Polling the Pollsters

The other day Book Riot published the results of a survey of Book Riot readers on What Counts as “AI-Written”, and I am fascinated by some of the results around AI editing. 

It left me wondering if a follow-up survey to dig in a little deeper would help. I’ve been personally trying to figure out where the lines really are these days. You can see that in my last post. This survey couldn’t have been better timed for me. I have friends and acquaintances in various author communities on all ends of that spectrum. And for the record: I’m in the “no to generated text, but meh on editing, and good luck trying to avoid it in research” camp. Plus, I’m in the “good luck avoiding support of the mega corp billionaire oligarchy, but try anyway, and not just by shunning the LLMs but also the machine trying to turn your art into <cringe> content at every turn, camp.

I regularly read and see extreme statements like “there’s no ethical way to use LLMs, period.” When, from a purely technical POV that’s simply not true. Open, ethically curated, models exist, local desktop hardware can run them. It’s not impossible to use an LLM outside of the billionaire industrial complex. They can be used as tools in a bigger picture, kind of like how “Photoshopped” was a dirty word in photography until it wasn’t and until all the artists found the right way to use it as a tool.

View on Threads

SO I’ve been dabbling in that space, building a local editor that uses one of those open LLMs. I did things like using a lot of old-school processing to find some easy patterns in writing (Oh god, the Regex…) and only using the LLM when it was actually useful. I built strings of prompts that don’t tell the author what to write, only pointing out craft lenses, like “You go 12 chapters here without raising any tension or advancing character arcs, you asked about places to cut down your 250000 words. Maybe look here.” 

Which…first of all, RUDE, but also not wrong, and I’d ask you not to call me out like that in writing.

While I see no problem with that, I think I’m reading between the lines of folks dead against LLMs. I think they’re saying they’re dead against supporting those billionaires stealing work and actively putting people out of jobs. They’re dead against the endless “Now With AI! You can ask our AI now! You want to ask out AI. Right? Right!?” embedded in every. single. flucking. app and site… They’re dead against the people flooding the Internet and ebook markets with slop. They’re dead against the grifters and scammers trying to pass slop as real artist work to make a quick buck. LLMs just happen to be an easier target than KDP. 

So how can we have this conversation and get people reading between their own lines? How can we as an author community fight back against the companies and people devaluing our work, without attacking each other because one author used more than another was comfortable with? Even when that work contains zero slop, even when all the research is verified, even when the LLM in question wasn’t trained on stolen work.

Thanks for reading, if you have any suggestions for how or where this conversation can happen, I’m happy to help.