6 Productive Ways I Used AI (Ethically) to Help Write My Books

As a former English teacher, I understand the power of writing and its connection to thinking. So I don’t discount the arguments that using AI to help one write will make them lazy and lead to brain atrophy (especially for young developing brains).

But like most things, it isn’t black and white. It depends on how and when you use it. I feel no shame in sometimes completely turning my thinking over to AI.

Being an editor is a highly cognitive job, being a mom to three young kids is … well, if you know, you know. It’s okay to give my thinking muscle a break sometimes.

That said, AI often helps me think even more deeply. I’ve strengthened my problem-solving muscles by learning how to prompt AI more effectively. Some of the output from AI has helped me learn and understand things I didn’t before. And then, of course, I always analyze the output for effectiveness. Analysis is a great brain-exercising skill. Anyway, I could go on and on about this topic, but that isn’t the point of this blog.

As a reader, I appreciate well-written content. So I don’t discount the arguments that AI produces horrible writing. The problem with that argument is, without disclosure, you may not realize AI “wrote,” or helped to write, a well-written social media post or blog.

Most, if not all of us, can recognize poor AI writing. It has a recognizable cadence and pattern. So people use those as examples as “proof” that AI can’t write and can only produce slop. However, I know some great writers on LinkedIn who use AI, and it doesn’t sound like AI.

As an editor and supporter of authors and writers, I understand the frustration and fear of people cheapening your skillset by saying a computer can do it. I faced my first dry spell of my career because of that very thing. We can only hope that soon those who think this will realize how much the creative world needs humans.

And then, of course, there’s the fear of having your copyrighted work used without consent or compensation. Personally, I’d be flattered if someone wanted to use my work to train the model, but I’m not the norm.

As a human, I acknowledge the carbon footprint and energy consumption issues, though I admit that one is harder for me to relate to. I should care more about the environment.

In essence, AI has problems. I completely understand those who say they would never, ever use AI to help them write. I’m just not one of them.

I recently finished the first drafts of two manuscripts: Beacon Point’s Self-Editing Guide for Nonfiction Authors and Beacon Point’s Self-Editing Guide for Fiction Authors (titles are subject to change, and let’s be honest, probably will … I suck at titles and headlines), and I did use AI. Definitely not to write the books in their entirety. I wrote the majority of the words, but AI drafted some sentences and paragraphs verbatim.

So let’s dissect the six ways I used it and which cases needed a disclosure.

Disclosure:

I wrote every single word of this blog post. Just wanted to clear the air there. Yeah, I know, I’m not the best writer. I wish I were as good as the real writers I know. I’ve always hesitated to call myself a writer. I’m more comfortable in the editor role. But hey, someone’s gotta write books about editing, right?

My self-editing guides contain a variety of editing topics grouped into four passes. From my years of editing, I had a good idea of what topics I wanted to put in each pass.

To start, I added topics that I already had blog posts on (repurposing for the win) and grouped them in the correct pass. Then I listed additional topics that I knew were essential and had planned on writing a blog post on someday.

I had a good outline, but I wanted to ensure I hadn’t missed anything really crucial. So I asked AI. It gave me some great additional topics: scene entry and exit points (fiction), microtension (fiction), ensure accurate reading level (nonfiction), and maintain consistent tone (nonfiction).

I couldn’t use all its ideas (and quite frankly, I wouldn’t want to use some of them); that would make for very long books, but I picked those ones, and the books are better for it.

Disclosure: Using AI to generate topic ideas is widely accepted without the need to disclose. Ideas can’t be copyrighted, and I could have found the ideas in an internet search—just would have taken longer and wouldn’t allow for a back and forth.

While I’m using beta readers who will do this much better than AI, I wanted to have a cursory check for clarity on especially difficult concepts. So I fed each chapter with a difficult concept to AI and asked it to be brutally honest and tell me if the concept was clear to a brand-new writer and not condescending to an experienced writer.

After the analysis, I went in and made clarity corrections.

Disclosure: I didn’t disclose when I used AI to flag unclear sections, as it was just giving me feedback, not generating content.

A few times I was just stuck on how to make it clearer, and so I just asked AI to do it. This is a tool I wish I had when I was a teacher. Sometimes I just failed to explain something the way a particular student would understand.

After reading the output, I either used its output to write something different—but it gave me the idea—or I used just some of its output. Occasionally, I used its answer entirely. However, since the original input was my writing, the output still contained many of my words and was completely my idea.

Well, we now enter territory that could necessitate disclosure. But here’s the dilemma, I can’t mark it in the text. That would be ridiculous. Passages could end up looking like this:

➜But here’s the “problem: I can’t exactly insert a note every time AI helped” (AI). That would be ridiculous. “You’d” (AI) end up with “passages that looked” (AI) like this:

If you couldn’t tell, I asked AI to rewrite the previous paragraph but retain some of my original words.

Some would argue that in this situation one should put a disclosure in the front of the book that AI helped write the book. But what does that mean “helped”? It’s too vague, so some readers may question whether I’m an authority on the topic at all, thinking AI wrote most or all the book, which isn’t remotely true.

My hours spent writing this book shouldn’t be discounted because AI massaged some of the language here or there. I can ethically say I authored these books.

Disclosure: When the AI flagged a passage as unclear and provided suggestions, I either used its output to write it in my own words, used some of its suggested text, or used its suggestion entirely. Since the original input was my own writing, I didn’t disclose each occurrence of this use.

Occasionally, I worried my tone sounded wrong. Was it too harsh? Did it sound too lectury?

My husband and best friend have helped me see that sometimes the words I use express a tone—and, thus, a message—I didn’t mean. I’m always baffled. How did they get that? You’d think being someone who works with words I would know this, but hey, we all have our strengths and weaknesses.

I used AI in the same way I did for #2. Just an analysis in some cases; massaging the wording in others.

Disclosure: When the AI flagged a tonal issue and suggested a fix, I either used its output to write it in my own words, used some of its suggested text, or used its suggestion entirely. Since the original input was my own writing, I didn’t disclose each occurrence of this use.

AI can’t edit as well as a human, and I’m not just saying that because I’d like to keep my job. But it can be an effective partner with a human who does have that skillset.

I fed it the lessons from my nonfiction guide that deal with line editing (strengthening word choice, ensuring good sentence fluency, and eliminating weaker verbs) so it had a great knowledge base to work with. Then I asked it to flag instances of weak writing (e.g., using expletives like “there are” or “there is”—a topic specifically mentioned in my guide) and offer a suggested fix. Sometimes I used the fix it gave me; sometimes I tweaked it; sometimes I wrote my own. But it was nice to have it flag the spots I needed to focus on.

It’s hard to edit your own work. Plus, I just didn’t want to spend hours upon hours line editing them. But, wait, aren’t your books teaching authors to do that very thing: edit themselves? Yes, but before you engrave a scarlet H on my forehead for hypocrisy, let me explain.

Self-editing is valuable for three reasons: It strengthens your writing ability, reduces your cost of professional editing, and ensures a higher-quality manuscript after the editor finishes (an editor can only take it so far).

I’m a great editor, and I spend a good chunk of my day editing, so I get plenty of practice strengthening that muscle (e.g., I just edited out four uses of expletives, “there are,” “it is,” etc., in a manuscript I’m editing).

Plus, I didn’t just say, “edit this for me.”

Using AI as my line editing partner also will help me achieve the other two reasons: lower editing costs and a higher-quality manuscript in the end.

So why can’t you just skip the grunt work of self-editing and have AI edit for you? I mean, you could. It would improve some things, but you could do better.

I didn’t skip the grunt work entirely. I used AI as a partner.

And the important part here is, I’m an editor, who

• knows how to prompt AI to edit effectively (I took two courses on it)
• knows what output to straight-up ignore, what to tweak, and what to use wholesale
• fed it my guides as a knowledge base

Is it as good as me? Heck no! Most likely, I have more areas of weaker writing in my drafts than it flagged for me. But it was good enough. And I didn’t want to fall into the trap of editing them to death, fine-tuning each and every word.

So I used AI to help with the self-editing.

And maybe you can too. If you have the editing skillset (which you will if you read my books once they’re published) and know how to prompt AI effectively, then go ahead and use AI to help you with your self-editing.

Just don’t rely on it as the only editor. It doesn’t replace a professional editor. I’m paying for an editor for my books too

Disclosure: I didn’t disclose each specific instance where I use AI as a self-editing partner. It would be annoying to cite which sentences AI helped me with. And this is how it is with a professional editor: Authors often use my entire suggested edit, and they didn’t need to credit me for having written the words, sentences, and paragraphs.

After writing the first draft, 8 out of 49 total lessons (across both guides) felt incomplete. I could have researched some more to complete them on my own. And I would have except … well, AI can do it faster. Of course, faster isn’t always better, especially when it comes to writing and editing.

There’s something to be said about struggling and wrestling through what to write. About researching the topic further. I taught this very thing to my students. And I had plenty of those moments in writing my books—don’t want my brain to atrophy.

But after editing all day (working in the business) and then spending my free time working on my business (I’m going through huge business growth right now) and writing these guides until midnight, I did ask AI a few times to expand on what I had, then I edited it from there.

But even when I did that, I didn’t outsource all my thinking. I still had to analyze the output, decide if it fit my understanding of the topic, and add in my own ideas here and there. I didn’t just copy, paste, and call it a day. I often had follow-up prompts, asking it to rewrite what it had written to include specific concepts or to strengthen and expand on certain areas with my own ideas.

One topic was new to me. AI had suggested it (see use case #1). So I read several blog posts on the topic to understand it. After doing so, I worried the content would sound too much like those other people’s blogs since they were fresh in my mind. So AI drafted it, I edited its output, then added an entirely new section to it to complete the chapter.

Okay now we’re getting into the territory of entire paragraphs potentially coming from AI.

A few of my social media posts have been like 80 percent written by AI but 90–100 my ideas. And I don’t disclose this. But I’m also not profiting off my social media posts like I could with the book. So, yes, better to be transparent.

Disclosure: Where applicable, I added this as a footnote: “Portions of this topic (insert name of topic) were drafted with AI assistance, then I edited and reviewed the output.” With the one it initially drafted, I had this footnote: AI initially drafted the main portion of this lesson, then I added to it and edited its output

Each topic in the guides has a “Self-Editing in Practice” section—an example passage for readers to practice revising for the topic, followed by my revision.

For some topics, I wrote the “bad” version example. I had the idea for it and did it.

For others, I already had examples from my clients’ work (I got full permission). In those cases, I had written a blog post on the topic soon after editing that book, so their manuscript was fresh on my mind.

So that left several without a practice example, especially in the fiction guide. I’ve edited nearly 300 books at this point, so I can’t remember which manuscripts contain a perfect example for a given topic that I could use with permission.

I’m an editor, not a fiction author. I’m better at improving what’s already there than creating from scratch, especially when the creation needs to be an example of something specific without being too obvious.

So I turned to AI.

I fed it the entire lesson and asked for a “self-editing in practice” example. To get a decent or great example, I often had to give it follow-up prompts because its first iteration made the issue too glaringly obvious.

This isn’t self-editing for kindergarteners.

In some cases, I went through up to ten iterations before copying and pasting the example in. It was my knowledge and understanding of the topic guiding AI to produce good examples. For the less complicated topics, the output was fine on the first try. But for the most part, on its own, AI couldn’t have produced what ended up in the book. It needed guided prompts from someone with a deep understanding of the topic.

Disclosure: AI wrote some of the “bad” examples in the “self-editing in practice” sections. All such instances are noted with “Generated by ChatGPT” or “Generated by Claude.”

Additionally, I acknowledged this in the introduction,

“The ‘Self-editing in practice’ sections came from various sources:

• My original examples
• AI-generated examples from ChatGPT or Claude (always noted)
• Examples from previous clients’ manuscripts (anonymized and used with permission)”

While I consider the above uses of AI productive, not every approach to AI passes the same test. I’ve seen firsthand how overreliance can ruin a book.

Earlier this year (2025), I was hired to edit a nonfiction manuscript that had so obviously been entirely written by AI. Maybe a few sentences were the “author’s,” I don’t know. But the use of AI was rampant, and the book was bad. It was all theoretical, not practical, very repetitive, and not at all enjoyable to read.

Every other sentence followed an “it’s not this; it’s this” pattern, and lists of three were everywhere. And em dashes up the wazoo. None of these alone indicate AI; it was the sheer number of them.

*Side note: Every single em dash in my blog (and the vast majority in my book) I placed intentionally, not because I copied and pasted from AI. But when they’re sprinkled in where they don’t belong and used like the only punctuation that exists, then … um, yeah. That can be an indicator.*

This is NOT good writing. I nearly cried trying to edit it. The content was meaningless. I won’t get into the weeds of how I ended up with this job in the first place. But suffice it to say, it was the most painful thing to read, let alone edit.

I edited another nonfiction manuscript that wasn’t nearly as bad. It was clear this author had actually put some thought into it. I think the previous one had just said, “Write a book on this topic.” This author had some of their own stories in there, real actual applicable ideas, etc. But, again, the overuse of the AI cadences and writing style made it difficult to enjoy reading it.

Then I had someone contact me on LinkedIn who hired me to read his novel that was entirely generated by an AI he had built. He wanted to know if it was readable and marketable as is. I bet you can guess the answer.

I had to use AI to help me “read” the whole thing so I could analyze the plot arc. And by that, I mean after chapter 8, I gave up and asked it to give me bullet points of events for each chapter. I couldn’t read it anymore because the writing was not enjoyable, and I was so confused about what the heck was going on. But I wanted to report on AI’s ability to follow a plot structure.

For any of you who are curious to learn what AI did well and what it did not, you can read my report here.

Bottom line: Don’t use AI to write an entire book or to handle all your editing.

If you have a skillset, like editing or writing, AI can be a powerful partner to augment that skill. If you don’t, your prompts—and therefore your results—will be poor.

Like most tools, AI’s value depends on the person using it. Hopefully, sharing my process helps you find your own balance between creativity, authenticity, ethics, and efficiency.

To get notified whenever I publish a new blog, learn useful writing and publishing tips from other professionals, get discounts on my resources, see what I’m reading and recommending, and learn when I have editing openings, sign up for my newsletter.

I’d love to hear from you. Comment below using any of these prompts to guide you:

• Which use case do you disagree with? Why do you feel it crossed a line?
• Where’s your personal line for disclosing AI help in a book?
• Which use case would help you most right now?
• What’s a “never do this” AI rule you’re adopting for your own writing?