#333: “Explain photosynthesis to a Swiftie:” How Students Remix and Subvert “Answer Engine” AI Chatbots into a Dialogic Tool

As GenAI tools enter classrooms, their interactional role in supporting students is not well understood. This study examines 53 chat transcripts from two sixth-grade photosynthesis classes to investigate how students engage with a general-purpose chatbot. Rather than treating the bot as a one-shot “answer engine”, students actively shaped the exchange, prompting changes in voice, audience, and genre to explore the same core ideas in multiple ways. We examine three focal cases and trace how exchanges either works, wobbles, or breaks, as students calibrate explanations and responses to refusals or rigidly literal replies. These interactions show how style and audience design can support revisiting core mechanisms, not just adding play. Our analysis highlights the work students do to sustain dialogue, and the implications for designing more inclusive and dialogic AI systems.