Author Information
Ruben Dawagne, Michigan State University, United StatesAbstract
As generative AI becomes more present in language learning, educators often face a tension between overreliance and avoidance. If students can generate language instantly, what does it mean to actually understand it?
This session introduces an approach where AI use becomes a process of debugging rather than simple generation. Drawing on recent research in AI literacy, I present three adaptable approaches in which AI-generated texts become objects of analysis rather than final products. Knoth et al. (2024) show that AI literacy develops when learners iteratively refine prompts and reflect on changing outputs, while Preschern et al. (2025) highlight the importance of explicitly structuring elements such as audience, role, and context. This session extends these insights into language education by showing how existing classroom activities can be adapted rather than replaced. Across these approaches, students work with AI-generated language in ways that require them to identify breakdowns in meaning, revise inputs, and examine how linguistic and cultural choices shift across iterations. Rather than asking teachers to replace existing activities, this framework shows how familiar tasks (such as reading, writing, and discussion) can be adapted to include AI-generated output as part of the learning process. Rather than treating AI as something to trust or ban, this approach positions it as something to question. AI errors become opportunities for deeper interpretive work, helping learners notice nuance and engage with intercultural meaning. Participants will leave with adaptable tools for a range of learning contexts.
Comments
Powered by WP LinkPress