AI That Works: Teaching AI Literacy Through Scientific Argumentation

What if AI could help middle schoolers write stronger scientific arguments while teaching them to question AI itself? This session shares a classroom-tested model that embeds AI literacy directly into scientific argumentation using the Claim Evidence Reasoning (CER) framework. Through a four-phase writing process supported by aiEDU, Common Sense Media and Colleague AI, students receive responsive questioning and rubric-aligned guidance that scaffolds their thinking without giving away answers. A multi-submission revision cycle turns science writing from a high-stakes summative task into an authentic formative process, as students improve their work and learn to assess AI output for accuracy, bias and limitations. Developed as part of the Presidential AI Challenge, this approach meets two critical needs at once. You will leave with strategies for implementing AI-supported revision cycles, designing prompts aligned to learning targets and using dashboards to spot patterns and misconceptions, plus CER resources, rubrics, annotated student examples, AI literacy links and an adaptable framework for any science unit.