After a wave of AI-generated slop inundated their submission piles, Magazine editors like those at Asimov's Science Fiction and Apex Magazine published rules prohibiting submissions that used AI in any fashion. Here are their stated policies:
Asimov's Science Fiction Magazine:
"We will not consider any submissions written, developed, or assisted by these tools Attempting to submit these works may result in being banned from submitting works in the future. [This includes using ChatGPT for plot, character, setting, and other story development ideas.]Apex Magazine's policy is similar:
Apex Magazine does not accept submissions of any kind that were written, developed or assisted by tools such as ChatGPT. While we love publishing and reading stories about AI, we want them created and conceived by flesh and blood beings. Any attempt to submit AI-aided work may result in being permanently banned from our magazine.
Personally, because I take pride in both my originality and integrity, I would never put my name to something not entirely my own, unless I am sharing due credit with another collaborator/other collaborators. This includes anonymous collaborators such as AI--meaning, I would never put my name to something made with AI.
As for art or writing generated primarily by AI, I am cautiously fine with it being published as memes or ad hoc illustrations, provided it is never misrepresented, nor allowed to compete with human artists and writers. This usually means no human is credited or paid. Possible exceptions include sufficiently sophisticated and artful human-directed creations, including certain animations.
Setting to one side environmental and existential concerns, I have tried to keep an open mind about AI. I have even experimented with images and, very recently, with stories. I need to explain the latter very carefully.
The illustration above is Generative AI, for example. It loosely illustrates one of my stories, but I would never publish it as a book cover, though I see no harm in publishing it here. The story in question was written entirely and exclusively by me: story, characters, setting, dialogue and description, etc. But I did use ChatGPT in ways I thought were not "cheating". I'll let you be the judge:
The story involves human space travelers interacting with human colonizers of an alien planet. For plot reasons related only to internal logic, I wanted a planet which might take less than ten years to reach by a ship accelerating at 1-G for half the voyage, then decelerating for the other half. I asked ChatGPT to suggest a known, possibly habitable exoplanet within that zone. It suggested Ross 128b, which orbits the red dwarf, Ross 128, about 11 light years from Earth. Note that while I'm clinging to scientific plausibility wherever I can, I'm simultaneously taking as much slack as I require for the sake of story.
The scientific boundaries are arbitrarily drawn.
I asked ChatGPT what conditions on Ross 128b might be like for human colonists. It noted that the planet is so near, it is probably tidally locked, but allowed that a theoretically feasible, synchronous orbit (a local year would be 9.9 earth days) might allow convenient fluxuations in daylight. For purposes of description, it helped me visualize the relative size of the local sun in the alien sky. Being over-helpful, ChatGPT suggested the phrase, "like a hammered copper coin", which meant that description was now dead to me. I would have to "coin" my own description (and I did). The actual relative size and even the color of the local sun had no great impact on the story. The Ross 128b sky and landscape are glimpsed only briefly as a shuttle is landing on the surface. Had I not chosen to base the star and planet on real celestial objects, this detail would have mattered to no one.
To avoid confusion between the star Ross 128 and planet, Ross 128b, I decided the colonists would have renamed it. I also needed names for three colonies. I let ChatGPT generate suggestions--purely as placeholders, so that my writing flow wouldn't come to a halt while I spent half an hour thinking up realistic names. Of course, I swapped out all the placeholder names before calling the story finished.
When the story was done, I asked ChatGPT for editorial feedback.
My purpose was partly thoroughness, partly emotional need (writing is lonesome work). But here, let me testify, that the feedback it offered proved surprisingly adept. It pointed to redundant exposition (this information was already given, why have a second character repeat it?); it highlighted dialogue that dragged or deviated from optimal tone (levity here distracts prematurely from mounting tension), etc.
I have human beta readers who've always offered me this sort of feedback. Feedback from ChatGPT had gratifying immediacy, but it was a slippery slope, as I'll illustrate.
After I revised my story to incorporate the tidbits of advice I appreciated, the finished story came to just over 3,000 words. I looked at places to submit the story, and found one that only permitted stories of 3k words or fewer. So, I submitted the story to ChatGPT again and asked where it thought I should most easily trim the story. It offered several options. One of the best suggestion sharpened a bit of dialogue. But this was a trap: When you say the same thing but in fewer words, there is far less margin for rewording. Ergo, a temptation to accept the suggested dialogue verbatim.
And so, I did, but immediately afterwards, regretted this choice. I knew I was going to have to scrub out any phrases that belonged to ChatGPT. My final ask of ChatGPT was for it to help me identify any of its own work, so that I could replace it.
At last, the story is finished and it's all mine. But could I honestly submit it to Asimov's or Apex? Alas, their conditions clearly prevent it. [They are such a highly competitive market, it would be quite an achievement for my work to every be accepted anyway.] Even though I wrote the work entirely before asking for feedback, and even though no meaningful trace of ChatGPT remains in my final draft, I can't claim the AI didn't assist me. It told me what to remove.
Well, lesson learned. Months ago, when I first began experimenting with ChatGPT, I instructed it to NEVER create anything nor SUGGEST anything unless it was explicitly requested. (I had noticed that it's default was to end every conversation with a battery of suggestions.) The moment it offered a poetic description of Ross 128, I should have realized that this instruction had lapsed.
On the bright side, I have a new story and it's good. I don't regard it as corrupted. Other (less prestigious, lower-paying) markets exist where it is not categorically disqualified. But even if I never get it published in a magazine or anthology, at least this story will almost certainly be included in my future science fiction collection.
(Frankly, I largely empathize with Asimov's and Apex. I am repulsed by the idea of submitting to any place that doesn't forbid MOST AI.)
