Ghost in the Machine: Artificial Intelligence and the Business of Writing
By Emilie-Noelle Provost | February 28, 2024 |
A few weeks ago, my publisher put up a post on the company’s private Facebook page. The small press had received a manuscript submission that seemed unusual compared to the hundreds of others they sort through each week. Although the manuscript was remarkably tidy in terms of compliance with grammar and style, the author’s word choices were odd in places, almost but not quite appropriate for their context. Much of the characters’ dialogue lacked emotion. Even the query letter that accompanied the submission felt somewhat strained.
On a hunch, the acquisitions editor ran a portion of the manuscript through an AI content detector and discovered that the book had been produced by an AI writing assistant. After Googling the name of the “author,” it was discovered that this person had already published a handful AI-generated novels, most of which were available for purchase on Amazon. As far as anyone could tell, none of the books’ listings disclosed that they had not been written by the person whose name appeared on their covers but had instead been created by a computer.
Several days later, I came across a post on the page of a Facebook writers’ group. The poster, who works for a children’s book publisher, was lamenting that many of the submissions they had recently received had been AI-generated. “Now is a great time to submit your book ideas to us,” she wrote. “We’re looking for stories written by real people.”
These posts and others like them have launched lengthy discussions among authors, writers, and editors and have raised a lot of questions about AI and the future of writing and publishing.
Currently, there is no definitive answer as to who owns the rights to AI generated content. Whether it will be the person who came up with the idea for a book or story or the owner or developer of the AI technology used to turn that idea into content still remains to be seen.
At this time, there is also no requirement that publishers disclose whether a book or other type of written material offered for sale to the public was generated by AI. Whether consumers have a right to this information has also yet to be decided.
Surprisingly, there is also no real consensus about whether putting one’s name on a piece of AI-generated writing and claiming to be its sole author is plagiarism.
Perhaps most important are the financial considerations regarding AI-generated content. Could or should an author using AI for any purpose (content generation, editing, proof reading, etc.) ever be obligated to share royalties with the owner or developer of the AI service they chose to use? If an AI-generated book or story were made into a film or other subsidiary content, who should be able to profit from it?
These concerns and others like them will eventually need to be resolved by lawmakers in the form of new regulations. If the business of writing and publishing is to remain fair, ethical, and intact, it’s essential that writers, agents, and editors, rather than corporate entities, have a say in the way these new laws are written. The recent acquisition of audiobook distributor Findaway Voices by Spotify and the company’s subsequent requirement that authors give Spotify an irrevocable, royalty-free license to their work, including unlimited royalty-free subsidiary rights, is just one example of why this is important.
One way that publishers might be able to help protect authors is by distinguishing books written by people from their AI-generated counterparts. Voluntarily labeling books and other published work with a logo similar to the “non-GMO” symbol that appears on some food products would be a good start. But in order for this to work nearly all publishing companies would have to agree to it, especially the largest and most influential.
It’s essential that the voices of authors and other creative people be heard before decisions on these matters are made for us. Talk to other writers, editors, and especially publishers. Contact your U.S. House representatives and state’s senators and let them know about your ideas and concerns.
Other helpful resources may include writers’ organizations such as The Authors Guild; the Romance Writers of America; the Science Fiction and Fantasy Writers Association; the Society of Children’s Book Writers and Illustrators; and the National Writers Union.
If enough creative people speak up, I’m hopeful that we’ll be able to turn the tables in our favor.
What are some ways that authors, writers, and other creative people can work together to ensure that their work remains relevant and profitable in the midst of AI-generated content entering the marketplace? What is the best way to organize such efforts in order to ensure they are effective?
Emilie-Noelle Provost is a writer, editor, and author who lives in a northeastern Massachusetts mill town with her husband, daughter, and three crazy rescue cats. She is an avid hiker and a fan of European crime dramas.
Emilie-Noelle has worked as an editor for four magazine titles and has written and published hundreds of articles. She is the author of two novels. The Blue Bottle, a middle-grade adventure with sea monsters, was published by North Country Press in 2018. Her second novel, The River Is Everywhere, the coming-of-age story of a Franco-American teenager, was a finalist for a 2023 National Indie Excellence Award, a 2023 American Fiction Award, and a 2024 American Legacy Book Award. The novel was published in March 2023 by Vine Leaves Press.
She is currently working on a series of novels set in Île d’Orléans, a historic island in the Saint Lawrence River near Québec City.
[coffee]
Thanks for this update, Emilie-Noelle, and for the action items. There are also the rights of those whose copyrighted works were used to train the AI.
Thanks for this post, Emilie. I don’t know why I’m surprised that someone would submit an AI-generated novel for representation, but I am. My second novel was among those used to train AI without consent; I did sign on to a complaint early on over that but I have no idea what, if anything, will come of it. Strange new world, and it’s important we keep up with its rapid changes as we can. Thanks again for your post and list of resources and potential action steps.
“Currently, there is no definitive answer as to who owns the rights to AI generated content. Whether it will be the person who came up with the idea for a book or story or the owner or developer of the AI technology used to turn that idea into content still remains to be seen.”
Maybe we can question the idea that everything is owned by somebody.
Who owns the dance of light and shadows in my home office right now, with the window open and the tree by the window moving in the wind? As far from AI as you can get—but someone designed the house, someone made the window, someone planted and pruned the tree, my spouse and I put the curtains up…
The copyright office in Washington already has a clear policy: AI generated works are not eligible for copyright protection, only human written works are.
In a traditional book contract, the author grants the publisher exclusive publishing rights (with limits and exclusions) and the publisher relies on, and indeed registers the book, for copyright protection.
AI generated works do not meet that requirement. Self-published works are a different matter, no such promises are made but no such protection is available either. Your AI written novel can be ripped off and there’s nothing you can do about it.
In a sense, seeking representation or traditional publication for an AI generated work is fraudulent. I wonder how long it will be before there are suits brought by the industry for that and, really, for wasting the time of agents and editors.
What a helpful post, Emilie-Noelle. Thank you. I think authors have to deal with this individually as well as from an organized group. For myself, I’d like to know what would be an appropriate line of text that authors can add to their work (beneath their copyright symbol?) that claims the work is not AI-generated. Something like “This story was not AI-generated” isn’t appealing and sounds pretentious. I’d rather say something positive like “This story was human-generated” but that sounds terrible too. Do you have any thoughts on what an author can do to let readers (and editors/publishers) know that the work they are reading was written by a person? Is anybody out there adding a statement to their work to clarify? I’ve put statements in my cover letters but these days, it may be necessary to identify AI written material from non-AI at the front of the story or book for the readers.
Paula, what I do is a statement to the effect that “No generative AI was used in the conceptualization, planning, or drafting of this work. No permission is given for the use of this material for AI training purposes” on my copyright page.
Hi Joyce. Your statement is quite fine! May I use it? It has a clear professional ring to it and covers all the necessary elements.
Absolutely, Paula! Go right on ahead–and thanks for asking.
Good idea, Paula. Works of historical fiction use a disclaimer to the effect that people, places, and events were real, but scenes, dialogue, and other details are products of the author’s imagination. I can see publishers requiring a similar statement.
I love this, “It’s essential that the voices of authors and other creative people be heard before decisions on these matters are made for us.” This is a unique time to gain a seat at the table and it will pass. All of this is happening in the context that 1) most opining, even in Congress, are behind as research pushes us towards further advancements and 2) that AI’s ability to write “stories” is a tiny, tiny part of the capability of AI though it gets lots of media. I recently heard a writer say AI hallucinations are great for creativity, but they aren’t- they are a sign a model isn’t working as intended which is concerning. I bridge both worlds – AI and writing – perhaps neither competently, and hope the writing community will understand more of what actually happens when an AI generates text so its voice is most effective.
Erma, you are right that the application of AI to fiction writing is a very small part of AI’s impact. If you’d like to delve, have a look at the comprehensive European AI regulation that is making its way to law over there. It is a national shame that in the USA we do not have anything remotely close to this level of thinking about AI.
https://www.europarl.europa.eu/news/en/press-room/20231206IPR15699/artificial-intelligence-act-deal-on-comprehensive-rules-for-trustworthy-ai#:~:text=This%20regulation%20aims%20to%20ensure,risks%20and%20level%20of%20impact.
I appreciate the issues you raise here. I teach (as well as write) and while I have thought about AI-generated material (papers, translations) submitted by students as plagiarism, I just hadn’t thought about the fact that published AI-generated works would of course be the same thing. I guess I hadn’t, before reading your post, really thought beyond AI-generated web-postings.
Hello Emilie-Noelle for your timely, informative post. I suppose the time isn’t far off when AI programs will have mastered all known structures in print, and replace humans. But for now, I like your “non-GMO” idea. My one encounter with AI occurred when a publisher licensed by The Peace Corps contacted me about a planned publication of RPCV memoirs (Returned Peace Corps Volunteers). The idea called for phone interviews in which Volunteers would memorialize their recollections of their first day as a Volunteer. I was happy to participate, and on the day of my phone exchange I happened to be in good form. I recounted the truly bizarre details of my first day as a member of the first-ever group of Volunteers to be sent to Micronesia. Afterward, I felt good–until I was sent a transcript of the phone results. It bore very little resemblance to what I remembered saying. When I called to ask about this, I learned that all the phone interviews had been fed through an AI program. The end result was generic, stripped almost entirely of the specifics that every WU member knows make up the essence of what we write. The person I spoke with told me I was not alone in being angry and disappointed. She promised to withdraw my entry.
Thanks for your post and all the responses. I will cling to what Don wrote. He and others in his position will become our protection from AI. It is a frighting prospect that we have long read about, maybe thought about…that machines will take over our ability to control words, content, even our emotions. Knowledge is power…beware the power of the machine.
More insults from the universe. Lots to think about. Therese says she’s surprised that someone would even submit an AI novel. Think of someone like George Santos. The point is not to write; the point is to be an author. And Barry’s story about the casual AI rape of his Peace Corps memoir shows the clear lack of understanding–even among people with good intentions–of what writing IS. I’ll bet the person behind that decision is someone who brags that they don’t have time to read.
I was surprised and encouraged, though, by the number of “AI content detectors” listed in Emilie-Noelle’s link. I didn’t know there was such a thing.
I think the problem in the near future is going to be one of sheer tonnage. Thousands of manuscripts are already pouring over the transoms of agents and publishers, and that’s just the real books. Gatekeepers will mean little when the gates are completely buried in good-looking, mass-produced “passable prose” books.
PS: They’re coming for movies, too. Soon we can watch Humphry Bogart fight Spiderman. Not really kidding.
You’re right, Michael, to imply I shouldn’t be surprised in today’s world.