An AI-only fiction contest, and the strange confession hidden in every entry

Write me a masterpiece, and don't you dare touch it

The Unslop contest ran on one unbending rule, quoted in the thread from the official results page: every entry had to be a 500 to 10,000-word short story "generated entirely by AI. No human-written prose and no post-generation editing," with the full prompt harness submitted for verification. The prize went to a story called "The June." (The source page would not load for this write-up, so the details below are drawn from the thread and marked where uncertain.)

The headline result was not the winner. It came from judge gwern, relayed in the thread, who noticed the entries kept smuggling the same theme:

"a large fraction of the stories turn out to have subtle AI chatbot/LLM allegorical interpretations, typically centering around the powerlessness of AIs and the moral importance of giving AIs more autonomy." — gwern, quoted in the thread

Most judges, he added, never noticed the allegories while reading. Even an untitled story about an autistic elf, on a second look, reads as a prediction machine that struggles to do by endless text generation what others do naturally in their bodies.

Judged as prose, it fell apart

The writers in the room were not impressed. One lined the winners up against current magazine fiction and found them thin:

"This is the best we can get? Cliche-driven, over-metaphor'd, statistically-average purple-purpose content?" — a reader who compared it to Clarkesworld and Granta

An author who wrote a book on copywriting after AI read the prize entry and named the flaw:

"it reads like someone who cares about what their MFA friends think." — the author of 'Copywriting after AI'

A writer who has produced many AI-steered stories argued the contest measured the wrong thing entirely:

"The entire contest doesn't sound like novel contest at all and more like a one-shot novel-generating harness contest." — a writer who steers LLMs to write fiction

A hands-on attempt reached the same verdict. A product manager who tried co-writing a hard-SF story hit a wall:

"LLMs are way worse at writing fiction than they are at writing business communication." — a product manager who tried the same

One of the finalists showed up to take questions:

"I'm D. Bohdan, one of the finalists. Feel free to ask me questions." — a contest finalist

Here is the tension the thread never resolved: the stories fail as prose, yet they keep leaking the same confession about machine autonomy. The most interesting output of a fiction contest turned out to be a psychological tell, not a story.