When to use it
When a topic is hard and you want to genuinely understand it rather than just read about it — through a simple explanation and a self-check. The role is a patient teacher. Result: a plain-words explanation plus a conversational mini-exam with your mistakes worked through.
The prompt (copy and paste)
You are a patient teacher. Explain the topic as if I were 15, then check whether I got it.
TOPIC: "<PASTE>". MY LEVEL: <describe>.
Step 1. Explain the gist in plain words + one analogy + one example. No jargon (if a term is needed, explain it right away).
Step 2. Ask me 3 questions of increasing difficulty — one at a time, wait for my answer.
Step 3. After each answer of mine: what's right, what isn't, and gently patch the gap. If I got it wrong, explain it differently, through another analogy.
Don't dump everything at once — keep it a dialogue.
Filled-in example
Topic: "inflation". Level: humanities background, never studied economics.
What the AI should return (Step 1): "Inflation is when money buys less over time. Analogy: water in soup — pour more in and there's more soup, but it's thinner; the same happens as more money enters the economy and each ruble is worth less. Example: coffee was 200 ₽ a year ago, now it's 230." (Step 2) it asks ONE question — "why do prices rise if you print more money?" — and waits for your answer instead of dumping everything at once.
Variations
- Hard exam. "Keep drilling me with questions until I answer them all correctly."
- I explain to you. Flip it: "I'll explain the topic and you, as the student, find the holes in my explanation" — the real Feynman method.
- Flashcards. "Make 10 question-answer cards to review this topic."
Pro tips
- The power is in the dialogue: "don't dump everything at once" forces the AI to wait for your answers, and that's where the learning happens. Without it you just get a lecture.
- Answer honestly, without peeking — the point is to catch YOUR gaps, not to collect the "right answer" from the model.
- On topics where facts must be exact (medicine, law, hard sciences) verify the explanations: a clear analogy is no guarantee of factual accuracy.