So. AI, eh?
Like most folks of an artistic persuasion who are active on social media, I’ve been seeing more and more about advances and applications of AI technologies within creative fields.
In particular, talk of two apps—ChatGPT, which produces AI-driven writing, and Midjourney, which does the same with visual artwork—have been all over my Twitter feed of late, at times in eager celebration of their capabilities, but more often expressing concern around issues of copyright and compensation of the artwork that’s used as a base for the AI to learn/improve, academic dishonesty, and a general devaluing of human expression, effort, and experience.
These complaints came to a head on December 9 when design manager Ammaar Reshi posted a Twitter thread explaining how he used both ChatGPT and Midjourney to create the prose and pictures for a children’s book called Alice and Sparkle, which he then self-published to Amazon’s Kindle Direct Publishing.
I spent the weekend playing with ChatGPT, MidJourney, and other AI tools… and by combining all of them, published a children’s book co-written and illustrated by AI!
Here’s how! 🧵 pic.twitter.com/0UjG2dxH7Q
— Ammaar Reshi (@ammaar) December 9, 2022
This AI-assisted project, particularly the writing aspect, immediately sent me into a prognostication spiral as to what this could mean for the future of fiction publishing.
I don’t by any stretch consider myself a visionary, and I’m certainly not any kind of publishing professional, or even published at this point. But in this case, I saw a possible path forward quite readily and clearly, perhaps just so because I’m actively—and with no small amount of difficulty—still trying to realize my publication goals.
So what are my predictions for the future of publishing?
In short, I believe AI will be absorbed into the publishing industry, both trad and indie, in a huge way—and that it will happen fast. Perhaps as long as five years hence, but part of me thinks it could be as few as two.
AI-bstract art
I’ll start with art, for although I’m not a visual artist, it was this aspect that kicked off my entire line of thought, it being the aspect of Reshi’s project that garnered the greatest amount of criticism that I saw (which makes sense: children’s books are highly visual so the art is the first thing people see. So too is the cover-art the first thing people see on a novel).
welcome to the industry! Since your new to publishing I thought I might offer some feedback pic.twitter.com/ZfeBfOKsdM
— Corey Brickley Illustration (@CoreyBrickley) December 12, 2022
At the moment, AI-assisted art is kind of funky-looking, often throwing off all sorts of oddities like multiple sets of eyes or missing limbs. However, this will improve. This isn’t even a prediction but the very definition of machine learning.
AI images will get considerably better. And even if they don’t, most people will just stop caring, coming to prize convenience and low cost over quality the same way folks prefer MP3 or online streams of music on their phones over the vastly superior sound quality of CDs.
People will come to see AI artistic flaws as quaint idiosyncrasies of the medium—a new form of abstraction—assuming they notice them at all. The harshest critics of visual art will always be other trained artists, which the majority of the population are not.
Existence & insistence
Moving on to writing, in truth, I believe that AI-assisted prose is already being published, both by trad-pubbed authors and indies.
This is not necessarily a bad thing. AI may be a useful tool in helping people with something to say yet difficulty saying it have a voice (e.g. people with disabilities that make writing challenging; people writing in a language that’s not their mother tongue; people for whom the time constraints and stress of a hectic and demanding life leave little extra time for writing).
This, to me, speaks to the same general philosophy that underpinned the rise of indie publishing—the use of self-directed technology to eliminate gatekeeping and reduce inequality.
But AI-assisted writing is generally frowned upon at present, so not many authors will admit to it beyond conceding their use of grammar aid apps like ProWritingAid, which, but by a matter of degree compared to ChatGPT, is also an AI program.
I believe AI-assisted writing will become even more widespread, and also overt, largely because traditional publishing will force authors to use it to crank out titles faster. This will come about by way of increasingly shorter deadlines in contracts, particularly in genres where readers are prolific, like romance, YA, and with multi-volume SFF series.
More titles in the hands of prolific readers of course represents huge revenue increases for publishers.
Likewise within the indie sphere, where rapid-release is often already an industry expectation, higher customer demand for more titles will lead some authors to use AI to help speed the writing process.
Enforced partnership
A regular refrain of writers in online spaces, both sincerely and somewhat tongue-in-cheek, is how difficult the writing process is and how much we hate it. This sentiment will likely come back to bite us in the collective ass as traditional publishers present AI as a necessary solution.
If authors balk against either shorter deadlines, with or without an AI writing partner, publishers may cut out authors entirely and get employees to create AI-assisted works while on the clock.
And unlike with the average mid-list author, publishers will actually market these AI titles since they’ll have been relieved of the burden of paying advances, royalties, or cover artists1 (hello again, Midjourney—the inevitable next step where currently there’s already been an upsurge in cartoon-style covers, which are cheaper in not requiring a photoshoot).
To the credit of whoever art directed/designed, your changes are vast improvements to the original stock image, but also come on the fact that you had to draw on legs kind of undermines the premise that you didn’t know this was ai generated pic.twitter.com/N5NmHK2OFB
— Corey Brickley Illustration (@CoreyBrickley) December 9, 2022
Marketing will finally fit the bottom line. Indeed, AI will likely help there too, creating algorithmically-driven marketing copy for use on TikTok, Instagram, and whatever the next big thing in social media may be.
Mechanical translation
Right now, in my opinion, the text produced by ChatGPT is bland and emotionless. But for the undiscerning reader it gets the job done. I’ve seen the same quality/lack thereof produced by humans as well; indeed, with publishing’s strong focus on marketability of stories, artistry and innovation of prose is often overlooked.
The majority of people read for the story itself—the plot, the tropes, the actions of the characters—not the words that are used to tell it.
But again, the AI will improve.
When it comes to literary fiction, which does concern itself with prose, I believe it will take a bit longer for AI to produce comparable output to humans, and even then it will seem somewhat “off” under close scrutiny, lacking the depth that comes from human imagination and lived experience.
For that reason, as well as philosophical aversion within lit circles, I think quality literary and upmarket fiction will remain free from AI involvement longer.
Still, though, people will get used to AI litfic, perhaps perceiving it in the same way as a novel translated from another language.
Dividing lines of identity
While I do believe in the inevitability of AI’s integration within the publishing landscape, I don’t think it will be the sole product on offer, at least not for a while.
Currently, what makes a lot of stories interesting, noteworthy, and/or marketable is the identity/identities (often marginalized) of the author, especially when they write from their lived experiences.
As much as marginalized authors often lament being pigeonholed into writing about identity (and often in trauma-driven narratives), doing so, I believe, will become an important selling point for “all-human” fiction since AI doesn’t have an identity, or at least doesn’t have one yet.2
It may be that indie publishing, already a haven for creativity and innovation compared to risk-adverse traditional publishing, will become even more prominent through its promotion of all-human works. Some sort of indie-led all-human certification may develop in contrast to trad, which finds a more lucrative market with AI.
Also within indie, though, we may see more people who don’t actually consider themselves writers—folks like Ammaar Reshi himself—exploiting the indie infrastructure with trope-heavy, genre-friendly AI-assisted titles and series to make quick gains. This could continue for some time, leading to a whole new indie gold rush, until that market saturates and a new equilibrium is set.
AI & M-E
I could be wrong about all of this. I’d like to hope I’m not, but don’t really think I’m too far off. Which finally leads to the $64,000 question motivating this whole line of speculation: how will my publishing prospects as an unagented, unpublished, aspiring author be affected?
The short answer is that they’ll probably worsen, for everything in publishing that stands to make the industry more wealthy is generally worse for writers on an individual level, particularly those who have yet to break in.
The longer answer, however, is that I don’t really know. Right now, AI doesn’t seem like something I want to work with. I value my personal writing voice and individual story ideas too much.
But I’m also a very slow writer who’s not the sort that’s bursting with ideas for future WIPs, plus I have a demanding job that siphons off huge amounts of mental focus at my writing’s expense. I could well be the perfect candidate for AI-assistance in my writing. It could well be the thing that helps me achieve tangible success, and years from now, if I do actually manage to break in, the thing that keeps me on the in.
As I said before, I’m not a visionary, least of all within my own life, which is actually supremely limiting. It’s also the reason I never say never anymore, especially related to technology (I once said I’d never get a smartphone), and especially related to writing, which means more to me than most anything else in the world.
All I can truly say for now, for me, is not yet.
~
1Tor Books, whether they knew it initially or not (and whether they’re the only publisher currently doing so or not), is already using AI art on the cover of the forthcoming Fractal Noise by Christopher Paolini.
2Okay but for real I gave myself chills when I wrote this because that is definitely coming too.
A Frame Around Infinity 
As one of them (the literary fiction novelists), I hope there will be both a market for and a yearning for my kind of fiction – the kind where it matters who wrote it – for a long time.
There’s already plenty of stuff that reads as if it were recycled – and its buyers/consumers. They were never potential readers for me, because, well, because books produced quickly are going to be a little thinner gruel. At least thinner than I cn produce if you let me have a couple of decades. It’s a lot to expect an AI to be coherent for a novel-length piece. I haven’t been impressed yet – it always turns out the stuff generated is designed to fool people into thinking an academic produced it and people in the field are so used to bafflegab they don’t actually listen, parse, and try to understand it. At which point their brains would crisp.
I’ll keep trying to tell stories I want to tell – and leave the AIs to do the boring tasks of writing, such as turning the facts from a basketball into a ‘story.’ Or one more weather report.
Hope you’re getting time for your own writing, and that next year is even better.
LikeLiked by 1 person
I think that AI output will definitely improve, and if nothing else will give writers who care to (or are forced to) use it a base from which to work, molding it into better shape. I definitely believe that big changes are on the horizon, driven equally by a push to make writing easier (from an author’s standpoint) and cheaper (from a publisher’s standpoint) to produce. The only thing really stopping it now is philosophical aversion, yet many things to which society was averse are now wholly acceptable, both for better and for worse.
LikeLiked by 1 person
I create a plot and a style, and something else does the work? Hmmm. Tempting.
I think it’s often called ‘fan fiction’ – color me a bit skeptical of the results, which I will happily yield to if they are of acceptable quality. Considering, though, how much work goes into getting the details right, and how often I completely surprised myself, I wonder how much the human author will have to provide before, during, and after, to end up with Gone With the Wind.
Here’s an experiment to try: START with GWTW, and see what sequel an AI might ‘write,’ as the object produced by the one asked to do it is something I won’t read or even skim again because it was not anything like what I believe Margaret Mitchell would have written.
That would be a fair comparison to try every few years: a novel completed by someone else vs. a novel produced by an AI that was fed some limited version of the same future as a starting point that Alexandra Ripley had.
But fiction isn’t chess, which at least by definition has a ‘best’ way forward from many positions, though it quickly gets into numbers of paths that must be tried that defeat even the current computers. Fiction is currently, as it should be, heavily influenced by its author and their life.
Not saying it can’t and won’t, just saying ‘Prove it to me.’
LikeLiked by 1 person
A well-considered piece Janna. I hope this dystopian-looking future is further down the road than you say. I’ll live with the belief that no amount of AI will come near to generating the various emotions, complex detail and nuances that a talented writer can engender. There is though still a mountain of (in particular) indie stuff which is flat and mundane enough to qualify. Best of luck as ever with your own writing and a Happy Christmas.
LikeLiked by 1 person
Thanks very much, Roy. To be honest, I too believe that AI will likely only go so far in replicating human emotions in a convincing manner. But I also think that AI will evolve in ways that are utterly beyond way we’re able to imagine and predict, and that the art that results this will utterly fascinating in its foreignness (perhaps terrifying as well) and lead to whole new fields of media analysis and critique. Only time will tell. Happy Christmas to you too!
LikeLiked by 2 people