Send message to Author

Paper short abstract

Using Rie Qudan's AI-written short story "Kage no ame" and its published chat log as a case study, this paper analyses AI-assisted literary production through Azuma Hiroki's theory of database consumption, arguing that LLMs transform authorship from writing to database operation.

Paper long abstract

After admitting in January 2024 that a chatbot had written 5% of her Akutagawa Prize-winning novel "Sympathy Tower Tokyo" (2023), the Japanese advertising magazine Kohkoku invited author Rie Qudan to take part in an experiment: creating a story composed 95% by a large language model (LLM) and only 5% by herself. The result was the short story "Kage no ame" (2025), in which the narrator – suggestive of an AI network – reflects on the extinction of humanity. Alongside the story, Kohkoku also published the chat exchange between Qudan and the LLM, providing insight into the process behind the text.

Using "Kage no ame" as a case study, this paper conducts a close reading of both the short story and the accompanying chat exchange to analyse the creative process behind generating a literary work in collaboration with AI technologies. The analysis is informed by Azuma Hiroki’s theory of "database consumption", as articulated in "Dōbutsukasuru posutomodan" (2001; translated into English as "Otaku: Japan’s Database Animals" in 2009), describing a postmodern cultural logic in which narrative meaning emerges from recombining elements from a database rather than from a unified authorial intention. Drawing on this framework, the paper argues that using LLMs during the writing process shifts the author's role from writer to database operator, with creative contributions taking the form of prompt design, constraint-setting, and targeted intervention.

The use of AI for literary text production has provoked intense debate in recent years, given that LLMs draw on preexisting texts to generate their replies, raising concerns about issues such as copyright infringement and plagiarism. Yet, as the case of Rie Qudan demonstrates, AI-assisted literary writing is already in active use, highlighting the importance of a careful and nuanced analysis of the author's role within this process.