You, the Machine, and the Blank Page
- Mar 17
- 13 min read
Updated: Jul 4
First, a disclaimer: I love writing my books and I'm passionate about making sure every imperfect word is mine, and mine alone. So while I'm not living under a rock (I have to use AI a LOT at work), I do fall into the camp that feels a bit icky about letting AI create art for us. But I want you to decide how you feel about it for yourself!
Now, let's skip the part where I immediately wag my finger at you and start flinging around terms like "brainrot." You've heard it all before and you're bored by it.
Here's a more interesting question: there's a tool sitting right there that can spit out a whole story in ten seconds. So what do you actually want to do with it? Not what you're allowed to do. What kind of writer do you want to be on the other side of that choice?
That's the real thing to figure out, and nobody can figure it out for you: not your teacher, not your parents, and definitely not the AI. But thankfully, I'm not your parent or your teacher! I just get to be the (hopefully) cool author friend who is ready to talk to you about this stuff without an agenda.
But first! The AI landscape is changing SO FAST. So check the date on this post and verify facts before you make any decisions.
Okay, so let's take a walk with your cool author friend and talk through it: the cool stuff AI can do for your writing, the stuff you should probably keep for yourself, the parts that are a bigger deal than they look, and how a bunch of real writers and artists are landing all over the map on this.
Then you get to decide where you land.
First: your imagination is the whole point
A screenwriting professor named Frank Deese once put it like this: letting AI do your writing for you is a bit like paying a robot to do your workout at the gym. The robot gets stronger. You just watched.
That's the trap, and despite what some of your teachers might be shouting, it's not actually a moral one.
For me, the whole reason to write is that writing is the thing where my specific brain gets to exist on a page in a way no one else's ever could. It's the weird metaphor only I might think of. The character who talks like my best friend. The ending that made me cry while I typed it. If I hand that part to a machine, I didn't save time. I gave away the one thing that was mine.
So the honest north star for everything below is this: use AI for the scaffolding, not the soul. Let it help you build the ladder, but make sure you're the one climbing it.
What AI is actually good at (that won't cost you your voice)
Used as a thinking partner instead of a ghostwriter, AI is genuinely useful. Here's where it earns its keep:
Breaking a blank-page freeze. When you've got nothing, it's a spitballing partner that never gets tired of your "what if." And please, don't take its first suggestion! Let it throw twenty bad ideas at you so you can find the one good one hiding underneath.
Untangling structure. If your plot is a knot, AI is decent at helping you see the shape of it. It can help point out where the middle sags, which beat is missing, and why the ending feels rushed.
Interviewing your own characters and world. Ask it to ask you questions about your character's backstory or your world's rules. You do the answering. Seriously, I do not hear enough about this "AI hack." Having it tirelessly ask you questions about your ideas, your plot - it can be life changing! Suddenly you know things about your story you didn't know you knew.
Getting unstuck. Not "write this scene," but "here's my scene, ask me three questions that might make it sharper." It's that questions trick again, and it works.
Boring research legwork, with a huge, red asterisk. AI makes up facts constantly and says them with total confidence. Never trust it on a real-world detail (a date, a law, how a gun works, what year a war ended) without checking a real source. Treat every "fact" it gives you as a rumor.
Real life sidebar: I mentioned that I have to use AI a lot for work, and for my day job, I build websites that utilize a lot of different databases. The AI I was using confidently told me that a database's rows run vertically, while its columns run horizontally. Reader, that is 1000% backwards and 1000% false. When I pushed back, it doubled down. I had to push back again using references from the internet, and it finally admitted its mistake.
And here's the line worth drawing in big block letters and probably in permanent marker:
Things you should protect and keep for yourself when it comes to working with AI, especially if you want to become a writer:
The actual sentences.
The voice.
The jokes.
The emotional beats (the moments that are supposed to land that you're hopefully basing on real-world experiences and feelings!)
The choices only you would make.
Prompts that keep you in the driver's seat
And guess what? You can set ground rules with the AI up front, in plain language, and it'll mostly stick to them. Here are some copy-paste starters. Tweak them to sound like you.
The ground-rules opener (paste this at the very start of a project):
I'm a writer and I'm writing this story in my own words. I do NOT want you to write prose, dialogue, or full scenes for me. Not ever, unless I specifically ask. Your job is to be a brainstorming and structure partner: ask me questions, offer options I can accept or reject, and point out problems. When I share my writing, don't rewrite it. Instead, ask me questions that help me make it better myself. If I ever slip and ask you to write something for me, remind me of this rule first.
The brainstorm (ideas, not answers):
Give me 15 quick "what if" premises for a [genre] story about [topic]. Keep them one line each. Don't develop any of them. I want raw sparks to react to, and I'll pick.
The outline / story beats (a skeleton you'll flesh out yourself):
Help me sketch a rough beat outline for my story. Here's my premise, my main character, and how I want it to end: [paste]. Suggest the major turning points as a bulleted skeleton only. No prose, no dialogue. I'll write every actual word myself.
The character interviewer:
Interview my main character by asking me one question at a time, like a journalist digging for the real story. I'll answer as the character. Don't answer for me and don't summarize, just keep asking good questions.
The unstuck button:
Here's a scene I'm stuck on: [paste your own draft]. Don't rewrite it. Ask me 3–5 questions that might help me figure out what it's missing.
Notice the pattern: every one of these makes the AI a coach, not a substitute. You're still the one doing the reps. 💪
Where your words actually go
Most of today's AI tools got smart by being fed an unbelievable amount of writing and art, a huge chunk of it scraped off the internet without asking the people who made it, and without paying them.
Real authors have found their own books inside these systems. In 2025, one AI company agreed to pay around $1.5 billion to settle a lawsuit from authors whose books were used to train its model.
This is why a lot of writers and artists are genuinely angry, and it's worth understanding before you feed a machine your own work. Because when you chat with an AI, your own words can become training food for the next version, too, unless you turn that off.
Here's roughly where the setting lives right now (note! companies love to move these around, so if the menu looks different, search "[tool name] turn off training"):
ChatGPT: Settings > Data Controls > switch off "Improve the model for everyone." Its Temporary Chat mode also isn't used for training.
Google Gemini: This one's tied to your whole Google account. Look for Gemini Apps Activity (sometimes called "Keep Activity") and turn it off.
Claude: Profile > Settings > the privacy/training toggle. Its Incognito chats aren't used for training.
There are a couple of catches:
Opting out only protects future chats. Anything already used to train a model that's already out there can't be pulled back. So the earlier you flip these switches, the better.
If you're not logged in, your stuff may get collected regardless of the setting. Log in, then opt out.
Do this on every tool you use, and you keep owning what you make. It's also worth checking periodically to make sure the setting stays off.
When the book is really yours: copyright + being honest about it
Say you actually finish the thing. Here's how ownership works, minus the boring and confusing legalese.
You already own it. In the U.S., your writing is under copyright the moment you write it down (computer file, paper & pencil, toilet paper & crayon - it doesn't matter, it's yours!) You don't have to do anything for that basic protection.
Registering gives you more muscle. If you ever want to seriously protect a finished work (like before you publish), you register it with the U.S. Copyright Office at copyright.gov through their online system. There's a fee, and some of the paperwork and payment stuff is easier with a trusted adult's help, especially if you're under 18. While minors can absolutely hold their own copyright, you can hit snags with online payments and contracts. Note: it's a "loop in a grown-up you trust" thing, not a "you can't do this" thing.
The Copyright Office has been clear: copyright protects human creativity. Purely AI-generated text cannot be copyrighted - only your human contributions can. In 2026 the Supreme Court left that rule standing. The Office clarified that just typing prompts, even really detailed ones, generally isn't enough to make the output "yours" in the eyes of the law.
So what does that mean? The more of the book is truly you, the more of it you actually own.
And you have to be honest about it. When you register a work that used more than a tiny bit of AI, you're required to disclose it and briefly describe what you did versus what the machine did. Fudging that can get your registration cancelled. Being straight with the Copyright Office isn't optional.
Disclosure to readers is its own thing. If you self-publish or publish, being upfront about how much AI you used is becoming a real expectation. There's now a "Human Authored" certification mark (run by the Authors Guild) that writers can put on books to signal the text is human-written. And there are cautionary tales going the other way: in early 2026 a publisher pulled a horror novel called Shy Girl after readers noticed what looked like heavy AI use. Readers care. And trust, once you lose it, is expensive to buy back.
A simple test: would you be comfortable telling your readers exactly how you made this? If yes, you're probably fine. If you'd rather they didn't know, that's your answer about whether you've crossed your own line.
The whole spectrum of disagreement
Nobody agrees on this, but that's not a problem for you to solve. For now, it's just a map you get to place yourself on. Here's a rough stab at defining the ranges:
The "no thanks" camp
Loads of creators are pushing back hard, and they're loud about it:
More than a thousand UK musicians - including big names - released a protest album in 2025 called "Is This What We Want?" It's almost entirely silent: recordings of empty studios. The track titles, read in order, spell out a sentence telling the government not to let AI companies take music without permission.
In 2026, around 10,000 writers "published" a completely blank book called Don't Steal This Book to protest the same thing.
Artists started slapping a "Human Intelligence" / made-by-a-human badge on their work - a badge artist Beth Spencer created - to proudly mark art made without AI.
Legendary animator Hayao Miyazaki (Studio Ghibli) has long been openly repulsed by AI-made animation, which made the internet's 2025 "Ghibli-style" AI-image craze land like a slap to a lot of fans.
Popular Hollywood screenwriter, Seth Rogen (Superbad) flatly stated "if your instinct is to use AI and not go through [the writing] process, you shouldn’t be a writer, because then you’re not writing." (Salty language warning at that link).
Closer to home for young writers: NaNoWriMo: a once-beloved write-a-novel-in-a-month challenge with a whole Young Writers Program for people 17 and under, defended AI use in 2024, and big-name authors quit its board in protest. The whole organization shut down in 2025. AI wasn't the only reason, but it was a big crack in the dam.
The "middle ground" camp
A lot of working writers use AI, but carefully - and they'd tell you there's a world of difference between scaffolding and ghostwriting.
Surveys in 2025 found roughly 45% of authors using AI somewhere in their process, but mostly for research, brainstorming, outlining, and marketing copy - not for writing the actual prose. Screenwriter Frank Deese (quoted above) said he'll let AI help him rough out the boring mechanics of a chase scene, but never the heart of the script.
👋 I started out pretty cranky about AI (and can still be super curmudgeonly about it) but I can see myself falling into this camp in some areas. I'd never use it to write a book for me, but I can see the usefulness of using it to help outline, work through beats, or even have it ask questions about a character or a line of action I'm stuck on.
This camp agrees, it's okay to use it to help build the ladder, but every step must be your own (write every real word yourself), and tell people the truth about what you did.
The "all in" camp
And then there are people going full throttle.
There are AI tools that advertise "full-length novels in seconds," and self-publishers using a workflow where the AI drafts most of the raw text and the human polishes the last slice. Their argument is that it's faster, it's cheaper, it lets people who'd never otherwise finish a book actually ship one, and readers can decide with their wallets.
The counter-argument is just as real: you can't fully copyright it, the quality is often thin, online stores are drowning in AI "book mill" sludge, and (see Shy Girl) readers who feel tricked don't come back.
You don't have to pick a team today.
But it helps to know these camps exist, because "everyone's doing it" and "everyone hates it" are both false. The truth is: everyone's arguing, and you're allowed an opinion of your own.
The readers can't agree either
On one end are readers who genuinely don't care. They want a good story, and if AI helped make it, so what! They'll judge the book, not the process.
On the other end are readers who feel the opposite, and they feel it hard. To them, AI in a book (or on a cover) is an absolute dealbreaker, full stop, do not pass go.
Real life sidebar: I belong to a support group for indie authors where many post drafts of their book covers for feedback. The covers! Not even the book itself. And posters will get absolutely slaughtered if the group thinks AI has been used, even for concept art.
Some readers will one-star a book, boycott an author, or pile on online at the faintest whiff of AI, sometimes even when the author didn't actually use it, or used it in a tiny, disclosed way! This backlash is real, it can get loud and personal, and it's not always fair or well-aimed.
Real life sidebar: My own writing, run through a fancy AI detector like the kind my kids' teachers use, often flags as 98% AI when I wrote it myself! All because I (mostly) understand how em dashes and semicolons work (though please note, my blogging software mangles em dashes, so you'll mostly find single dashes here).
But should that (reader rage) be the thing that decides how you write? Probably not, at least not the main thing. If you make every creative choice out of fear of the angriest possible reader, you're not really deciding what you believe; you're just dodging a mob.
And while I live in general fear of mobs and vindictive one-star-reviewers, it's a terrible and shaky way to build anything. Besides, the mob will eventually move on to something else to be outraged about.
But it's worth knowing the crowd exists for a couple of reasons. First, it's more reason to be honest up front: quiet, clear disclosure protects you far better than getting "caught" later. And second, it's a nudge to know your own audience: the readers who love a heartfelt, human-made story are exactly the ones most likely to notice, and mind, if it wasn't.
Let all that inform your choices, not make them for you.
So where do YOU land?
No lecture here. No test! There's no answer key, and if there were, I've lost it anyway.
The voice test. Write one short scene completely yourself. Then ask an AI to write its version of the same scene. Read them side by side out loud. Which one sounds like you? Which one would you be prouder to show a friend? Sit with why, but know this - no published writer sat down, wrote the first thing that fell out of their pen, and immediately got a book deal. Why do you think writing stories over and over again (all by themselves!) helped? How could doing the same help you?
Remember that real life sidebar up there? The one about the databases? Here's the hottest of my "cool author friend" tips: The thing I've discovered while using AI for work is that while everyone is shouting about how the computers and robots are coming for our jobs, there are still areas where your core knowledge is really important. If I didn't already have years of experience coding and working with databases, the AI might have slipped one past me. But I knew my craft after years and years of practicing it. So while you can probably find ways to use all these new tools without sacrificing your brain, there's real value in knowing your art first.
The "is this the part I care about?" question. For any task you're tempted to hand off, ask: is this the part of writing I actually love, or the part I'm trying to skip? Skipping the boring parts and skipping the parts that make you you are very different moves. Try to be conscious of the move you're making and why.
The golden rule. You now know a TON of writing got fed to AI without anyone asking permission. So: how would you feel if you poured a year into a book and it got scraped, uncredited, and no one ever paid you for it, all to help a machine write the next thousand books? Let your honest gut reaction inform how you treat other people's work, including the AI's training data.
p.s. This got really long so I tried to break it all up with dividers. Hopefully it helped!
