ISLS 2026
Demo & Interactive

#1250: Evidence-Centered Design for AI-Supported Assessment of Scientific Argumentation

Wed Jun 17, 4:15 PM–5:45 PM · ALP 1700

This demo showcases an AI-supported, evidence-centered assessment design that scaffolds middle school students’ scientific argumentation in the context of ecosystems learning. Grounded in Evidence-Centered Design (ECD), the assessment links constructs of scientific reasoning and argumentation with AI-driven evidence detection and feedback generation. Students investigate ecosystem dynamics through a NetLogo simulation, construct Claim-Evidence-Reasoning (CER) arguments, and receive adaptive feedback from an LLM-based chatbot. Participants will experience the task, explore reasoning analytics, and see how ECD structures both assessment and AI design. The demo highlights how ECD can ensure a valid, interpretable, and instructionally useful AI system that advances students’ reasoning and supports classroom instruction.

Authors

Lei Liu, Field M. Watts, Yi Song, Teresa M. Ober, Xiaoming Zhai, Yun Wang, Xin Xia, Ninghao Liu