Generative AI programs are spurring plenty of ongoing discussions surrounding artistic integrity, labor, and originality. But that’s not stopping people from already attempting to profit from the systems. On Tuesday, Reuters profiled multiple individuals turning to text generators such as ChatGPT to churn out book manuscripts that they then fine-tune and subsequently sell via systems like Amazon’s self-publishing e-book platform. The rising deluge of chatbot-assisted stories is now so bad that it’s even forced a temporary submissions hiatus for one of the internet’s leading science-fiction magazines.
According to Reuters, Amazon’s e-book store includes at least 200 titles openly listing ChatGPT as an author or co-author. Such titles include space-inspired poetry, children’s novels about penny-pinching forest animals, and how-to tutorials on using ChatGPT to supposedly improve one’s dating life. “I could see people making a whole career out of this,” one of the AI-assisted authors told Reuters.
[Related: No, the AI chatbots (still) aren’t sentient.]
Because Amazon currently has no explicit policies requiring individuals to list generative text programs as authors, the library of AI-assisted titles is likely much higher.
“All books in the store must adhere to our content guidelines, including by complying with intellectual property rights and all other applicable laws,” Amazon spokeswoman Lindsay Hamilton told Reuters via email.
But although AI-assisted titles are proliferating in literary markets like Amazon’s Kindle Store, other outlets are being forced to halt all submissions in order to develop new strategies. In a blog post published last week, Neil Clarke, publisher and editor-in-chief of popular science-fiction website Clarkesworld, announced the site would be pausing its unsolicited submissions portal indefinitely due to an untenable influx in AI-assisted spam stories.
[Related: Just because an AI can hold a conversation does not make it smart.]
Clarke revealed in their post that spam entries resulting in bans from future submissions rose precipitously since the public debut of ChatGPT. Within the first 20 days of February, editors flagged over 500 story submissions for plagiarism. Before ChatGPT, the magazine typically caught less than 30 plagiarized stories per month. While there are a number of tools that can help detect plagiarized material, the time and costs make them difficult to utilize for publications like Clarkesworld operating on small budgets.
“If the field can’t find a way to address this situation, things will begin to break,” Clarke wrote on his blog. “Response times will get worse and I don’t even want to think about what will happen to my colleagues that offer feedback on submissions.” While he believes this won’t kill short fiction as readers know it—”please just stop that nonsense”—he cautions it will undeniably “complicate things” as opportunists and grifters take further advantage of generative text’s rapid advancements.