Sunny Greenhill – How to Write a Books with ChatGPT (страница 5)

A beginning author often associates authorship with "wrote it by hand." But in reality, authorship is not hands. It is the mind. It is architecture.

There are novels that were written in collaboration by several people: one built the plot, another handled the dialogue, a third polished the style. And the book still has authorship and rights – because it contains human decisions.

With AI, it is the same, except that instead of a human assistant you have a tool. Which means your task is to make sure the book contains your architectural decisions.

And here is the main principle I want you to remember:

creative contribution is where you choose, shape, and change the result in terms of meaning, not cosmetics.

You are not merely correcting mistakes. You are building.

Where the Contribution Is Visible "Inside the Text"

There are several things that always reveal the author, even if that author worked with AI.

The first is structure.

The order of chapters, the dramatic construction, the pace, where you place the turn, where you give a pause, where you close a scene, where you open a new semantic block. AI can suggest an excellent structure, but it rarely guesses yours. And if it does, it is usually because you yourself are already holding it very clearly in advance.

The second is the choice of focus.

AI often writes "about everything." A human being chooses "what matters most." For example: a scene can be about the weather, the interior, movement, emotion, conflict… And the author decides: what is this scene really about?

If you know how to hold the focus, that is already authorship. That is a decision, not generation.

The third is voice.

Not style as in "beautifully" or "dryly," but precisely voice: how you think in the text, how you look at the world, what rhythm your phrases have, how you place emphasis, how you measure explanation and omission.

AI can imitate. But voice appears when you align the text to yourself and stop accepting it "as is."

The fourth is semantic knots.

These are the places where the text "holds": scene endings, transitions, the hero's key decisions, the thought that remains after the chapter. These places are always authorial. Even if AI suggested a variant, you choose the one that corresponds to your book.

The fifth is selection.

You may be surprised, but selection is one of the strongest forms of authorship. Because a book is not the sum of paragraphs. A book is what remains after you throw out everything unnecessary.

AI almost never knows how to throw things out precisely. It tends to "add." An author knows how to cut.

And this is where a beginning author's stereotype usually breaks down: "But I am just choosing…"

Yes, you are choosing. And that can be creative work – if the choice is not mechanical, but semantic.

Why "Minimal Editing" Almost Never Works

I see the same mistake again and again.

A person takes a generated text, changes 5-10 words, slightly rearranges the phrases, removes a repetition, fixes the punctuation… and believes that "they reworked it."

But the problem is that this kind of reworking remains cosmetic. It does not change the architecture. It does not create voice. It does not make the text "yours" at the level of decisions.

If you want your contribution to be obvious, the edits must not be small touches, but intervention in the fabric of the text:

restructure the scene;

change the point of view;

intensify the conflict;

remove an unnecessary layer;

replace explanation with action;

change the rhythm so that the scene can breathe.

You do not have to do this on every page. But in every chapter there should be places where it is clear: this is not a stream of generation – this is a book built by an author.

Where the Contribution Is Visible "In the Process"

Now an important part: sometimes the contribution is visible not only in the text, but also in how you work. And, by the way, that can become decisive if someone needs to be convinced that you really are the author.

I would call this the "trace of authorial work."

If you run the process in such a way that you preserve:

an early outline;

several scene variants;

your edits;

different versions of the chapter;

notes on "why I changed this";

discarded fragments,

then you create provability that is useful not only in a conflict. It is useful to you as well: you begin to see that the book did not "arrive" – it was "made."

And there is one simple technique I advise all beginners to use: at the end of each chapter, keep a short note (for yourself only) on what decision you made.

For example:

"I made the chapter shorter because the pace needs to accelerate;"

"I removed the explanation and left the action;"

"I changed the final line so that it would expand the meaning."

This is the discipline of an author, not an operator.

A Practical Criterion: Can You Explain Why It Is This Way

There is a good "maturity test" for authorship.

If someone asks you:

why does the scene begin exactly this way?

why does the hero say exactly this?

why does the chapter end with this line?

why is there a pause here and acceleration there?

…and you can answer not with "that is how it generated," but with "I did it this way because…" – then you really did work as an author.

This is, in fact, the main internal marker. If you yourself cannot explain why the text is the way it is, then the text is controlling you, rather than you controlling the text.

A Model for How to Make the Contribution Obvious in Practice: "Draft – Editing – Voice"

I will give you a working model that almost always leads to a stable result:

First, you let AI help with the draft – quickly, roughly, as with "raw material."

Then you switch into editing mode: you cut, rearrange, assemble the structure, and strengthen the semantic knots.