Defining AI Literature: Data-Driven Genetic Mode and Collaborative Turn

As data-driven systems have entered literary production, the label AI Literature has come to encompass practices that differ in how computational systems participate in composition. This article addresses this definitional instability through two complementary concepts: the data-driven genetic mode and the collaborative turn. The first identifies a transgeneric category of works whose production involves systems that learn patterns from data, while requiring intentional human organization and poetic or interpretive relevance. The second describes an asymmetrical redistribution of literary production across human and computational operations, without implying co-authorship or equal intentionality. A historical reconstruction of documented works from 2016 to 2026 traces this field from statistical and custom neural systems to pretrained Transformers and conversational interfaces. Rather than showing a replacement of human authorship by autonomous systems, this trajectory reveals growing experimentation with relations among models, archives, interfaces, editorial procedures, and human competencies. The article describes AI Literature as a historically coherent but technically heterogeneous field of literary practice. These works do not delegate writing to automated systems: they engage AI critically, using literary practice as a way to deconstruct technological tools.