Iterate the prompt in plain language and validate the simulation on technical, physical, and pedagogical levels¶

Strategy · All strategies
Evidence · no claims cited

Description¶

After generating a simulation from the four-element prompt, the teacher runs it, notices incorrect behavior, describes the problem in plain language, and sends a correction prompt; the article notes "The loop itself is short: run, notice what behaves incorrectly, describe the problem in plain language, and send a correction prompt." The result is then checked on three levels: a technical check (tracking and smooth mirrored motion), a physical check (correct relations between quantities, such as color ordering in the visible spectrum), and a pedagogical check (the gesture is natural and does not compete with the physics for attention).

Design Implications¶

Context¶

Requirements¶

• Willingness to run several prompt-and-correction iterations; two of the prompt's sentences were added this way, such as the phase-continuity requirement added after the wave shook with every wavelength change

Constraints¶

• AR simulations are especially sensitive to edge cases that appear only when a real hand moves in front of a real camera
• In class it helps to keep the hand at a roughly constant distance from the camera, where tracking is most stable

Target Learners¶

• teachers designing physics demonstrations
• students generating their own simulations

Target Learning Goals¶

• producing physically correct, stable, and pedagogically usable AR simulations of physics phenomena

Related Strategies¶

• Use generative AI as a no-code development tool so instructors can design laboratory software around their own pedagogical objectives

Examples¶

-

Key Sources¶

• Ofek Levy, Joshua Glazer, Noah D. Finkelstein, and Yossi Ben-Zion. (2026). From Prompt to Embodied Simulation: Using Generative AI to Create AR Physics Learning Tools. https://www.google.com/search?q=%22From+Prompt+to+Embodied+Simulation%22