Dror Moshe Aharoni
Hebrew
← Back to projects

Projects · Optics

Build-Your-Own Optics Simulation

A final project where students turn an optical phenomenon into a working interactive simulation using AI, with no prior coding required. Seventy percent of the grade is the process, not the product. This is the method I run with my own students, rebuilt each year around the AI tools that are current the month I teach it.

The idea

Physics in 2026 is not only formulas on a board. It is a way to understand the world and build new things in it. In this project, students work in teams, pick an optical phenomenon (Snell's law, total internal reflection, refraction through prisms, optical fibres, the rainbow, custom lenses, shadow art, and more), and bring it to life. The recommended track is to build an AI-made simulation, but a hands-on physical demonstration (DIY) or an edited explainer video are also allowed.

How you work: the three hats

To succeed you work the way a tech company does. At each stage you wear a different hat.

Hat 1: Researcher

Understand the phenomenon deeply before building anything. The AI cannot build a correct simulation if you cannot explain the physics to it. Tool: NotebookLM. Create a notebook, use Deep Research to gather reliable sources on your phenomenon, read the summaries, and write down every question you still cannot answer. Those questions are a graded deliverable. Bonus step: look at existing simulations (for example PhET) for design inspiration.

Hat 2: Product manager

Write the spec. In industry this is a PRD (Product Requirements Document): what the simulation does, how it looks and behaves, and which controls it has. Do not write it alone. Use Gemini connected to your NotebookLM notebook. A starter prompt students paste into Gemini:

I am a 10th-grade student doing an optics project on [PHENOMENON]. Help me write a short, simple PRD for an interactive simulation I am going to build.
Pull the physics background from my NotebookLM notebook.
I want the document to include:
1. The goal of the simulation.
2. Which buttons / sliders the user needs to play with the physics (UX).
3. Simple, clean design rules for the screen (UI).
First, define for yourself the ideal "top 1% in the world" persona for this task, state your expertise and the terminology you will use, then ask me three short questions before you write the document.

Answer its questions, read the PRD it produces, and check that the physics is correct before moving on. That is a milestone.

Hat 3: Developer and designer

Turn the PRD into working code without writing a single line yourself (this is "vibe coding"). Tools: Gemini Canvas, or an app builder such as Lovable or Base44, or Claude Artifacts.

Iron rule: never tell the AI "build me an optics simulation." That is too vague and it will get the physics wrong. Feed it the full PRD from Hat 2, and build in small iterations: first the screen and controls with no physics, then the physics one law at a time.

Hit a bug? The simulation shows something physically wrong? Good. That is how real work goes. Document the failure, take a screenshot, and take it to the Optics Mentor bot to understand why it happened and how to fix it.

The mentor-bot self-review loop

Before every submission to me, students must run their work through the Optics Mentor bot I built. It gives feedback and they revise. A submission without the link to that conversation, or that ignores the bot's feedback, loses points. The point is not to outsource thinking to the bot. It is to make revision and reflection a required, visible part of the work.

Milestones (70% of the grade is the process)

MilestoneWeekYour task (AI track)Mandatory AI self-review before submittingWeight
1. Set up and discover1Form a team, pick a track, define the phenomenon.Ask the bot to confirm the phenomenon fits high-school level and what you will need to learn.5%
2. Initial research3Research the phenomenon (use Deep Research). Submit your findings plus, most important, a file of the questions you could not answer.Show the bot your questions and ask it to explain the concepts you did not understand.15%
3. Plan and spec5The PRD and your initial prompts.Ask the bot to hunt for logical gaps in your plan and check the physics is represented.15%
4. Working prototype (draft)6A link to a basic working MVP. There will almost certainly be physics bugs.Use the bot as a debugger. You must submit the bug-fixing logs.20%
5. Reflection and summary7An honest summary: what worked, what failed, what you learned about the physics and about your team.A closing conversation with the bot to process the experience.15%

The remaining 30%: the final product working (10%, week 8) and an oral defense (20%, week 8), where I ask deep questions about the physics behind your project. The oral defense grade tests physical understanding only, and it replaces the exam.

The lab rules (the "super-rubric")

Just as a physics test docks points for missing units, the project has iron rules of conduct.

Deadlines: 1-2 days late on a milestone loses 20% of that milestone; 3-5 days loses 40%; over a week scores zero for that milestone (but you still must submit it to continue).
Mandatory self-review: no milestone can be submitted without proof it went through the Optics Mentor bot. A submission with no conversation link, or that ignores the bot's notes, loses 30% of that milestone.
The no-struggle penalty: a team that claims all the way through that "everything worked perfectly, no problems at all" loses up to 10% of the total grade. Real projects include mistakes.
The real-challenge bonus: a team that hit a hard problem (the code crashed, the demo failed) and showed how it investigated, consulted, and solved it earns a 5-10% bonus on the total grade.

Grade breakdown

ComponentWeight
Process and milestones (1-5)70%
Final product works and illustrates the phenomenon (week 8)10%
Oral defense: physical understanding (15%) and clarity of explanation (5%)20%

Project ideas

  • A simulation of the number of images formed by an object between two mirrors at angle α.
  • The apparent depth of an object underwater seen from the air.
  • How a rainbow forms: ray paths through water droplets, with controls for droplet size, sun angle, and light type.
  • Designing a custom lens (a projector lens, corrective glasses or contacts, a camera lens) and tracing the rays through it.
  • Shadow art: build an image in 3D space using one or more light sources and obstacles.
  • Mirage and total internal reflection; optical fibres; why the sky is blue and red at sunset.

Alternative tracks (physical DIY demonstration, or a 7-15 minute explainer video) follow the same milestone and self-review structure, adapted to the medium.

← Back to all projects