Note #4: Let’s Come Clean: The Struggles and Possibilities of AI

Interesting things happen when writers are transparent with editors about how AI shaped their nonfiction.

Ah, the story forever circulating on editor Slack threads: a client sent over a manuscript-in-progress, and it was quickly clear to the editor that several of their chapters were AI-generated.

Luckily in my case, I didn’t jump into “gotcha” mode.

That’s because the writer I worked with was transparent about their AI use at the start and I’m deeply curious about how AI is informing the work of writers today, for better or worse. So, we were able to really dig in and explore the content for what it was: a launching pad for his book. Here’s some things we discussed from the get-go:

• The writer, who is an expert in their discipline, explained that because they spent a lot of time with AI, adding content and prompting, narrowing or widening the interaction, they felt that the content was now theirs, to a certain degree. Fascinating, right?

So our next topic of discussion needed to address the fact that content created solely by AI cannot be copyrighted. That consideration should weigh heavily on our mind as we gauge how much of the content is purely AI and where we need to add human-generated content (I have been relying on The Author’s Guild for advice on this).

• While some of their chapters were AI-heavy and had a hollow tone, others were narrative-driven and had a lot of personality. So this contrast presented the perfect opportunity for a restructuring of chapters, where the author would lead with narrative before transitioning into drier, concept-driven text.

This conversation was a great segway into deeper discussions about how we could remedy lack of voice. Below I explore the high stakes when we don’t carefully approach AI, some ways to approach AI-generated text, and why writers might not want to admit they are using it.

Soulless Syntax: When AI Kills the Writer’s Voice

I often ask nonfiction writers: What’s the engine behind this chapter or book? And what I mean by that is: What’s the animating purpose that propels the prose and makes it feel urgent? And how is that purpose consistently expressed through phrasing, tone, rhythm?

AI lacks such an engine. While we can prompt it to respond to a specific rhetorical situation (just as I prompted it to give me an outline for a blog post directed to authors and editors struggling with how to deal with AI), the resulting prose often lacks soul and narrative coherence—and readers will sense this.

While this issue is existential for all writers, for developmental editors specifically helping nonfiction writers hone their books, the stakes feel uniquely heightened, since good nonfiction is so reliant on things that only a human can provide: authority, trust, relatability.

So what do we do with this AI-generated text? I use the following tools both as a writer and developmental editor.

• Force yourself to work from only the outline

One of my tried-and-true practices is to push myself to fill in the content within an AI-provided outline (which I did with this post—I’m fessing up!) instead of attempting to make long blocks of AI text my own. That way, writers have the space to generate content that is so much better than what AI could generate. What’s more, by using outlines only, authors also avoid inadvertently citing incorrect, hallucinated content.

• Explore your emotional connection to the topic

Nonfiction writers should never underestimate the importance of personal emotion and conviction, especially when we humans will best AI in this regard every time. I commonly interview my clients on the deeper reasons why they are driven to write their book, and then we find ways to weave those reasons into the book as a throughline, whether spoken or unspoken. In terms of AI-generated text, we might work to repeatedly explicate how exactly this AI text connects to the larger thrust of the book.

• Layer voice into AI-generated content

For those blocks of AI text writers do not want to part with, writers have to be sure they tether the text to their voice. To achieve this, I often help them add anecdotes, tangents, or unique examples. I suggest writers start a new document in order to achieve some cognitive distance, since it can be hard to divorce oneself from the generated text once it is there.

The Shame Spiral: AI Admission Problem

Because I am a writer myself, I understand why writers wouldn’t want to come clean about using AI, especially when developmental editors can carry an intimidating bias for the human-crafted written word (and justifiably so). Authors may fear they will be seen as lazy, fake, or less skilled.

But transparency about AI matters, not only because it forms the bedrock of the author-editor relationship but also because transparency prevents missed opportunities. That’s why the time is ripe for developmental editors to find ways to offer safe disclosure spaces and help authors blend rather than purge.

For me, my somewhat accepting attitude toward AI probably stems from my background as a former writing professor: In the same way in which I began open conversations with students in an attempt to understand why they plagiarized so that I could help them avoid plagiarizing in the future, I invite my clients to put it all on the table in terms of how they are using AI so I can help them use (or not use) it to write powerful, effective books.

I remain very concerned about the environmental impact of AI, the way in which it does not compensate or credit writers when it trains with their work, its use of our personal data, our tendency to trust it as it hallucinates “facts” and citations, and the list goes on…but the horse is clearly out of the gate.

TL; DR: Let’s Navigate the Unknown, Together

While I’m just beginning to understand how AI can help (or hinder) nonfiction writers, I believe that the only way forward is for writers and developmental editors to discuss its use in a transparent, open way. When we stop seeing its use as a moral failing or stumbling block, we can develop editorial trust, experiment together, avoid AI pitfalls like loss of tone and fractured coherence, and hopefully discover better ways to communicate ideas.

Thanks for reading,

Natalie

the annotation room shares strategies on shaping nonfiction content that informs, persuades, and connects with readers. Natalie Tomlin—developmental editor, writing coach, and award-winning writer—hopes that this bi-weekly letter engages with a community of nonfiction writers and developmental editors and shares actionable advice on how to edit with purpose and precision. Click here to subscribe.