The rapid uptake of generative AI in schools collides with a changing ecology of reading characterized by multimodality, fragmented attention, and increased reliance on short-form content. Literature education therefore faces a dual challenge: sustaining close reading and interpretive depth while teaching students to use AI responsibly. This Policy (2) a taxonomy of seven didactic functions for AI in literature teaching (reading companion, Socratic questioner, writing studio, curator, differentiation, assessment support, creative partner), each aligned to core literary goals and paired with low-tech alternatives; (3) ready-to-teach mini-models across levels (primary to upper secondary), including the Revise–Locate–Justify (RLJ) routine that strengthens evidence placement and warrants while protecting student voice; and (4) assessment and ethics guardrails that standardize line/page citation and lightweight process artefacts (evidence ledger, warrants list, rejection log) to stabilize validity in AI-present classrooms. The framework is designed for classroom adaptation and iterative improvement. The framework is designed for classroom adaptation and iterative refinement rather than as a claim of causal effectiveness; its purpose is to specify implementable guidelines and monitoring points that reduce common risks (hallucinated quotations, authorship displacement, bias, privacy breaches, and over-reliance) while preserving the human-centred practices of literature education: interpretation, argument, and reflective authorship.
Mašát et al. (Wed,) studied this question.