AI Slop: One star reviews
Most authors have had one: a one star review, together with hurtful words like ‘slop’, ‘unfortunate’, and ‘confusing’. If the reader didn’t like the book, they make up things about it and they even accuse authors of using AI when they haven’t. These reviews aren’t helpful. Some authors complain they won’t pass a plagiarism test because even though the concept, characters and narrative are from a human mind, it’s already on the LLM site.
The one star review is not a critique of the book, or of the reading experience for that particular person. Authors can’t please everybody; that’s why some authors ‘write to market’, but that’s another topic. The reader is driven by fear of AI, and the way the world is changing rapidly in ways they don’t like. I aim to tell you how you can judge if an author’s used AI in their writing process in context and a bit about where publishing is now, so you know the context authors are working in right now.
Intro
The age of fiction text by AI is on us. Many fear AI in general, but it’s here now and we just must cope with it as best as we can. This isn’t a discussion of the ethical and social impact of AI, the scope is fiction texts. For convenience, I use the term ‘AI’ to reference any of the machine tools mentioned below.
What does this mean?
It means, as an author, there is writing software that generates the word count, chapter by chapter, or scene by scene, to a series of author prompts, and editing software that covers functions such as structural edits through to proofreading and spellcheck. This means (to a degree decided by every author), they can allow AI to write their book for them. Of course, many authors write the old fashioned way, grinding out every word by hand.
It means, as a reader, you might not be getting the fiction concept (or the thoughts) of an individual first-hand. ‘Might’ in that sentence is doing a lot of heavy lifting, and there’s the rub. How can you tell? The books might be filtered through AI, spewing out words and phrases that have been ‘scraped’ from thousands of successful books and famous authors. Also, there are faults with the machine process (as it stands at the moment) which diminish the authoring and the reading experience.
Why this matters is discussed in further detail below. It’s not necessarily bad, though if you’re interested in books from any angle you should be aware, but first, I must make clear: this is my personal opinion. I’ve gone to every effort to check my sources, but they won’t always be accurate and I’ve also made assumptions which might be more opinion than fact.
In this series of posts, I analyse some of my recent favourite books, and look for clues that AI has had some input in the work, though I should make clear, again, (Disclaimer) I have no proof the authors have used whatever machine process in a book. It’s just guesswork, but from the volume of fiction I consume, I’ve learned to recognise certain ‘tells’ that AI has been used in the process.
Having said that, let’s dive in.
What is AI text in fiction?
Writers use software (or apps) to generate story, plot, characters, or text, in a genre they’re writing, according to prompts the author provides. There are popular examples online. Just key in ‘ai fiction writing’ to the search engine and you’ll find some. You must pay for these. The output is varied, depending on the skill of the author at constructing the prompts.
The Copyright Conundrum
The Large Language Models (LLM) that underlie this software are produced by ‘scraping’ the text of published authors which have been fed into the system like some enormous digital mill in which grain is put in and flour comes out at the other end. Some would call it ‘stealing’. Copyright law (which differs country by country) has been disrupted by it and the USA Library of Congress in Washington, the largest copyright registration holder in the world (I think), hasn’t ruled on it at the time of writing, though there’s recently been a case in the US brought by authors against one of the biggest LLMs in the world where the court ruled for the plaintiff upholding copyright and the defendant must pay out compensation, and presumably, costs.
Authors can check whether they’re on the list here.
More Traditional Writing Apps Here, Some Include AI
There are also editing apps which I don’t use in my author process, so this will be a short paragraph. Dave Chesson, founder of Kindlepreneur, explains the different writing and formatting apps and choices here. Also, in addition to word processing software, there are online thesauri, translators and spellcheckers, including that famous one which I found so annoyingly intrusive I had to uninstall it.
As reader, how can you tell?
Faulty story logic
This is the most important. AI fails on story logic. As an author, it’s a particular bugbear of mine. A reader picks up threads to understand the story; they follow a series of clues like a breadcrumb trail through a forest. AI sometimes contradicts what it previously established and leaves a reader frustrated and confused. They might give up on the book and make the author vulnerable to being slung in the dreaded ‘Did Not Finish’ (DNF) category on goodreads.
Repetition
To move onto repetition, the point is — in my opinion — the author’s style doesn’t always come through. It results in characters sounding ‘samey’ and dialogue styles which aren’t different from another character’s. Worse than that, repetition is a big problem ranging from concepts being repeated to the style of the syntax, duplicate phrasing over and over, similes and metaphors, and in the worst cases, text being duplicated word for word.
Same vocabulary, metaphors and phrasing
Readers can tell by finding the same clusters of phrasing repeated and the same metaphor(s) applied to everything. This is tiring, can slow the pace, and makes the text hard to wade through.
Mistaken Meaning
Often, words are homonyns, signifying that words mean something different but sound the same. An example is ‘taut’ and ‘taught’. The words sound the same but there’s a world of difference in the meaning. To my mind, this a sure fire way of telling AI input either in the generative text or editing tool. A human editor worth her salt would’ve picked it up. It might not annoy others but it knocks me clean out of the story, and that’s the point; a fiction has to immerse readers. That’s why they buy it.
I suspect authors don’t read through their final drafts any more given the pressure on them to produce more content but that’s what beta readers or proof readers are for.
If you find all of the above in a text, you suspect AI has been involved.
The most cited reason for detecting AI input is the ‘m dash’, like a long hyphen, but I don’t think that’s the case. More obvious are the choppy text, the annoying use (or overuse) of full stops (periods). Passages which should be gentler and more lyrical are subject to the same treatment by AI, leaving no room for a build-up of tension and a faster pace.
Confusion and Uncertainty
That’s what I feel when I read faulty story logic, along with rage and disappointment to varying degrees, and I assume other readers feel that too.
Uncertainty is also compounded by the suspicion that AI has been part of the writer’s process. The reviewer doesn’t know how much the story — from concept to finished version — is not the author’s but AI input, and this leads to doubt about star ratings and reviews. How much praise is due to that specific author or is the praise due to someone else?
A brief history of the publishing world
At some point, when the big publishing houses got bought out by greedy corporations, the bean counters cut out the editorial process. What the new publishing landscape demanded was a ‘clean’ manuscript from authors, a final draft in which all the editing had been done. What mattered was the bottom line, not producing a quality book that book lovers had been anticipating.
There used to be ‘the big six’, now ‘the big five’: Hachette Book Group, Penguin Random House, McMillan, Simon & Schuster, HarperCollins. A few. years ago, a contract with one of these and an author had made it. Now self-publishing is all the rage. You can gain insight into the lion’s den of publishing here.
Books as products
The current scenario: books are treated like products. The online platforms are partly to blame, though unknown authors have established themselves by publishing on Amazon through its Kindle Direct Publishing (KDP) platform. Together with the new formats, like epub and mobi, the online platforms came about with the hardware, tablets like Kindle, Kobo, and Apple Mac’s ipad.
People have Changed the Way they Consume their Content
Having said that books are treated like products, we must also be aware that consumers have changed the way they take their information. Reading capability (or habit) has dropped by 40% in the US. This is partly because of the teaching system we have these days, but more importantly, consumers prefer videos or podcasts. Also, attention spans have shrunk, supposedly due to the popularity of mobile phones. A decreasing number of consumers have the will or the patience to get through an entire book. There is even a claim we’re the ‘dumb generation’ now.
Authors are Content Producers
In response, authors must not have a flair with the pen, not be skilled with words, but skilled in prompting the AI tool they use. Many authors choose self publishing over the Big 5. They publish on Apple Books, Google Books, Barnes & Noble, Waterstones and other online platforms through Draft2Digital. They promote their books on Facebook, through Amazon ads, on TikTok, and online through a variety of channels.
Different functions of editors
Specifically, the bean counters cutting out the editing process, which is an essential part of producing a quality, unforgettable book. You might say nobody cares, but the consumer who’s spent cash on a book, in whatever format, to read on the beach, on a train journey, or to while away hours at an airport, or simply because they love books, cares.
The editing process includes: Structural or developmental (incl. story logic), copy editing, line editing and finally proofreading. Many authors use AI tools for these, and in my view, they’re not up to it. Reading and writing fiction is a human, emotional experience and AI doesn’t have emotions; it’s just a machine. Of course, AI might improve and it probably will, but we’re still a long way off machine consciousness.Find more about editing here and here.
Online Psychopaths
You might have seen reports that AI has passed the Turing Test, but in my view, that means the machine has learned to mimic humans at the required skill level; it doesn’t have a soul and emotions. An essential basis for fiction includes human emotions to varying degrees, that’s the point, otherwise why write it? AI is online psychopaths, intelligence without the human emotions and awareness, making a binary decision by matching the request against defined criteria, rigid, inflexible, unable to judge mitigating circumstances.
Stone Gray
Science Fiction Romance books I read within the context of AI by this author. Stone Gray is a new author to me and I found her because Amazon suggested her titles. I test new authors (to me) by trying out their books in Kindle Unlimited. I also check the ‘Look Inside’ for formatting and other things (button underneath the cover image on the sales page).
Taken for Tribute
Taken For Tribute. I wouldn’t have guessed AI was part of this author’s process (allegedly) until Chapter 19 when dialogue was repeated, and also the story claimed the protagonist went into the Observation Gallery (on the spaceship) for the first time late on in the story but she’d spent time hanging out there in Chapter 6. Faulty story logic. I am making a huge assumption here, but a human would have caught such a howler, an editor or beta reader. I note the book has been marked down in the reviews for AI but I stand by my review on Amazon and I’m not editing it. I really enjoyed the story to this point and I enjoyed it until the end. The comments include spiteful ones about AI, especially the one-star rating ones. I ignore one star reviews because they’re not helpful and they say more about the reviewer than the book, and about some online booksellers treating books as products.
Her Other Books in Science Fiction Romance
I’ve read The Wrong Star and The Alien Envoy series (2 books). The concepts of all her science fiction romance are brilliant, the characters distinctive, the structure and world-building I haven’t seen before. My enjoyment of the stories was marred by mistakes which could or couldn’t have been made by AI but that’s not to say the concepts weren’t new (to me) or brilliant.
More Science Fiction Romance
Darren, Book 1 of Harmonious Mates Agency series, by Ember Hunter. On Amazon. More Science Fiction Romance.
Conclusion
I admire authors hugely. It takes enormous physical and mental skill, creativity and stamina, to write a book. Self published authors need to be skilled at business and promotion on top of that. Even if you have a contract with a publisher, you need to be savvy with promotion. AI is used to produce music, art, works of fiction, and other texts as well. I worry about the ethics of using it and the social impact, but that is not discussed in this article. It’s the huge subject of our time and this article’s scope was the extent authors use AI tools in their published works, how readers can tell and readers’ reaction to it.
No AI was used in the planning, structure and text of this article.
Sources
Header image by the mighty picjumbo.com
Advice about Editors: Bryn Donovan, CreativIndie.
YouTube Channels: Jared Henderson, The Guerrilla Publisher.
More sites of interest: Kindlepreneur
Recent Comments