Who Wrote the Book?

I wrote my novel Terminal Autonomy in fifty-eight days. The ideas, characters, story, and prose are 100% mine. It’s an amalgamation of two stories I’ve always wanted to write. The first was about the choices we have to make in EMS that might put us at odds with allies. The second was about the effect automation will have on the industry I’ve dedicated my entire working life to.

It’s my novel, but I never would’ve finished it without the help of AI. I used it for research, criticism, continuity checks, and revision discussions—the same territory where I might otherwise have consulted editors, subject-matter experts, or extremely patient friends (not that I didn’t bother plenty of you anyway).

I worked so much faster with an artificial intelligence sitting beside me 24 hours a day. When I was awake at 3am working on the novel, Claude and ChatGPT were both happily able to read the entire manuscript and tell me what the tone of a conversation was ten chapters back. AI could find the answers in seconds without having to do perfect word searches or fully articulate what I need. I could discuss.

Sometimes, I pasted their answers to each other and let them duke it out. AI fight club can get entertaining.

It also left me curious about what would happen if I moved the boundary.

So I am conducting an experiment.

About fifteen years ago, my children and I invented a story around the dinner table. A group of kids travels back to the age of the dinosaurs and accidentally prevents their extinction. When they return, sixty-five million years of evolution has produced a world inherited by intelligent dinosaurs instead of humans.

The premise existed long before ChatGPT. The original creative spark belongs to my children and me. What never existed was the book.

Now I am building it with AI.

I establish the premise, choose the characters, set the rules, reject ideas, resolve contradictions, and decide what the story means. The AI helps develop the structure, tracks the style sheet, builds the outline, and produces draft prose under my direction. I then do a line-level edit, evaluating what survives and cutting what doesn’t. I make changes and, most importantly, decide whether the finished work feels like something I can honestly put my name on.

Authorship has never been as simple as counting keystrokes. Editors restructure books. Researchers supply facts. Spouses rescue endings. Writing partners trade scenes. Ghostwriters produce entire manuscripts that appear beneath someone else’s name. Every book emerges from a process, even when the cover lists only one author.

Generative AI can participate at almost every level. It can brainstorm, criticize, outline, rewrite, and draft. The uncomfortable reality for me is that AI generates prose that I often like better than my own work. But it isn’t human, and that comes out when you let it run free.

That’s why this experiment. There’s a huge debate going on about what it means to be an author when a machine can write. Those debates aren’t new. They happened when cameras made images only painters could previously make. They happened when word processors made composition easier. They have always happened when technology arrives with a better way to do the work, but the human is always necessary.

That’s the crux of Terminal Autonomy. It’s also the point of this exercise.

I intend to be open about the process. I am keeping records of the conversations, decisions, rejected alternatives, and revisions. I want to see how much of the initial draft remains in the finished book and how much human labor it takes to turn generated prose into something worth reading.

The primary goal is still a good book. An experiment that produces a bad novel is not especially interesting. Neither is using new technology merely because it is new.

But if the book works, the process may be worth examining as closely as the result. If so, I’ll write another book to share how I did it. I already have the title: Who Wrote the Book?

Who supplied the premise? Who established the rules? Who made the consequential choices? Who wrote the sentences? Who decided which sentences deserved to survive?

Who’s writing the book?

I am. Let’s see if using the most modern tool we have makes it better and quicker, or just mediocre.