This is a series about writing books with AI, and most of it covers what the tool does badly and how to fix it. So I’m starting with the opposite, because it’s true and hardly anybody in the argument will say it plainly: AI is a useful tool for writing books, and used well it saves a working author a lot of time and a lot of misery. I use it, and I wrote this series partly to defend using it well. I won’t apologize for that, and you shouldn’t have to either.
The trick is being precise about what it saves, because the hype and the backlash have both buried the answer. The hype says it writes your book. It won’t, and the rest of this series shows why. The backlash says it only produces hollow garbage and no real writer would touch it. That’s wrong too, and it costs honest authors real time. I’ve got no patience for either camp. AI is bad at writing your book and very good at the jobs around the writing, the ones that eat your hours and your will.
Does AI really kill the blank page?
The most expensive moment in writing a book is the one before there are any words. The blank screen, the cursor blinking, the chapter you know you have to write and cannot make yourself start. Every writer knows that moment and most of us lose hours and sometimes days to it. It is not laziness. Starting from nothing is genuinely hard, and the hardness is not the part of writing that makes the book good.
This is the thing AI is best at, and it is not close. A rough, flawed, imperfect first draft of a scene, generated in thirty seconds, is a different psychological object than an empty page. You are no longer creating from nothing. You are reacting, cutting, fixing, arguing with something already on the page, and reacting is enormously easier than originating. The draft can be mediocre and still do its whole job, because its job was never to be good. Its job was to end the blank page, and it ends it instantly.
For a lot of authors, that alone is enough. I’d much rather a writer argue with a mediocre machine draft than lose a week to an empty chapter, because the week is gone for good and the draft costs nothing to throw away. A bad draft you can fix beats a blank screen you can’t start, every time.
AI generates faster than you can type
You react faster than you originate, and the machine originates faster than you type. Put those together and the whole cycle of writing, draft, read, react, revise, speeds up, because the slow step was never your editing, it was waiting on yourself to produce the raw material to edit.
This matters most when you know roughly what a passage needs to do but do not relish the doing. The connective tissue between two big scenes. The third paragraph of physical description you have written a hundred times in a hundred books. The recap of where things stand at the start of a chapter. This is the tedious, necessary, unglamorous prose that every book is half made of, and the machine drafts it in seconds for you to shape. You spend your real attention on the scenes that deserve it and let the machine rough in the rest.
That’s the prose I’d hand off without a second thought. Nobody ever bought a book for the recap at the top of chapter twelve, and a writer who spends a morning hand-building one has spent that morning away from the scene readers will remember.
AI makes experimenting cheap
Here is a quieter advantage that I think makes books better, not just faster. Experimentation in writing is normally expensive. To see whether a scene works better in first person, or opening on the argument instead of the arrival, or from the other character’s point of view, you have to write it, which costs an hour or an afternoon, so mostly you do not bother. You commit to your first instinct because testing the alternative is too costly, and your first instinct is not always right.
The machine drops the cost of trying to near zero. Show me this scene three ways. Now I open on the argument. Now I am in his head instead of hers. You read the three, and usually one of them shows you something your first instinct missed, and the scene gets better, not because the machine wrote it well but because cheap experiment let you find the right shape before you committed your real writing to it. The machine is a fast way to be wrong cheaply, and being wrong cheaply is how you get to right.
AI is a tireless second pair of hands
It’ll summarize your own earlier chapters when you’ve lost the thread of your own book. It’ll give you ten chapter titles and won’t get tired or hurt when you reject all ten. It’ll draft the synopsis and the back-cover copy you dread. None of that is the real writing, but all of it is work, and it drains the energy you’d rather spend on the writing. Offloading it is no more cheating than using a word processor was, and I don’t understand writers who treat it as a moral failing.
What is the catch with using AI for books?
Notice what every one of these has in common. They are all advantages at the drafting layer, the layer of raw material and momentum and tedious necessary prose. Not one of them is a claim that the machine makes your book better. It does not. It makes the work faster and less miserable, and it does that by being excellent at exactly the parts of writing you least want to do yourself.
And the parts it’s good at come bundled with parts it’s bad at, which is the catch. Left alone, the same machine that ends your blank page fills it with its own defaults: the inflated words, the tame scenes, the safe redemption arc, the facts it quietly invents. It takes back in revision the time it saved up front. Treat it like a tool and you’ll do fine. Let it decide what your book is and you’ll hand readers a book nobody made a real decision about, and I’ve got little sympathy for that.
The series comes down to that, and it needs saying before any of the criticism that follows: the criticism is in service of the upside. Everything after this article, the off-ness to name, the eight layers of tells, the prompting craft, the full method, exists so you can keep the real advantages of this tool without paying the hidden costs. Start with the next piece, why AI books feel wrong, which names the catch precisely, and the rest of the series teaches you to beat it.
