AI really wants to read your books: a writer's guide to AI.
- Julie R. Neidlinger
- Jun 19
- 8 min read
In an email from the self-publishing company I use, I discovered that:
As you may have read, artificial intelligence (AI) developers are starting to approach book publishers, seeking to license books for the purpose of training Large Language Models (LLMs). Books are prized for LLM training due to their long form narrative structures which can be used to develop Natural Language Processing, and a few book publishers already have AI training rights deals in place. AI training rights are poised to become another source of income for authors, yet so far, indie authors have not been part of the conversation.
A survey then went on into a bit more detail:
In recent months, a growing number of AI developers have begun approaching large publishers, seeking to license books for the purpose of training Large Language Models (LLMs).
Books – both fiction and non-fiction – are highly prized for LLM training due to their long form narrative structures which teach Natural Language Processing.
Common uses for these LLMs include powering personal productivity applications such as customer service chatbots, virtual assistants, and the drafting of written communications for marketing, customer service, and internal communications.
What are your AI training rights worth? There’s no hard and fast rule to answer this question because each licensing deal is different.
Some early compensation models for news publishers suggest the equivalent of about $100 per license for LLM training rights for a 75,000-word novel, which works out to a little over 1/10th of a cent per word.
Some experts believe training rights for long-form book content justifies higher compensation for training rights than news content.
Further complicating the question of fair compensation are other variables of interest to authors and publishers, such as the size of LLM being trained (larger, more complex LLMs are typically expected to pay more), output restrictions, and the applications built upon the LLM (internal/external; competitive/non-competitive).
The email's recipients were encouraged to respond to a survey. A few weeks later, the response came back a resounding no. The company wouldn't be offering this income option. The authors must have responded quickly and loudly.
I agreed with the results.
Yes, others are doing it. Yes, the AI will be trained eventually. Yes, some author made a few extra bucks today at what expense tomorrow. Yes, online writing is already being scraped. Most authors are always scrambling to find a way to earn income from their work, and AI is this strange monster lurking out there, threatening to kill you while also offering you snacks and stock tips.
When AI first exploded onto the scene as ChatGPT in all its overt and unhidden glory, I was unsure of what to think. I work as a writer and artist, the two things AI initially seemed to be angling for, and most obviously. It didn't take long until nearly every blog had AI-generated images that were so full of drama and chiaroscuro that I could never compete. There was a time, years ago, when people would pay me to paint images for their blog headers.
I've thought about AI a lot in terms of what it means for people, for artists and writers, and for how we think. If Google-brain already has people unable to hold onto facts or remember things, choosing instead to just search for something on their phone rather than fight to recall it, what would AI do? Specifically, how would AI affect my own work, specifically my writing?
Because yes, AI is now in every tool I use, from my computer's operating system to Google Drive to typo checkers now wanting to rewrite entire paragraphs for you.
It is likely this blog post will be updated as AI changes.
The biggest problems with AI for writers.
The biggest problems AI presents for writers in terms of using it as a tool in their work are:
Plagiarism is way too easy, even for people who wouldn't think of plagiarizing, because you don't know if AI is returning their own words or someone else's.
AI sometimes fabricates facts and returns questionable results, without always providing links to the sources.
It strips away some of the humanity in your writing, and you start to sound like other people using AI.
AI has the ideology of its programmers and the materials it consumes (often left-leaning).
AI is diminishing your ability to think creatively for problem solving, which is cumulative over time, and makes you not just lazy, but unable to do what you used to do on your own.
As long as you know this going in, you can understand what to do with the results AI gives you.
I saw a fellow jokingly say that typos are going to be the last bastion for writers in that they'll be proof that it was written by humans instead of AI. There may be truth to that, though you can still pick out AI writing if you're a writer and reader (though this is changing exponentially each day). It has a vague, circular, jargonized sound. I've read and identified AI copy that goes on for pages and while I can see that it's English, the grammar appears in order, and it kind of makes sense, it is wildly confusing in that it's still garbage-sounding.
A lot of the tools I use now have AI baked into them, and it's easy to forget that AI is at work in apps like Google Docs, Grammarly, Canva, Wix, and so on. I have to guard against "suggestions" for improving my writing at times. It's okay to disregard AI suggestions for "improving" your writing; I've noticed they remove some of the peculiarities of my style, and while a good edit is useful now and then, AI would strip my writing down to a common standard that had no differentiation from anyone else. Sometimes, that imperfect or roundabout way of stating something is what comes from being human.
So now we get to it, the much-needed writer's guide to AI.
A writer's guide to AI that you can honestly live with.
I have come to view AI as a tool for efficiency that I use in three ways:
Handle rote tasks that eat up a lot of time and creative energy.
Help with research.
Create outlines on topics I'm unfamiliar with.
AI can handle your rote tasks.
I do writing tasks for clients (and myself) that have a heavy brain load but low return. These include writing job descriptions, some web copy, meta descriptions for blog posts—you get the idea.
For those, I go to either ChatGPT, Microsoft Copilot, Grok, or use the web builder's built-in features and prompt it to do what I ask.
"Write a job description for _____ using these specific details: ____(client's job details)."
I don't have a problem with doing that generally, though sometimes the results are hilariously bad. There are times it's just better to do it myself.
AI can help with research.
AI can be a research tool if you know which platforms do the best work and also understand some of them generate nonsense.
While working on a client book, I needed an illustration story for a particular point. I tried as hard as I could to think of some historical story, some kind of way to drive home the point, but could not. I felt like I knew of a story but it wasn't coming to mind. So I headed over to one of the AI tools I was using at the time and gave it a very specific prompt asking it to find an event in history that contained specific aspects.
It came back with a strangely convincing, yet totally outlandish, story. There was no way it was true. I decided to backsearch and used every historical name, place, and event it detailed elsewhere to see if I could find anything.
Nope.
It had made it up, complete with dates.
AI can help with research, but it can also screw you over. For research purposes, you need an AI tool that is built for research and provides you with links to the sources it got its information from. For that, I'd recommend:
Not only do those source links help you verify the information, but you'll also need them when you cite your sources. For the record, Perplexity is my favorite general research tool.
If you're still wary, I understand. The problem is that typical internet searches in most browsers are now returning AI summaries as the top result. If they provide sources, great. But you have to be extra insistent now that your research is from a good source.
Whatever tool you use, if you're not sure about the veracity of the results, you can simply ask it how it arrived at its answer. Whether it provides the details you need is another story.
How I do NOT use AI.
I do not use AI to write the content that I consider the meat of my work.
I write my own blog posts, books, and articles. I don't care if the world around me turns to AI; I can't live with myself if I let AI do the very thing that brings me joy through challenge.
I'm not going to give any writer a pass on this one. You are a writer. You do the work of writing, even the ugly early practice writing. You train yourself to think. You do it, or you'll never develop as a writer. Research, typo-checking—that's something else. But writing? That's your job.
Yet I know that content creators are using AI to structure their books, sift through large amounts of information to pull out key ideas, rework old content, help with characterizations, and change the tone and feel of a passage of writing. There's so much gray area there. As a writer, part of the work and end result is sifting through the information and figuring out the structure on my own. Leap-frogging that prep work just to get to the writing is going to leave you with less of a grasp of the material and will affect your writing (and future writing growth).
Some writers who are using AI have found out the hard way, particularly if they left the AI prompt in the copy they dropped into their manuscript. I know that the excuse they are using is that it was just for brainstorming, but the word is brainstorming. As in, your brain.
One of the most popular searches that brings people to my website is for brainstorming worksheets, but I wonder at what point that will become obsolete as we let the machine come up with ideas for us.
When AI bots have scraped writers and other creators for so long that humans decide to turn their craft and creativity over to AI in part or whole, we end up with an ugly ouroboros in which AI starts eating its own offal. Its demise is as looming as the human creators it first scraped and then knee-capped, because it's now feeding off of the pablum the internet is flooded with, generated by itself. As humans learned there was profit in creating blogs that churned out nothing but AI content, those AI bots are eating the garbage they produced.
No wonder the dead internet theory has such legs.
Somewhere, between accepting that internet research has replaced ye olde library card catalogue and letting AI develop characters, structure, and entire content, we have to find a middle ground that keeps creativity humanized.
NOTE: I will likely be updating this post over time, as AI is a constantly changing subject that I am always digging up in my reading.
