The controversy over AI in publishing and writing has been growing, with many having their own views on whether AI training infringes on copyright, whether it’s OK to post a blog that was written by AI, whether publishing a novel that was written by AI is acceptable, etcetera. As editor-in-chief and publisher at Pelican Book Group, I have always looked ahead to technology, embraced it when I thought it should be and rejected it when I felt it was unnecessary or dangerous.
As it stands today, I both love AI and am cautious of it. But regardless of my personal feelings towards it (or yours, if I may be so bold to say), it is a technology that will only advance and will not go away. Like all industrial and tech advances that at their infancy were poised to have global application, we cannot shy away from the reality of progress. For instance, my mother told me that when she was a child, her mother didn’t want to get a television because TV was a fad that would never usurp the radio. I’m sure during the dawn of television, many people felt that way. Their feelings didn’t change reality. Telephones, TV, personal computers, cell phones, digital printing, e-readers and e-books… are all advancements that changed daily living for better or worse depending on one’s viewpoint.
Whether you embrace AI, fear it, or are indifferent to it, none of these stances will change the unfolding future. The only thing we can do is face it head on and learn to use AI in a productive, non-harmful way. It isn’t the tech that’s necessarily bad, it’s how humans choose to use it.
So, today, I want to take a look at the acceptable ways for an author to use AI, because there is a major difference between using AI as a tool to help you develop plots, outline characters, edit and promote a book and having the LLM (Large Language Model) write a story and then slapping your name on it as “author” (when in reality you didn’t really author anything).
The Value of Human Authorship
As authors, we each have a unique voice. Even if AI can somehow mimic that voice, it cannot reproduce it with precision. (which is a blessing, by the way). It cannot “feel” when a certain word is emotionally more impactful or an ellipsis will evoke the right amount of suspense. When we first start writing, our voice needs to be developed, skill honed. I’ve said on several occasions that in order for authors to effectively break grammar rules, they must first learn how not to break them. It takes skill and effort to learn how and when to introduce a fragment or a one-word paragraph; but more than learning how, one must know why. And that takes instinct and understanding—understanding that comes from study (eg. knowing the proper grammar rules) and instinct that comes from human emotion and experience. An LLM might be able to mimic and learn technique; it might even be programmed with the reasons/whys certain authors used certain literary devices, but AI doesn’t have instinct honed from experience or the emotional understanding to be individually creative, to know from first-hand experience how humans react to different situations. That takes a human, animated with a soul, who can draw from his own perspective and observance. Think of this in Christian terms for a moment: God granted humans free will rather than creating us with an unbreakable set of parameters (even though sometimes it seems as if the world would be a lot more harmonious if we didn’t have the choice to choose evil). He gave us free will (even though sometimes we choose wrongly) because love counts only when it is freely given, not forced. Robots that are programmed can’t love, can’t feel, nor can they have a unique thought. Machines are limited to their programming.
Can We Use AI as a Ghostwriter?
So where does our unique voice fall if we use AI as a ghostwriter for our novel? After all, human ghostwriters exist. What’s the difference? There are several things we need consider. The first being that of true authorship. Even if we take into account that human-to-human ghostwriting is a true profession, we have to acknowledge that the human ghostwriter has his own talent and unique voice which he is lending to the by-line author. In most cases, the by-line author has presented the idea to the ghostwriter, and the ghostwriter is following the demands of the by-line author.
On the contrary, when we prompt AI to write a story, we have no idea what the LLM will produce (even if we’ve given it a skilled prompt). The AI’s voice and “talent” are nothing more than an algorithm that is dictated by input—yes massive amounts of input including our prompt text, but never actual talent. Unlike a human ghostwriter, with AI, there is never a completely unique output; simply a rearranging and regurgitating of previously received programming. Again, highly sophisticated programming that can quickly deduce and induce uniqueness, but no actual uniqueness. That’s why at present, there is so much repetition in AI output.
Yes, we can edit the AI-generated story, try to “make it our own”, but in that instance, we are an editor and not an author, because we’ve simply taken a “frankensteined” output from billions of coded (not unique) input and put our own spin on it. And if that is true, then how, could we sell that product with our name on it as “author”?
At this point, you might wander back in thought to hiring a human ghostwriter and it being acceptable to put your own name on that work. This is completely different from using AI. Why? Because there is a direct exchange of services with complete understanding between the human parties. The ghostwriter is expected to create a completely unique (i.e. not plagiarized, nor rehashed) story. As we’ve established, AI cannot output something completely unique. Even it’s skillful responses will repeat words, phrases, sentences; not just for “you” but for “anyone” who asks it for similar output. Would you pay a ghostwriter to sell you the same paragraph he sold to fifteen other by-line authors? Would you want to put your name on that paragraph as “author” only to see other “authors” incorporating the same paragraph in their works?
This limitation and truth about AI highlights current ethical concerns. We’ve already seen controversy and lawsuits where industry professionals have sued AI companies for allegedly training LLMs on copyrighted works without permission—purportedly infringing on intellectual property by repurposing human-authored content. These cases spotlight how AI’s “rearranging and regurgitating” (as opposed to actual creation) might cross legal and moral lines if we’re not careful.
While technically both human ghostwriter and AI LLMs have learned the craft of writing by taking in a lot of input (what human author hasn’t studied books in his genre, taken writing classes, learned grammar rules, etc.), the human author understands when he is reusing the same sentences. The AI doesn’t because all it’s doing is spitting an output that matches the user’s request. This AI repetition and formula is one reason AI detectors work (disclaimer: AI detectors aren’t 100% accurate and some aren’t accurate at all). As technology evolves, legal battles (and hopefully greater understanding without legal battles) will undoubtedly inform how we decide what is ethical AI use. One thing is certain: we must ensure AI remains a human aid rather than becoming a replacement for human creativity.
Acceptable Uses of AI in the Writing Process
In what capacity is it OK for authors to use AI? While there are myriad and nuanced answers to that question (because, let’s face it, the list of how and why AI is used cannot be exhausted here), a general “rule” (my own self-defined rule) is to ask the question of purpose: What is the purpose of using AI on this project? The answer to this question, regardless of how or why one is using AI, should give sufficient insight into whether the AI’s use is acceptable. If the AI’s purpose is to be a tool, then it’s OK.
- Is the AI being used to check grammar? 👌
- Is the AI researching a topic for a human author to learn? 👌
- Is the AI expected to write a story for you to sell (even with human edits)? 👎
- Is the AI being used to help create an outline for a story you will write? 👌
- Is the AI being used to find plot holes or lacking characterization, etc.? 👌
- Is the AI writing from a non-fiction, first-person viewpoint in which the reader is led to believe the “author” actually experienced whatever is conveyed in the text? 👎 (This is to deceive the reader, and that is never OK)
- Is the AI being used to write brief promotional copy that will shed a spotlight on your human-authored story? 👌 (Why? Because marketing copy [even most blog posts] is more science than creativity and is not sold as an original, unique work of the human author)
- Is the AI writing a short story that will complement the human-authored main story? 👌 👎 (If the AI-produced story is not sold but given as a promotional piece, and it is not billed as authored by a human, then it remains a promotional tool. That’s OK. If it is purported to be written by a human, and/or it is being sold for an exchange of cash without disclosure of AI involvement, then a human author has just lied to the readership, and that can’t be OK.)
When AI is used as a tool, a human author is still in complete control of creative output. The human wrote it, the human edited it; the human merely used the AI to help with pre-writing tasks (eg. Outlining, research gathering) or post-writing tasks (eg. grammar check, checking for inconsistencies and a well-wrapped conflict, edit suggestions, promotion of a human-authored work).
To illustrate my point further, I deliberately used AI to help me outline this article. Rather than brainstorming my idea on “paper” or “pants-ing” the article (which I’ve done, historically) I asked AI to help me write an outline. In no particular order, I explained to the LLM the points I wanted to make and asked it to produce a structured outline for me, which I could use as a guide. Below is the outline. As you can see, this helped organize my thoughts very well. However, every part of the article is mine.
Even though if AI had brainstormed an idea for me, that would have been OK, in this instance, the idea for this article was mine, and the writing is mine. You’ll notice that I consolidated or condensed some of the ideas, and I also completely ignored its “Conclusion” suggestion. I did this mainly because the suggested conclusion topic didn’t fit with my true purpose for the post. However, I was impressed with the way in which the LLM organized my ideas without me having to spend extra time doing that. The LLM was a fantastic, time-saving tool.
So what’s the TL;DR version: AI is a tool, not an author. Use AI as often as you like, but use it well and always, always use it ethically. If you have a well-formed conscience, you’ll know. You know your own motives. If you’re thinking of using AI and you wouldn’t want someone to find out, then maybe think twice about using it in whatever way you’re thinking of using it. (That actually goes for any activity, right? If you can’t own up to it, don’t do it.)
One last thought/disclaimer: This topic is a hot-button one that will, I’m sure, evolve as the technology evolves. AI is awesome and scary. (Although, I don’t think the Sarah Conners of the world have to worry too much at the moment.) As things change, my thoughts on what is an OK way to use AI might also change. But one thing that will be true no matter how tech evolves: always use it ethically and with keeping well-formed morals intact.
Happy writing, human,
Nicola,
Pelican Book Group Editor-in-Chief
Guideline Outline for this Post
I. Introduction
- Brief context: The rise of AI in publishing and writing.
- State your main viewpoint: There’s a crucial difference between using AI as a creative aid and using it to generate the actual prose.
- Thesis: AI can be a valuable tool for outlining, plot development, and self-editing, but when it writes the text, the human ceases to be the true author.
II. The Value of Human Authorship
- What it means to be an author: voice, creativity, and personal experience.
- The irreplaceable human touch—emotional resonance, unique perspective, and authenticity.
III. Acceptable Uses of AI in the Writing Process
- A. Outlining
- How AI can help brainstorm ideas, structure plots, and organize chapters.
- AI as a “sounding board” for creativity.
- B. Plot Development
- Using AI to explore “what if” scenarios, develop subplots, or flesh out character arcs.
- AI as a collaborator, not a creator.
- C. Self-Editing
- Grammar, style, and consistency checks.
- Suggestions for tightening prose or clarifying meaning.
- Emphasize: In all these cases, the human author remains in control and makes the final creative decisions.
IV. The Line Crossed: AI as the Actual Writer
- What happens when AI generates the prose/text itself.
- The loss of human voice, originality, and ownership.
- Ethical and creative implications: Is the “author” still the author?
- The risk of homogenized, soulless writing and loss of literary diversity.
V. Why This Distinction Matters
- Integrity in authorship and publishing.
- Reader expectations: wanting to connect with a real human voice.
- The future of literature—preserving creativity, diversity, and authenticity.
VI. Conclusion
- Restate your position: AI is a powerful tool, but not a replacement for the human author.
- Encourage thoughtful, transparent use of AI in the writing process.
- Call to action: Authors, publishers, and readers should value and protect the unique role of human creativity in storytelling.
