Google Research Pilot Lets Teachers Generate Custom STEM Simulations

Google Research Pilot Lets Teachers Generate Custom STEM Simulations
Key Takeaways

  • Google Research opened sign-ups in September 2026 for a pilot letting teachers at Google Workspace for Education schools request custom AI-generated STEM simulations built around their own curriculum goals.
  • The system uses Gemini’s generative UI capabilities to produce game-like interactives with progressive levels, hints and worked solutions, without requiring teachers to have programming or design skills.
  • A public library of more than 30 teacher-reviewed interactives across six STEM subjects is already live, and automated checks screen each generated interactive for pedagogy, solvability, mechanics and visual quality before teacher review.

Google Research opened sign-ups in September 2026 for a school pilot that will let teachers describe a curriculum goal and receive a playable STEM simulation in return, no programming required. The program, called Learning Interactives, is open to schools on Google Workspace for Education and targets middle and high school science and math curricula.

How the Simulations Work

Teachers input a topic and learning objectives. The system, powered by Gemini’s generative UI capabilities, producesย a game-like activity with progressive difficulty levels, contextual hints, feedback and worked solutions. The interface itself is generated dynamically rather than pulled from a fixed template, a design the team calls generative UI (GenUI).

A physics teacher, for example, could describe pendulum motion and receive a simulation where students adjust pendulum length or gravity and observe the results in real time. The system is engineered to guide students through a problem’s mechanics rather than surface answers directly, a distinction that reflects the learning-design guardrails Google built into the pipeline.

Automated checks evaluate each generated interactive for pedagogical soundness, solvability, interface mechanics and visual quality before teacher review. A public library, launched alongside the pilot, already holds more than 30 interactives in English covering physics, chemistry, biology, mathematics, computer science and earth science.

Teacher Approval at Every Stage

The workflow keeps educators in the decision loop throughout. Teachers submit curriculum goals, review and revise the AI-proposed learning objectives, then approve them before the system generates the simulation. Interactives in the public library were AI-generated and reviewed by teachers, and Google describes the pilot as an early step toward more interactives for public use.

Google’s Broader Education Push

Google launched the LearnLM model family for educational use in May 2024, followed by a “Guided Learning” mode for students in Gemini and interactive image capabilities in Gemini, both released in 2025โ€“2026.

In July 2026, a Google DeepMind randomized trial in Sierra Leone reported that Gemini’s Guided Learning produced a gain of about 0.26 standard deviations in math scores among 1,763 students over eight weeks, which Google translates to roughly 1.2 to 1.7 years of typical progress, an estimate rather than a directly measured outcome. The study was led by research director Irina Jurenka, according to Google. Google’s August 2026 announcements also detailed new Classroom tools letting teachers tailor student AI experiences and track progress in real time. Eligible U.S. college students can claim a free year of Google AI Pro, which includes 4x higher Gemini usage limits, access to Gemini Spark and 5TB of storage, through December 31, 2026.

What Comes Next

The pilot is explicitly a research experiment, not a finished product. Its integration into Google Workspace for Education lowers the deployment barrier for institutions already on the platform, but the program’s longer-term viability depends on whether the generation pipeline can maintain pedagogical quality at scale without placing an unsustainable review burden on teachers.

Feedback from participating schools will feed directly into model refinement. Google has not published a timeline for moving Learning Interactives beyond the pilot stage.

Morgan Blake
Morgan Blake

Morgan is a technology analyst covering enterprise AI strategy, automation, and business transformation. Morgan tracks how organisations are deploying AI at scale.

๐Ÿ“ฐ Journalists welcome โ€” cite Auton AI News with attribution. Press & Media โ†’ | press@autonainews.com