Accepted Paper

Between Creativity and Control: Views on AI and Language in Rie Qudan’s Tōkyō-to Dōjō-tō (2024)  
Michaela Oberwinkler (University of Düsseldorf)

Send message to Author

Paper short abstract

This paper analyzes Rie Qudan’s Tokyo-to Dōjō-tō as a critical examination of AI within the narrative. Despite Qudan’s own use of AI in her work, the paper demonstrates that she takes a fundamentally skeptical stance toward AI-generated writing.

Paper long abstract

Rie Qudan’s novel Tōkyō-to Dōjō-tō (東京都同情塔; translated by Jesse Kirkwood as Sympathy Tower Tokyo) was awarded the Akutagawa Prize in 2024. At the award ceremony, Qudan revealed that approximately five percent of the novel had been written with the assistance of artificial intelligence. This statement led to an intense public debate, as the use of AI in literary production is widely considered controversial. However, beyond this societal discussion, the novel itself offers a far more critical and nuanced exploration of the linguistic capabilities and limitations of artificial intelligence.

The narrative alternates between the perspectives of two protagonists: Sara Machina, a successful architect, and her boyfriend, Takuto Tōjō, a poor fashion sales assistant. Sara enters an architectural competition to design a luxurious prison in the form of an immense tower in central Tokyo, while Takuto supports her in developing an appropriate conceptual framework for the project. Throughout the novel, both characters use AI for different purposes and to varying degrees, resulting in divergent outcomes.

This paper examines the use of AI in Tōkyō-to Dōjō-tō from multiple perspectives through close textual analysis, supplemented by interviews in which Qudan further articulates her views on artificial intelligence. In addition, Qudan’s experimental collaboration with AI in Kage no ame (影の雨; Shadow Rain), a work in which she designed a prompt that enabled AI to generate 95 percent of a novel, published in March 2025 in the journal Kōkoku, is contrasted with Tōkyō-to Dōjō-tō to show her critical stance toward AI-generated writing. The analysis demonstrates that Qudan consistently takes a skeptical position toward AI, both within her fictional narratives and in her creative practice, questioning its linguistic competence and, in particular, criticizing its excessive and stylistically problematic reliance on loanwords.

Panel INDMODLIT001
Modern Literature individual proposals panel
  Session 6