award-winning Clarks World Magazine has helped launch the careers of science fiction writers for nearly two decades, and regularly features the work of Hugo Award nominees and winners such as Elizabeth Baer, Peter Watts and Catherine M. Valente. But now, in a very ironic situation, she finds herself battling the most sci-fi of modern trends: AI.
according to to recent articles According to Neil Clarke, editor of Clarkesworld, more than a third of submissions to the magazine this year were written by artificial intelligence and then submitted by humans who cheated. And it’s getting worse fast. The first half of February saw more than twice as many AI writing entries for him as for the entire month of January, Clarke said. Kotaku There were 50 today alone.
Since the article was written, Clarke has tweeted that the post is completely closed at this time. “It’s not hard to guess why,” he adds.
Clark said the decision to end the submission was made “at that moment” Kotaku By e-mail because the number was flooded this morning. “I could play Whack-a-Mole all day or close my submission and work on a legitimate submission.”
The rate of rise in this situation is very impressive. In his own blog post, Clark said he had to deal with plagiarism for a long time, but it wasn’t until the end of 2022 that the problem became rampant. And in his first month and a half of 2023, it escalated to such a magnitude that the magazine stopped entries altogether.
How can Clarkesworld tell AI-generated stories?
Clarke doesn’t explain in his blog how you can tell which entries are written by AI. This is for the very sensible reason that you don’t want to arm your cheats with information that will help you evade his detection.but he explained Kotaku Currently, it’s not that hard to find.
“The ‘authors’ we banned,” said Clarke. Their works are formulaic and of poor quality. But he also suspects there are already layers on top of these. However, “it is inevitable that the group will grow over time and become yet another problem.”
It’s not a problem Clark faces alone. The editors report that others in a similar position are facing the same challenges, and if it’s happening at Clarksworld, it happens wherever submissions for publication are possible. And while in most cases such submissions are removed simply because they are not good enough for publication, searching for fakes is an expensive and time-consuming process.
Clark said third-party detection tools that should be able to recognize plagiarized or AI-generated content are not the solution, given the number of false positives and false negatives, and indeed the cost of such services. Other short-term measures, such as regional bans on the regions of the world with the most fake entries, are also not the answer. As Clark notes in his article,
It is clear that business as usual is not sustainable and we fear that this path will lead to increased barriers for new and international authors. needs These people.
Of course, this is not an easy problem to solve. The pace of AI chatbot improvement is enough to jot down ideas for sci-fi short stories, and future tweaks will soon make them harder than ever to find. But it may be a long time before AI can create stories that are truly worth reading. I asked Clark if he thought this was likely to be the case. I didn’t want to guess.
But this is not very comfortable. “There are still ethical concerns about the means by which these works were created,” Clark said. Kotaku“And I wouldn’t even consider publishing a machine-generated work until such concerns have been remedied.”
ChatGPT and Chatsonic’s attempt at a sci-fi story
services already exist. chat sonic Boldly promote it as a means of creating blocks of non-plagiarized writing that students can use. I’ve previously had thoroughly fruitless discussions with the AI itself about how this is clearly cheating. The AI is very resentful and defends itself with roundabout arguments and the determination that simply asking a bot a word on a topic is a creative act. itself.
In fact, while writing the previous paragraph, I asked ChatSonic to write a science fiction, Hugo Award-winning 1,000-word short story about AI. For some reason he only got to 293 words (bloody freelancer), which sucks, but it took seconds.
ChatGPT, on the other hand, put in a much better effort, increasing their word count and writing some creativity behind it. Ultimately, it’s still a terrifying, hilariously self-aggrandizing, but mind-bogglingly capable:
(Well, if you’re dying to know how the second half ends, I’ll try to paste it in the comments.)
Can AI Surpass Human Creativity?
Clark said there are a number of ethical concerns that need to be resolved before considering publishing the AI-generated text. But how could such a thing happen? If AI can generate original stories worth reading, does it make sense to publish such stories? You need a tool, not a remix, but a true imagination, at which point you can Comparable to the best authors, but not necessarily better. “
Of course, “better” may not be the final deciding factor. Clarke adds: Machines can outrun human artists and drown them in noise. “
And in case all this hasn’t worried enough, let me conclude the short story I asked earlier with a chilling final paragraph on ChatGPT.
Of course, some were still skeptical. They believed that AI could never be truly creative and was just regurgitating information programmed into it. But SciFiGenius fans knew better. They knew AI wasn’t just spewing pre-made stories. They knew they were true artists capable of creating works that touched the hearts and minds of millions.
By the way, you can Support Clarkesworld Magazine in many waysIt’s becoming more and more important. Amazon will end its Kindle subscription service later this year..