AI Chatbot Scams Are Now Mixing With Sci-Fi World

An electric brain coming out of the nodes.

picture: Yuichiro Chino (Getty Pictures)

award-winning Clarkesworld Magazine He has helped launch the careers of science fiction writers for nearly 20 years by regularly featuring the work of Hugo Award nominees and winners such as Elizabeth Bear, Peter Watts and Catherynne M. Valente. But right now, in a rather ironic situation, it finds itself battling most modern trends against science fiction: AI.

by the way to a recent article More than a third of the submissions to the magazine this year were written by artificial intelligence and then sent by deception, according to Clarkesworld’s editor Neil Clarke. And it’s getting worse fast. In the first half of February, the number of AI-written entries more than doubled compared to the whole of January, Clarke said. kotaku There were only 50 people today.

Since the article was written, Clarke tweeted that as of now, applications are completely closed. “It shouldn’t be hard to guess why,” he adds.

Clarke said the decision to close the applications was made “at that moment”. kotaku via e-mail, as the numbers flowed this morning. “Either I can mole all day, or I can turn off submissions and work with legitimate submissions.”

The rate of increase in this situation is quite remarkable. Clarke notes in her blog post that she had to deal with plagiarism for a long time, but it wasn’t until the end of 2022 that the problem became so common. And then in the first month and a half of 2023, it escalated to such a scale that the journal entries suspended altogether.

A graph showing the increase in blocked entries to Clarkesworld Magazine.

Clarke’s graph showing the massive increase in bans.
graphic: Neil Clarke

How can Clarkesworld know that a story was produced by artificial intelligence?

Clarke doesn’t explain on her blog how she figured out which entries were written by the AI ​​because she doesn’t want to equip the cheaters with information that could help her circumvent her detection. However, he explained kotaku It’s not hard to spot right now.

“The ‘authors’ we’ve banned,” Clarke said, “are very clearly sending machine-generated text. That stuff is stereotyped and of poor quality.” However, he suspects that there is already a layer on top of them, not so obvious but suspicious enough. “None of them are good enough to warrant spending more time on them,” he explains. “

It’s not a problem Clarke faced alone. The editor reports that others in similar positions are facing the same challenges, and if this is happening to Clarkesworld, it’s clear that it will be wherever applications are open for publication. And for the most part, such submissions are weeded out for not being good enough to be published, while examining fakes is an expensive and time-consuming process.

Clarke adds that third-party detection tools that need to be able to recognize plagiarism or AI-written content are not the solution, given the number of false positives and negatives, and indeed the cost of such services. Other short-term measures, such as regional bans in parts of the world where most of the fake entries come from, aren’t the answer either. As Clarke points out in her article,

It’s clear that business as usual won’t be sustainable, and I worry that this path will lead to an increasing number of barriers for new and international writers. short fiction needs These people.

And of course, this is not an easy matter. The pace at which AI chatbots are evolving is enough for you to write ideas for a sci-fi short story, and it’s likely that soon-to-be tweaks will make it harder than ever to spot them right away. However, it looks like we still have a reasonable way to create stories that are truly worth reading. I asked Clarke if she thought this might be the case. “At this time, substantial improvement is required,” he said, not wanting to speculate on exactly how long such a leap will take from now on.

But this does not provide much comfort. “We still have ethical concerns about the means by which these artifacts were created,” Clarke said. kotaku“and until such concerns are addressed, we won’t even consider publishing machine-made studies.”

ChatGPT and Chatsonic’s attempts at a sci-fi story

services such as ChatSonic Boldly promoting themselves as a way to create plagiarism-free writing blocks that students can use. I had previously had grueling and futile arguments with the AI ​​itself about how this was clearly cheating, becoming extremely angry, defending itself with circular arguments and a determination that asking the robot for words on a topic was a creative act. itself.

In fact, as I was writing the previous paragraph, I asked ChatSonic to write me a 1,000-word short story about a Hugo Award-winning artificial intelligence writing science fiction. For some reason it only got to 293 words (damn freelancers) and it sucks, but it took a few seconds:

A ChatSonic short story.

Screenshot: ChatSonic / Kotaku

Meanwhile, ChatGPT put in a lot more effort, hit the word count, and wrote something with some creativity behind it. Ultimately, it’s still a terrifying story and ridiculously self-aggrandizing, but frustratingly competent:

The sci-fi story of ChatGPT.

Screenshot: ChatGPT / Kotaku

(Er, I guess I’ll paste the second half in the comments if you’re wondering how it will end.)

Can AI surpass human creativity?

Clarke mentioned above that she has many ethical concerns to resolve before considering publishing AI-crafted articles. But could such a thing ever happen? If AI could generate original stories worth reading, would it ever make sense to publish such things? “First,” said Clarke, “you need these tools to be able to write something that goes beyond the dataset. Real imagination, not remix. At that point, it can It may rival our best writers, but it cannot be guaranteed to be better.”

Of course, “better” may not be the ultimate defining factor. As Clarke adds, “the biggest difference and what’s causing us problems right now is speed. A machine can outlast a human artist and bury him in the noise of it all.

In case all this didn’t worry you enough, let’s end with the chilling last paragraph of ChatGPT’s short story I asked earlier:

Some people were still skeptical, of course. They believed that an artificial intelligence could never be truly creative, merely spewing out the information programmed into it. But SciFiGenius fans knew better. They knew that artificial intelligence could do much more than tell pre-written stories. They knew he was a true artist who could create works that touched the hearts and minds of millions of people.

Meanwhile, you can Support Clarkesworld Magazine in many different ways. This is something that is about to become even more important, when Amazon is discontinuing Kindle subscription services later this year.



Leave a Reply

Your email address will not be published. Required fields are marked *