www.semioticstandard.com Open in urlscan Pro
2a06:98c1:3121::3  Public Scan

Form analysis
0 forms found in the DOM

Text Content

 * Home
 * Previous
 * Photography
 * About

 * Home
 * Previous
 * Photography
 * About

 * 


THE SEM10TIC STANDARD

R. Leigh Hennig. Horror author. Editor.




UNITING AGAINST AI IN FICTION

February 21, 2023  /  R. Leigh Hennig

Few are unaware of the meteoric rise in AI-generated content. But have you heard
of people using it to generate fiction, spam your favorite genre magazine
thereby crushing their first readers under mountains of spam, and forcing them
to close to submissions?

Yeah, well, it’s happened. Clarkesworld is now closed to submissions.

Already our small markets have to fight for survival, and now they have to put
up with this?


TLDR: I’M FORMING AN OPEN ADVOCACY GROUP, COMPRISED OF PUBLISHERS, EDITORS,
FIRST READERS, AUTHORS, AGENTS, DEVELOPERS, DATA SCIENTISTS, AND OTHERS
CONCERNED WITH ADDRESSING THIS PROBLEM. WE ORGANIZE, DISCUSS, PROPOSE, AND TAKE
ACTION.

It’s not that automated systems are spamming submissions queues, and it’s not
that editors are going to be ‘tricked’ into publishing a story written by an AI
(we’re not there yet, but we may get there eventually). The problem is that real
people are the ones who are submitting, and their junk submissions are burning
through all the cycles that slush teams have.

Take a look at the numbers Neil posted for January and February. Here are the
number of people Clarkesworld has had to ban for bogus submissions:

Source: Neil Clarke

They’re orders of magnitude higher than they were previously. That’s clearly not
sustainable. But the problems go deeper than that. Consider the following:

 1. Slush readers and editors can pretty easily tell apart a good story from a
    bunk one written by AI

 2. Bunk AI submissions sent by humans is skyrocketing, drowning the slush
    teams. They effectively function as a human-powered DDoS against first
    readers.

 3. Current AI-detection tooling is not keeping up. First, humans are submitting
    AI-written stories (bots are not submitting most of these stories, as may
    have been speculated), which can make them harder to filter out. Second, the
    tooling that scans for plagiarism or AI-written stories is imperfect,
    susceptible to false positives and misses.

 4. Because tooling to filter these bogus submissions out isn’t keeping up,
    markets are considering other means to protect their submissions queues.
    Those methods should be pursued as well, but some of what I’ve seen proposed
    lead us to a very dark place. For example: known-author lists.

One of the worst ideas I’ve seen so far to combat AI-driven submissions is by
closing submissions to only those authors who are already known. A curated list,
or database. This is an awful idea for two reasons:

 1. How can you possibly ever maintain such a database? Information would become
    stale very quickly.

 2. This will have a chilling effect on new authors, especially marginalized
    ones. Already they face an awful, uphill battle in getting published, and
    the reasons for this are myriad. People have been fighting that battle for a
    while now. But then if they had to also face access lists? This idea is like
    using a bullet to cure a cold. Talk about silencing voices that desperately
    need to be heard before they even get a chance!

It’s not enough for us to simply identify the problem and hope someone else
comes along to fix it. We need awareness, yes, but we need more than that.


WE NEED TO ORGANIZE. THEN WE NEED TO TAKE ACTION.

 

We’ll build our own anti-AI tooling!

 

I’m not against AI. I’m against the abuse of AI in fiction.

I have a couple ideas as starting points:

 1. Organize a group dedicated to outreach, education, and feedback to the
    authors of such tools as ChatGPT. Get together and function as a kind of
    public advisory group that engages these companies (Microsoft with their new
    Bing Chat, You.com, Neeva.com, Google and their Bard AI when it comes out,
    the list goes on) and make noise. Get in front of them and tell them what
    the damage is that they’re causing. I know for a fact they have an interest
    in being a positive force in this area. We can be heard, if we choose to be.

 2. Reach out to the other companies that actually build anti-AI tooling and
    that have released products. Work with these people to help them refine
    their systems. Tell them how we work as editors, first readers, writers,
    publishers, etc. Educate them on how submissions are sent, what we look for
    when passing on a story in the slush pile, etc. Help them make better tools.

 3. Reach out to markets to help them understand their current submission
    process, and how it may be improved. Educate them on the tooling that’s
    available, and the resources they have. Make them aware of our efforts, and
    bring them into the broader conversation.

 4. Work with established submission tools (Moksha, Submittable) to help develop
    and integrate solutions into their products to help address these problems.


IF ANYONE HAS AN INTEREST IN THIS, DON’T BE SHY. DETAILS ABOUT A DISCORD
CHANNEL, WEBSITE, FORUMS, AND OTHER COLLABORATION TOOLS WILL COME SOON. IF
YOU’RE INTERESTED, PLEASE:

Fill out this survey

Join us on:

Discord

Mastodon

Twitter


0 Likes
Share

Comment 0 Likes
tags / writing, blackjack, hookers, ai
Newer  /  Older