I asked ChatGPT to outline a chapter—is that a problem?
Is it okay if I give a chatbot previous writing samples and then ask it to write for me, while editing the results?
Can Claude write my nonfiction book proposal for me?
So. Many. Questions.
The answers are rarely black and white. Even when people are dead-set against using generative AI, it’s embedded in many grammar checkers and serves responses to searches on Google.
How wide are you willing to open the door for AI in your writing process?
You might decide that you’ll never use it to generate text for you, but will let it help in research and tedious non-writing tasks. Does summarizing something you’ve already written count? What about brainstorming titles for a piece?
As writers, we have to figure out how to navigate AI use in a way that makes sense for our personal standards and objectives, our relationships with readers and publications, and others’ expectations.
(Sheesh, wasn’t AI supposed to make things easier?)
Here’s another filter with which to think about these decisions: risk. Specifically, are you putting your relationship with the reader at risk?
What if others knew?
We read the news articles about AI problems— most recently, a book reviewer who used AI and the novel Shy Girl withdrawn for suspected AI use. How, then, can we think of using AI in our own work? Is what we’re doing okay? Where do we draw the line?
Here’s a useful question to ask: What if your reader (or publisher) knew exactly how you used AI to create the text? Would that risk your relationship?
Some people are open about how they use AI. (More than a few employers actively encourage it.)
Others don’t want the world to know. AI-detected apps are imperfect. (I feel for today’s college students, worried about human-generated writing being flagged as AI!) Even people writing without AI are sharing advice on how not to be falsely accused by people or apps.
The better question isn’t whether people will find out, but how they would feel if they did know.
The simple transparency test
What if we lived in a world of perfect transparency? (We’ll probably get there eventually.) That changes the question from “What if they knew” to “What am I putting at risk in the way I’m using AI?”
For example, if AI writes part of your book manuscript, you might risk compliance with your publishing contract. (See Jane Friedman’s excellent FAQ on AI and publishing.) Sometimes you risk your own growth, if you are using AI-generated content to skip the deep thinking and internalizing part of writing.
Usually, however, you risk your relationship with the reader.
To think about what’s at risk with the reader, try this simple thought exercise.
Assume that the reader can see exactly how you used AI in the writing; from researching and outlining to drafting and revising. How would they feel about it? Here are a few options to get you started.
- They would be delighted or think it’s cool.
- They wouldn’t care as long as it meets their needs.
- They might feel cheated or misled.
- They would question your expertise.
- They might feel owed them more time and attention.
- They would be fine if you told them, but not if you didn’t.
- They wouldn’t want to read it.
- They would trust you less.
The answers for every project will vary, based on context and your relationship with the reader matter. Are you writing wedding vows? Website copy for your employer? Context is everything.
If you’re submitting pieces to a publication or press, substitute “editor” or “agent” for reader. If you’re writing on the job, try it with your manager, the executive team, and the customer as readers.
This test aligns with the concept of servant authorship—we write to serve our readers. If we stick to what our readers want and need, we can make better decisions about how and when to use AI as a tool to support the process.
Take the next step: Disclose
You can find out how people feel about it by disclosing your AI use.
Here’s my disclosure: AI never writes my blog posts or books for me.
I sometimes use generative AI to expand my thinking if I need a brainstorming boost. For example, my first draft of this post had a much shorter list of objections above. Claude suggested a few more, and I riffed on its suggestions. It expanded my thinking.
(I also find it amusing to use AI to make suggestions on a post about writing with AI.)
Claude also helped me come up with the post title. My brain freezes up thinking of titles.
Here I am telling you what I did. Does it change your opinion of me? Have I lost your trust? (I hope not!) We’re all figuring out this AI thing in real time together, and I share in hopes of helping advance the conversations.
Does the Transparency Test help you? Let me know.
Related Resources and Fun Stuff
This thought exercise also appears in the 10-year anniversary edition of The Writer’s Process. (Stay tuned for that in May 2026).
Until then, you can find a few resources I trust on my website: AI and Writing Resources.
Read Ann Handley’s post The em-dash is not an AI tell. You can buy a “Justice for Em-dashes” shirt here, too. 👕
