About
September 2026 Google Research paper on generating customized STEM simulations with Gemini. Its teacher-led pipeline turns learning objectives into scaffolded, multi-level interfaces, then checks rendered HTML and JavaScript for visual, solution, telemetry, and mechanical failures before refining the result.
Summary
Why This Paper Adds New Signal:
1. Pedagogy shapes the interface: Maps curriculum alignment, learner agency, scaffolding, and formative feedback to concrete UI behavior
2. Evaluates rendered programs: Critiques both source code and browser output instead of trusting a plausible screenshot or explanation
3. Measures iterative repair: Raises the share of simulations passing every critique from 3.5% initially to 69.3% after ten refinement rounds
4. Keeps educators in control: Teachers approve learning objectives and select among generated candidates before students use them
5. Includes practitioner evidence: A 12-teacher usability study and independent expert ratings across 40 STEM requests found an 86% acceptance rate
1. Pedagogy shapes the interface: Maps curriculum alignment, learner agency, scaffolding, and formative feedback to concrete UI behavior
2. Evaluates rendered programs: Critiques both source code and browser output instead of trusting a plausible screenshot or explanation
3. Measures iterative repair: Raises the share of simulations passing every critique from 3.5% initially to 69.3% after ten refinement rounds
4. Keeps educators in control: Teachers approve learning objectives and select among generated candidates before students use them
5. Includes practitioner evidence: A 12-teacher usability study and independent expert ratings across 40 STEM requests found an 86% acceptance rate
Tags
generative-uieducationinteractive-simulationsgenerate-then-refineteacher-evaluation