Teacher-reviewed simulations are promising prototypes
Could teachers commission useful learning simulations without writing software? Google Research describes a workflow that turns a teacher's request into proposed learning goals, then generates simulations after the teacher approves those goals. Twelve U.S. STEM teachers submitted 36 requests. Internal raters selected the best of five candidates for each request; three requests produced no candidate meeting the quality threshold. Teachers responded positively to the usable simulations, while a separate expert exercise examined 40 requests. The work demonstrates a promising creation-and-review process for instructional prototypes. It does not yet establish whether students learn more, retain ideas longer or transfer understanding beyond the simulation.
Why it matters for leaders
Ask a curriculum team to test a small set for scientific accuracy, accessibility and standards alignment. Build a lesson in which students predict a result, manipulate variables and explain what happened. Follow with an unaided task that tests transfer, and record the teacher effort required to correct or adapt the simulation.
The evidence
The vendor report examines teacher usability and expert pedagogical ratings. The public release focuses on middle- and high-school STEM experiences.
Who was studied
Teacher-reviewed STEM simulations.
What the evidence cannot tell us
Small teacher sample, vendor authorship and candidate selection limit generalization. Classroom effectiveness remains future work; favorable ratings are not student outcomes.
Read the original sources
Summary and leadership analysis from the Week ending 2026-09-18 brief. The edition’s selection notes explain source access and evidence limitations.