ChatGPT Study Reveals User Generated Thousands of Doki Doki Literature Club! Fanfics
The study quantifies how a small number of users drive the majority of AI-generated fiction, including niche fan content, and reveals the scale of sexually explicit material produced through chatbots.
Reporting from 1 sources: Automaton.
A University of Washington and University of Colorado Boulder study analyzing over 500,000 anonymous ChatGPT conversations found that a single heavy user generated thousands of fanfics for Doki Doki Literature Club! over several months. The study, published as "AI Fiction in the Wild," classified about 195,000 conversations as fiction, with nearly 30% containing sexually explicit material. The top 2% of users accounted for over 80% of fiction-related conversations.
The research team used the WildChat dataset, collected from April 2023 to May 2024 via a free chatbot on Hugging Face, to analyze about 573,000 English conversations. They classified fiction using GPT-4o-mini and manually verified a sample. Of the roughly 195,000 fiction conversations, about 52,000 were sexually explicit. The most extreme user, described as an "infinite story demander," repeatedly asked ChatGPT to continue a scene where Doki Doki Literature Club! character Natsuki suddenly gives birth, generating thousands of variations over months.
The study also identified "story cyclers," users who repeatedly request the same story before moving on, and noted that fiction-related conversations were concentrated among roughly 10,000 users. The findings highlight how generative AI tools are used for sustained, repetitive fan fiction creation, often with explicit content, by a small subset of users.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.