Can AI Tell How Well Teachers Notice? Researchers Are Testing It.
AI Summary: Researchers funded by the National Science Foundation are testing whether commercially available generative AI tools (Claude, ChatGPT, and a third selected tool) can accurately assess teachers’ professional noticing of students’ mathematical thinking and deliver rapid feedback. The two‑year, $489,316 project will have 150 in‑service and preservice elementary math teachers complete three established noticing assessments, with AI scoring de‑identified responses using custom rubrics, decision trees, and background materials. AI-generated scores will be compared to expert raters to determine validity and reliability, with the aim of producing a practical tool for teachers. The study emphasizes using off‑the‑shelf AI in realistic educational contexts rather than laboratory prototypes.