Ars Technica
A side effect of unlimited content-creation machines—generative AI—is unlimited content. Last Monday, the editor of the famous sci-fi publication Clarkesworld Magazine Office has partnered that he temporarily closed story submissions due to the large increase in machine-generated stories sent to the publication.
In a graph shared on Twitter, Clarkesworld editor Neil Clarke notes the number of banned writers who submit plagiarized or machine-generated stories. The numbers reached 500 in February, up from over 100 in January and a low baseline of around 25 in October 2022. The increase in banned submissions roughly coincided with the release of ChatGPT on November 30, 2022.
Large language models (LLM) like ChatGPT have been trained on millions of books and websites and can easily write original stories. They do not work autonomously, however, and a person must guide their output to a prompt that the AI model will then attempt to complete automatically.
Since 2006, Clarkesworld has published famous sci-fi authors and won several Hugo awards. Among sci-fi publications, it is known for having an open submission process and typically pays 12 cents per word. On its submission page, the publication says, “We are not considering stories written, written, or assisted by AI at this time.” However, that hasn’t stopped the number of submissions from skyrocketing, and Clarke explains that most get-rich-quick schemes.
“People who cause trouble come from outside the SF/F community,” WRITES Clarke in a tweet. “Mostly driven by ‘side hustle’ experts who claim to make quick money in ChatGPT. They drive it and deserve some of the disdain shown to AI developers.”
At press time, a quick search on YouTube for terms like “get rich with ChatGPT” and “make money writing with ChatGPT” returned several results, although we could not identify a video focusing on in Clarkesworld in particular.
Ars Technica
The problem with AI-written content is not unique to Clarkesworld. On Tuesday, Reuters wrote a report about the rise of Amazon’s AI-generated e-books. Reuters identified more than 200 e-books in the Amazon Kindle store that list ChatGPT as an author or co-author.
The influx of AI-generated content has left Clarkesworld in the awkward position of trying to keep the submission bar high enough to keep spammers away but not so high that it discourages unknown writers or writer from some regions of the world that may be unfair. targeted by geographically-based restrictions. In a series of tweetsClarke explained his situation:
We have no solution to the problem. We have some ideas to reduce it, but the problem will not go away. Detectors are unreliable. Pay-to-submit sacrifices are very legitimate authors. Print submissions are not possible for us. Various third-party tools for identity confirmation are more expensive than magazines can afford and tend to have regional loopholes. Adopting them is tantamount to banning the entire country.
We can easily implement a system that only allows authors who have previously submitted work to us. That would effectively ban new authors, which is unacceptable. They are an important part of this ecosystem and our future.
It’s worth repeating that until now, tools that purported to detect text written by LLMs had low accuracy rates (often returning false positives when tested with human-written text ), so they are not currently a viable solution. Despite these issues, Clarke said the magazine is not closing, and submissions will resume in the near future. But for now, the way forward is unclear.
“It just won’t go away on its own and I don’t have a solution,” Clarke wrote in a blog post Wednesday. “I’m tinkering with some, but this is not a game of whack-a-mole that anyone can ‘win.’ The best we can hope for is bailing out enough water to stay afloat.” In the meantime, Clarke encourages those who want to support the magazine to subscribe.