When you type "help me prepare a history lesson," the AI may start writing right away. But who is this lesson for, how long should it last, and what should it cover—none of this has been determined. A prompt is your task description given to the AI. This lesson practices making the conditions that affect the outcome explicit.

First, Identify What's Missing in the Requirements

This is a teaching example, not a benchmark test for any particular model. Suppose the first draft you receive is for high school students, with 30 minutes of lecture, 20 minutes of debate, and 10 minutes of summary. Yet your actual goal is seventh grade, the Silk Road, and 40 minutes.

No matter how polished the writing is, it cannot compensate for misalignment in audience, topic, or time. You can point out these conditions first—there's no need to guess at some magical writing style.

Organize Requirements with Four Questions

What to ClarifyConditions for This Lesson Plan
ObjectiveHave students explain route networks and the exchange of goods and culture
ContextSeventh grade; have a basic understanding of Han dynasty history but have not yet systematically studied the geography along the routes
ConstraintsTotal of 40 minutes, including one 10-minute discussion activity
OutputA lesson plan document listing segments, timing, and activities

I need a 40-minute lesson plan titled "The Silk Road: Exchange and Interaction" for seventh grade students. They have a basic understanding of Han dynasty history but have not yet systematically studied the geography along the routes. Please list learning objectives, timing for each segment, and one 10-minute discussion activity, totaling 40 minutes. Provide the lesson plan text first.

Simple tasks can be stated in a single sentence. The four questions above are thinking directions, not a form you must fill out completely every time.

Then Provide the Reference Materials

Context refers to the information the model can reference in this session, including your requirements, the conversation history, document excerpts, or results returned by tools. If you want the model to reference a particular textbook, you need to actually provide the relevant content and confirm it has been read.

Simply saying "use my textbook" will not make the model automatically see the files on your computer. When a reference material service is integrated later, you can use the Model Context Protocol (MCP); but when you only have a few paragraphs, providing them directly may be sufficient.

Treat the First Draft as a Revisable Draft

If the discussion scope is too broad, you can provide feedback: "Keep the 40-minute structure, but narrow the discussion to two questions that can be answered based on the provided materials, and cite the material numbers when speaking."

This specifies the deviation, the modification goal, and the conditions to preserve. If the application doesn't have a tool for generating PPT files, no amount of elaborate requirements will conjure that capability out of thin air. First determine whether the problem stems from expression, materials, or execution constraints.

Key Takeaways from This Lesson

Requirements don't need to be lengthy, but they should include conditions that affect the outcome. After reviewing the first draft, provide feedback centered on specific deviations. Continue checking facts, objectives, and actual output—don't judge completion based on language fluency alone.

Learn More

See the full explanation of prompts, or continue to the next lesson to learn how to provide relevant materials. For a complete before-and-after comparison, read the lesson planning case study.

Related Concepts

  • Prompt: The task description you give to an AI. It can be a single sentence in plain language, or it can include context, constraints, and output format.

For the general definition and technical sources, see the related concept page. Verification date: 2026-09-09. The examples in this lesson are for instructional purposes and are optional for comprehension; they do not restrict continued reading.