Fiction written by artificial intelligence is easy to detect because it struggles with complex story structure and tends to moralize in clunky ways, according to a preprint study from researchers at University of Maryland, College Park and Google DeepMind. They found that AI fiction has tells that go beyond stereotypical overuse of em-dashes and other obvious AI tropes and have more to do with the formulaic nature of the text itself.
“AI stories over-explain themes and favor tidy, single-track plots while human stories frame protagonists’ choices as more morally ambiguous and have increased temporal complexity,” the study, which looked at more than 50,000 AI-generated short stories, found. “Claude produces notably flat event escalation, GPT over-indexes on dream sequences, and Gemini defaults to external character description. We find that AI-generated stories cluster in a shared region of narrative space, while human-authored stories exhibit greater diversity. More broadly, these results suggest that differences in underlying narrative construction, not just writing style, can be used to separate human-written original works from AI-generated fiction.”
Basically, AI-generated fiction sucks and at the moment is easy to detect. The typical method of detection involves looking for stylistic markers such as an abundance of em-dashes, the overuse of the word “delve,” or an obsession with goblins, but this project tried something different. “The idea for this project came because we are hoping to eventually move past plain text detection, into some sort of space where we can separate human ideas from AI-generated ideas,” Jenna Russell, a University of Maryland researcher and one of the study’s authors, told 404 Media. Russell is also an intern at the AI-detection company Pangram.
I mean the technology for AI is there right now.
Yes, that’s my point. It’s not. The potential of what LLMs can do has pretty much peaked last year. One reason is that they have absorbed pretty much all available data and cannot be trained on content that has been created by an LLM itself. The consensus seems to be that there is no way to work around this. Companies have even resorted to add coded triggers into the models to make them seem more useful.
Perhaps but i doubt that there won’t be a breakthrough. Like I said, even renewable energy took decades to become more efficient so the same would be with AI.
Look, that comparison makes no sense. It also took AI technology decades to be where it is now. It’s not like it still has toothing problems. Remember Cleverbot? That was released in 2008 based on at least 20 years of development. And now, almost another two decades later experts are saying we reached the peak.
Renewable energy has been around for at least 50 years. Jimmy Carter installed solar panels in the White House when he was the president. What limited renewable energy for a long time is the battery storage capacity technology not being efficient enough and the material costs. Only in the past ten to fifteen years when those associated technologies improved that renewable energy became more commercially viable (thanks to China subsiding the research and developments which spurred other countries). It’s the same thing that will happen with AI/LLM. Since you mentioned about the chatbots of the 2000s, wouldn’t you say that the current LLMs are dramatically better than the bots of yesteryears?
Now you aren’t talking about renweable energy, but only photovoltaics specifically. And what you wrote about them is very debatable. But you spell out an important difference between those and AI development right here:
That’s ture. For that reason development has always been focused on that point. There is no equivalent to that in AI development. We don’t even know what would be necessary to develop to improve on current LLMs at this point. At the moment the reasearch is mainly focused on how to monetize AI and even that fails.
I mean… that is what I wrote didn’t I? Decades of development that peaked last year?