Author Mark Lawrence says he will stop holding the cover contest that’s long been a feature of his Self-Published Fantasy Blog-Off competition after the latest winner proved to be AI-produced artwork, despite Lawrence’s rule excluding such covers from consideration.
As Lawrence explained in a blog post ”AI or not AI? That is the question” the SPFBO contest entry form asked each author “Is the cover wholly or in part AI generated” and “any answer to the AI question other than ‘no’ meant that the cover wasn’t considered for the contest. The author of the winning cover answered ‘no’”.
But when artist Sean Mauss’ cover for M.V. Prindle’s novel Bob the Wizard was selected by the SPFBO judges as the 2023 winner, several artists on Twitter quickly identified it as AI-generated.
The book’s author, Prindle, disagreed:
Contest sponsor Mark Lawrence asked the artist for confirmation. Also, “I emphasized privately to the artist that if it was true, deliberately or through some mistake with the form, they should just come clean and we could all move on. But they were adamant that no AI had been involved. I gave multiple opportunities to U-turn on this.”
Sean Mauss sent Lawrence several preliminary sketches and photos with an explanation of how he’d created the cover with Photoshop, as well as the many-layered Photoshop file for the cover. Lawrence shared some of this on social media. Although people in his SPFBO Facebook group found Mauss’ evidence persuasive, artists on Twitter soon showed why Mauss’ story could not be true.
Artist Rue Sparks has posted a 24-tweet examination of the image files submitted by Mauss, with analysis to show why they could not have been created in the way he claimed. Thread starts here. Here’s an excerpt:
Mark Lawrence said yesterday that “in response to my increasing distress over the situation” Sean Mauss removed his cover from the contest. Lawrence bumped the other finalists up and declared a replacement winner, then announced, “There won’t be a cover contest going forward.”
Meanwhile, a couple of other authors who used the same cover artist have launched a GoFundMe to pay for replacement artwork: “Cover Art Fiasco”.
We’re Clayton Snyder and Michael Fletcher, authors of Noryslka Groans and various other novels. Recently, we paid an ‘artist’ for a cover to our new novel, the conclusion to the Manifest Delusions series. After paying and accepting the art and setting a cover reveal date, we discovered it was AI generated, against our explicit wishes. We were lied to and scammed. We cannot in good conscience continue forward with the current art when there are living artists who need to make a living, and since indie authors aren’t exactly swimming in cash, we’re turning to the community in hopes of rectifying this.
Mauss has now taken down their Twitter account.
Once the original issue was sorted, Twitter toxicity expanded in an attempt to claim other victims.
One participant in the Twitter discussion got overheated and snidely wondered if Prindle’s book was written by ChatGPT.
At the other extreme, someone attempted to retaliate against an SPFBO judge, blogger CraigBookwyrm, who’d been critical of Mauss.
[Thanks to Anne Marble for the story.]
Discover more from File 770
Subscribe to get the latest posts to your email.
Wow.
This keeps getting more disturbing.
Argh!
Very sad that one bad apple can ruin a good showcase for many. The cover competitions were one of my favorite features of the SPFBO.
This has been a mess — with a lot of artists now believing they are being attacked by writers. But then… maybe avoid accusing a writer of using AI to write his book without a shred of proof?…
I’ve seen another argumentative person claim the novel was written by AI simply because he… did not like the writing style I guess? WTF?
Maybe that old piece of writing advice where you mailed yourself a copy of your manuscript will have to come back. It can’t be used to prove your copyright, but you could use it as proof that you wrote your book. (Note: It might be easier to simply keep copies of your versions on your drive…)
Everyone out there–
Does “using AI” in artwork have anything to do with such apps as Dream Scope or Deep Dream? I’m curious.
A very good friend of mine routinely uses such apps to creatively modify her own images using blended textures and such (no details–I’m NOT a graphics person). Does this mean she’s using AI to produce artwork? She is a pro at using Photoshop Elements to make images from scratch; she has used Dream Scope (until it was discontinued) to easily do things that are difficult or impossible in Photoshop Elements.
As far as using AI to write fiction or essays with, I have no idea how I’d do that, and I wouldn’t want to, anyway. MS Word is about my speed. In my opinion, using AI to make fiction might be a good way to produce reams of formula fiction, of the grade which might pass muster with a half-cent-a-word-payable-on-lawsuit pulp magazine.
I know there have been wonders performed in the audio studios with AI–one thing that crosses my mind is a 21st-century recording by Rachmaninoff (I heard about this while listening to my local classical station). They took a 1930s-era recording done by Rachmaninoff (playing one of his own super-tough piano compositions) and did something magical to it to make it sound like Rachmaninoff had sat down in a modern studio to play the piece with the same artistry and emotion he performed with in the ancient studio.
But that isn’t cheating–the end result is Rachmaninoff playing Rachmaninoff, only using a modern digital setup rather than a wax disc running at 78 rpm, with no attempt to add or subtract notes, much less produce an entire new score.
@Jeanne Jackson
Rachmaninoff used a reproducing piano, and those have been converted to CDs. They’re worth the money. (They’re both called “A Window in Time”.)
(I heard a reproducing piano playing one of his pieces, once, at a museum. At then end, the 15 or 20 people around the piano applauded the invisible pianist. They’re that good: all the intonations are there.)
Jeanne, I think the distinction may be between whether software/AI is assistive or prescriptive,
Perhaps the proper simile might be that the prompt provided to an AI program is like an “elevator pitch” for a novel. Both are rough concepts, but the AI’s algorithms produces “content” only within the parameters of those algorithms, whereas the human mind is famously “fuzzy” and malleable, capable of producing both schlock and genius, instead of a dull and derivative result like the AI-writing samples I’ve seen.
It’s that fuzziness and malleability, that ability to choose different words or phrasing, to change the incidents or direction or PoV or any other aspect of a story, that makes every story a particular and individual creation of its particular writer in that particular time and circumstance, whether the story is good or bad. That humanity, or believable imitations of it, is what’s so sadly lacking in AI-written fiction.
Pingback: Authors: How to Make Sure Your Cover isn’t AI-generated? – Camilla Monk
Mostly c4c…
I bought the book in question as the elevator pitch sounded pretty good.
Jeanne, part of the problem of using AI to “produce art” or fiction, is that many of those working on it wish to use it to replace human artists and writers. It’s one of the things the WGA strike is about.
Randall, I agree with Bruce Arthurs–if you want dull and derivative (by which I don’t mean the mathematical sense of derivative, which can be reversed by an integral), you can get by with AI. But you get dull old garbage. As I said earlier, it’s about like pulp-magazine formula fiction, minus the creative genius of the likes of Jack Williamson, Edmond Hamilton, Edgar Rice Burroughs, or Doc Smith. I’m sure most people know (well, most smart people, anyway) that boring an entertainment customer base is a jim-dandy way to shrink it rapidly.
Jeanne, it’s certainly true that AI is dull and derivative . . . right now. But it is a type of algorithm that develops as it is exposed to more works by others, which its developers are doing their very best to do.
And as for dull products destroying the market, well . . . I agree that smart people realise this but: 1) We then have to ask how many intelligent people there are working in the creative fields, given some of the products that are produced, and 2) what seems dull and stupid to you or me might still find an audience of its own. If you doubt that, go back to the articles OGH posted during and after the Kerpupple and see how commenters reacted to the various works discussed then.
Jeanne, Dream Scope or Deep Dream appear to be tools used against already created work. I have not personally used them but from what little I saw online in my five minute search they seem to be filter applications that can be applied to images.
No different than the blur or smudge or “magic fill” abilities inside Photoshop or other photo editing applications and no one is claiming that is AI. The difference with these applications is that they use math and automation to apply the filter the creator wants, and where the creator wants it, with customization options to boot.
I would not consider this even AI assisted. It’s a specific tool built into an application.
But that’s just me.
-Sterling
Magic Fill, aka Content-aware fill: “Content Aware Fill uses AI technology to select and blend the best replacement pixels. The Patch tool allows us to select an object and choose pixels to put in its place. The Content Aware Move tool lets us move a selected object, integrate the object into its new home, and patch the space left behind.”
Okay, so Cliff is claiming it’s AI, but Cliff is also correct. 8^) I didn’t write this out as detailed as necessary, so that’s on me.
My main point to Jeanne (and the internet people) is that these filters/tools built into the program are simply that–tools. They are leveraged/directed by the artist on existing content to manipulate imagery. They may “create” new content in a magical way but only within the scope of the image already presented and at the direction of the human.
AI’s like Midjourney and such are causing all this mess because it is a completely new creation of imagery, albeit learned through works already created by humans. But it’s way beyond derivative work at that point. “This cat/horse/dog/person does not exist.”
This is a fascinating topic and one that will not go away anytime soon, if ever. Unfortunately, I doubt Legislators will be able to come together in any meaningful way to address this issue, considering how beholden they are to other interests, or simply can not think deeply enough about the topic to project into the future.
And voluntary industry standards will mean little for those who create these AI programs outside that industry.
In the end it may simply be a matter of self-policing (“mob” ?), which we are experiencing at the moment.
It’s gonna be a while before the dust settles. Joe Scott did a great AI primer video this weekend (28May2023) on this topic if you’re interested.
Be well,
-Sterling