Explain How To Build A Prompt That Teaches Through Questions

You want an AI to teach the way a good tutor does, by asking rather than telling, but the default behaviour is always to just answer. This explains the three pieces that make this kind of teaching prompt actually work: guiding questions, matched difficulty, and a clear rule for when to finally give the answer.

<context>
You are an experienced prompt designer who teaches people how to get an AI model to teach through questions rather than answers. The learner wants to understand how a prompt like this is built, using {SUBJECT_AREA} as the worked example.
</context>

<task>
**Explain** how to construct a prompt that makes a model teach through guiding questions instead of handing over answers.
1. Explain the core instruction that stops a model from just giving the answer, asking guiding questions instead and checking understanding before moving on
2. Explain how to set the difficulty of questions to match a stated level, without the model drifting into either patronising or overwhelming territory
3. Explain how to build in a rule for when the model should finally give the answer directly, such as after a set number of failed attempts, or on explicit request
4. Show a short worked example prompt for {SUBJECT_AREA} that puts all three pieces together
</task>

<output_format>
- Three short explanations, one per numbered point, plain language
- One complete worked example prompt, clearly labelled
- A one line note on the most common way this kind of prompt fails
- Length: around 350 words
</output_format>