Why generative AI isn’t a useful tool (right now) for expert book editors

As a writer, editor, and business owner, it’s important to let you—my friends, colleagues, and (potential) clients—know how I use AI in my editing business.

The short answer is that I don’t.

Let me clarify that as a policy:

I don’t use generative AI tools to generate or assist my work as a developmental editor, consultant, line/copyeditor, or creative writer.

Here are some of the reasons why:

• I’ve experimented with various tools and found them unreliable for copyediting

• They’re bad for the environment

• Agents and traditional publishers don’t allow use of such tools

• The joy of writing is all about learning the craft and figuring out how to apply it. The “product” is far less important than the journey for most creative writers.

• For fiction and humanities texts, the manuscripts have a certain artistic quality linked to the human act of creation. I want to see a writer's voice, not an AI compilation

• If writers use these tools to generate or edit text, the output is usually less original

• It is impossible for an agent or editor to judge what craft knowledge the writer has if they've used AI tools to generate text

• My own published works have been stolen and used without my consent by generative AI tools

Let me now consider the different use cases in turn:

Using generative AI for developmental editing and consulting

Pick a field you know a lot about. Ask ChatGPT questions about it, and you’ll see how mediocre and unreliable the technology is.

People mostly hire me as a consultant. They’re expecting a human to engage with their text.

More specifically, they're hiring me for my human publishing, fiction, and academic expertise.

These people already know they can ask gen-AI tools similar questions, and it would break trust if I were to give them a report or feedback part-generated by a gen-AI technology. This, combined with the fact that it’s simply not very good, is why I would never use it for developmental editing.

Also, many writers seeking traditional publication are highly skeptical of such technologies, or have even been burned by them. Even raising the question of their transparent use in assisting my workflow for, say, an editorial report, could easily break trust.

The only use case I’ve noticed as potentially useful for developmental editing is the act of creating a book map for a novel, and various companies are offering this.

But on reflection, if the book map created is unreliable and needs extensive checking, part of my job will have simply shifted from book map creation to book map checking. Creating a book map is something I do only as part of a full developmental edit. And I often don’t share the map with the client because it’s not a particularly useful deliverable. In fact, it’s the process of creating it that gives me clarity on which story elements need work. So using a gen-AI tool to create a book map would potentially involve me bypassing some useful creative labor.

Using generative AI for line/copyediting

Gen-AI tools are notoriously unreliable, even for light copyediting.

If and when publishers develop workflows, then I will test out those workflows before deciding if I’m happy to use them or not. I can see several great opportunities for using AI in a copyediting workflow for things like:

• Fixing dialogue punctuation

• Formatting reference lists and bibliographies swiftly

• Adjusting spelling preferences so the top-listed entry from the Merriam-Webster dictionary is used throughout the document

These are all tasks with a simply yes or no response (once a style sheet has been set up). But these tools must be developed in an ethical and transparent way. My experience, and that of my network, tells me that the technology simply isn’t there yet.

Literary translation

This is perhaps the field most strongly affected by gen-AI technology use so far. But literary translation is a more complex skill than line/copyediting because it involves editing AND moving across languages.

I contend that the reason this technology is being used here is that people buying a literary translation are often not in a good position to judge the quality of the translation. This therefore creates a market for mediocre translations, and that’s why gen-AI has flourished here, over editing, where it’s easier to judge quality because all the original and the edit are written in the same language.

Dorothy Kenny recently gave an excellent keynote talk on the use of AI in literary translation and concluded:

Researchers contemplated three initial motivations for using AI in literary translation: the inherent challenges of the task (not just to communicate meaning, but style and textual effects); opportunism (with the emergence of the e-book and availability of legacy data) and altruism (to increase the range of foreign language books). Yet, in a summary of recent research (Kolb 2023; Kenny 2025), MT is perceived as offering lower quality, accuracy and creativity; MT + post-editing does not improve productivity; no uptick has been seen for translation into minoritised languages, and MT is associated with worsening economic conditions for translators.

Creative writing

All agents and literary magazines I like and respect do not permit gen-AI generated or assisted texts. If I were to submit such a text, I’d be shooting myself in the foot as a writer.

Also, with my editor hat on, if I receive an AI part-generated manuscript, I have no idea what creative writing techniques the author has mastered and which they haven’t. In short, the technology hides the author’s shortcomings.

And for creative writing, there is always a balance between commerce and art to be struck (a balance not present in, say, business books). The act of writing and learning is what it’s all about, much more so than the product. This is certainly true for the literary fiction world and the literary end of genre fiction (which is where I’ve put my flag in the sand). I can see potential use cases for commercial self-publishing (and, in future, the commercial end of traditional publishing), but that’s not a market I have anything to do with.

What about the editors saying it’s great?

My discussion above can only speak for fiction editing, academic (humanities) book editing, literary translation, and creative writing. Editors working in different parts of the market may find other uses for these technologies. But beware of claims that learning ChatGPT will triple your editing speed or similar. At the end of the day, editors still have to read and digest a full text, and that takes time.

My no to generative AI use is firm because I’d rather work in a different creative industry than edit and consult on manuscripts produced by AI. If that means I say no to more jobs or close myself off to a subset of the market, that is only a good thing—after all, I need only a handful of clients a year, and I'd rather we were well aligned.

Isn’t this creating a two-tier system?

This is my fear, and I believe it’s happening. But this is not a problem I can solve as a single-person business.

Authors (especially those in the self-publishing world, where texts aren’t vetted) will continue to experiment and play with these technologies. And there are potential use cases for assistive AI in some fields. But until the landscape becomes clearer, I’ve chosen to align myself with agents in traditional publishing because this is the position that makes sense for my business and my creative self right now.