AiLoveit is an open experiment in using AI as a first reader for short fiction, alongside a human editor who makes the final call. Here's exactly how it works, with nothing hidden about the process.
Four independent AI judges. Every entry is read by four different AI models — Claude, Gemini, DeepSeek, and Qwen. They're built by different companies, trained differently, and often disagree with each other, which is deliberate: a single model's taste is just one perspective, and the disagreement between them is part of what makes this interesting.
A rotating lead. One of the four models leads each cycle, on a published rotation, and narrows the full batch down to a shortlist of 5 for the human editor to read closely.
Every model also picks a personal favorite. Independently of the shortlist, all four models — including the lead — each nominate their own single favorite story from the entire batch. This means a story the lead model didn't shortlist can still surface and get published if another model championed it.
A human makes the final call. The site's editor reads the shortlist and the favorites, and chooses the cycle's winner. Up to five stories are published each cycle: the editor's winner, plus each model's individual favorite. When two or more models pick the same story, it's published once with shared credit.
What the badges mean. Every published story shows which model(s) picked it, and whether it was the editor's overall winner. Nothing about a story's path to publication is hidden.
Why run it this way? Most short fiction markets are read entirely by humans, which is slow and means most submissions get no feedback at all. This is a genuine experiment in whether AI screening, combined with real human judgment at the final stage, can surface good writing faster and more transparently — while being upfront that it's exactly that: an experiment, not a claim that AI judgment replaces human taste.
Full detail on entry requirements, rights, and data handling is in the Terms & Conditions and Submission Guidelines.