weichelt is my attempt to make that difference visible. For free. Just for fun.
Where does AI contribute?
The parts of the process where AI can help produce, explore or transform material.
Contribution isn't the same as authority.
I find the usual distinction between “human-written” and “AI-generated” increasingly unsatisfying. Imagine an AI generates five possible arguments and I reject all five. AI participated in the process, but it didn't determine the direction of the work. Now imagine AI chooses the topic, proposes the argument, establishes the structure, writes the draft and suggests the conclusion. A human makes a few edits and publishes it. Both processes involve a human and AI. But they are not the same. The distinction I find more useful is between AI-assisted and AI-led work.
An AI Writer is a human writer who deliberately integrates AI into the writing process while retaining authorial judgment, creative direction and responsibility for the final work.
Carsten Jan Weichelt · The Human Decision Map
But writing a text is one thing.
What happens when AI helps build an entire world?
Good question, indeed. This question became more interesting to me when I started working with Kuh Alma. Alma is a fictional character and storyworld I develop with generative AI. AI can help me explore scenes, situations and visual possibilities. It can produce things I would never have produced in quite the same way myself. But generated doesn't mean accepted. A new image isn't automatically part of Alma's world. A generated scene doesn't become canon simply because it exists. I still decide who Alma is, what belongs to her world and where that world can go. That distinction eventually led me from thinking about the AI Writer to thinking about something larger: the author-led storyworld.
Generated doesn't mean canon.
Generative AI creates a peculiar problem for storytelling: possibilities become cheap. I can generate another scene. Another setting. Another version of Alma. Another visual interpretation. Another direction the story could take. There is always another possibility waiting behind the next prompt. But a storyworld cannot simply consist of everything that can be generated. Someone still has to decide what belongs. For me, that is where authorship becomes particularly interesting. The creative act is no longer limited to producing material. It also includes rejecting material, protecting continuity, deciding what characters would or wouldn't do and determining which possibilities become part of the world. Generation can become almost unlimited. Canon cannot.
This doesn't mean the author has to create every word, image or idea without assistance. And it doesn't mean using less AI automatically makes a work more author-led. The distinction is about creative authority. AI can suggest a character. It can generate a scene. It can visualize a place that previously existed only in my head. It can surprise me, and sometimes that surprise is exactly why I use it. But it doesn't automatically get to decide what belongs.
In other words: Author-led doesn't describe how much AI is used. It describes who can say no.
The bigger question is authorship.
The Human Decision Map grew out of a question I've been thinking about for a while: what happens to authorship when producing words, images and possibilities is no longer exclusively human? I don't think the useful answer is simply to keep AI out of the creative process. I use it myself. At the same time, I also write fiction without generative AI. What interests me is the space in between. What happens to judgment when generation becomes easy? What happens to voice when a machine can imitate one? And what happens to responsibility when part of the work has been produced by a system that cannot take responsibility for any of it? Perhaps that's why I keep coming back to the same question: Which decisions still make the work yours?
Continue exploring AI, writing and authorship
Carsten Jan Weichelt
Carsten Jan Weichelt is a writer and content strategist exploring how AI changes writing, creative work and authorship. He uses AI extensively in professional and experimental creative work while also writing fiction without generative AI. His work explores what machines can contribute, what humans still decide and what authorship means when the boundary is no longer obvious.
More about Carsten →Want to map your own process?
See where AI contributes — and where you are still making the decisions.
What happens when a writer can suddenly give a fictional world movement, music and film? I spent six months experimenting with Gemini to create a trailer for Kuh Alma. Somewhere between the prompts, failed clips and occasional horse, I started thinking differently about what AI storytelling can mean for a writer.