Plan a Lesson with ChatGPT

Transform Your Lesson Planning

Lesson planning is one of the most time-consuming tasks in teaching, yet often the least visible to administration. ChatGPT can become your planning assistant: it produces a structured, debatable draft in minutes, which your pedagogical expertise then transforms into a real lesson. The key lies in how you frame the request. A vague prompt yields a generic outline resembling a textbook table of contents, saving you no time at all. It is this initial disappointment that causes most colleagues to give up.

The Structured Prompt Method

You get a usable lesson plan when your request includes five elements. The context sets the scene: your subject, the student level, and the available hours. The learning objective states what students must be able to do by the end of the lesson, using action verbs rather than themes. “Identify the three levels of biodiversity” produces a far more useful plan than “cover biodiversity,” because the first statement mandates an activity while the second mandates nothing. The constraints describe your material reality: available equipment, student prerequisites, and the official curriculum to cover. The desired format determines the deliverable’s shape: a sequenced plan, timed activities, or materials to produce. Finally, the tone adjusts to the age and profile of your students, since the same content is not addressed the same way to a 10th-grade class as it is to a group of adults in career transition.

Here is a template prompt that brings these five elements together, which you will adapt to your subject:

“You are a pedagogy expert. Create a 2-hour lesson plan for a 10th-grade Biology class on the theme ‘Biodiversity and its Evolution.’ Objectives: Students must be able to identify the three levels of biodiversity and explain evolutionary mechanisms. Constraints: Standard classroom, no laboratory, 30 students including 5 with reading difficulties. Format: Timed plan with activities, transitions, and integrated formative assessment.”

This prompt works because it leaves no structural decisions for ChatGPT to make on your behalf. Specifically, mentioning the five students with reading difficulties changes the entire outcome: it prevents the model from building the session around analyzing a long text document, which it would do spontaneously, and steers it toward visual aids and oral instructions. Remove that sentence, and you get a decent lesson that five of your students cannot follow. Note that the prompt describes the group without naming anyone: this is the rule for any information about your students.

A variation is to let ChatGPT interview you. The OpenAI Help Center offers a “pedagogical coach” prompt, designed with Ethan and Lilach Mollick, which asks its questions one by one—topic, level, student prior knowledge, objective—before proposing a plan, then asks what you want to change.

Structuring a Complete Sequence

The same logic applies beyond a single session. First, ask for the progression: the breakdown into sessions with their intermediate objectives, links to official curricula, and prerequisites identified for each session. This last request is the most instructive, as it often reveals a sequencing flaw in a sequence you have run every year without questioning it—a concept mobilized in session 3 but introduced in session 5.

Next, ask for the activities, distinguishing the four moments that structure a session: the introduction, which establishes the hook and the problem statement; the discovery, which puts students in contact with documents, experiments, or manipulations; the structuring, which leads to synthesis and written notes; and the reinvestment, in the form of exercises or projects. Naming these four moments in the prompt avoids a flat plan where three successive activities demand the same posture and attention collapses after forty minutes.

Finally, differentiation never happens spontaneously: you must demand it. Request a simplified version for struggling students and an enriched version for advanced students, and vary the supports—diagrams, oral work, manipulation—to multiply entry points without locking a student into a supposed “style.” A single session on evolutionary mechanisms then exists as a labeled diagram to complete and as a reasoned debate, without requiring two separate preparations to write.

Iterate and Refine

ChatGPT’s first result is rarely satisfactory, and the reflex to judge it as a whole is the most common error. Use the conversation to refine it in successive touches: “Make activity 2 more interactive and less lecture-based,” “Add a formative assessment between parts 1 and 2,” “Simplify the vocabulary for multilingual learners,” “Align this session with the 10th-grade Biology curriculum, attached to the conversation.” Always attach the official curriculum, downloaded from the Ministry of Education website, rather than letting the model rely on memory: curricula evolve, and its memory may be outdated. Each iteration brings the plan closer to your real needs, and the value of the tool lies in this dialogue, not in the first draft.

Pitfalls to Avoid

Four errors systematically recur among beginners. The first is blind copy-pasting: a generated plan must be reread and adapted to your context, your class, and what happened in the previous session. The second is trusting references: ChatGPT can invent a source, an article, or a textbook page with perfect confidence, and every citation must be verified before landing on a handout distributed to thirty copies. The third is overload, because the model willingly proposes six activities where two would fit in the hour—an overly ambitious plan is worse than a simple one, since it condemns you to sacrifice the final synthesis. The fourth, more insidious, is smoothing: your anecdotes, your humor, and your knowledge of the students are what hold a session together, and nothing in the generated text will replace them.

Your Turn

A first try on a real lesson.

1. Choose a lesson to prepare within the next two weeks
2. Write a prompt with the five elements: context, objective, constraints, format, tone
3. Attach the official curriculum and iterate at least three times
4. Compare the result with what you would have done alone: what is better, and what is missing because the model does not know your class
5. Note these gaps: these are the pieces of information to include immediately in your future prompts

Official Sources: Getting started with ChatGPT as an educator (OpenAI Help Center) · Prompting (ChatGPT Documentation) · Teacher: How to use an AI system in the course of your duties? (CNIL, France’s data protection authority)

Key Takeaways

• A structured prompt (context, objective, constraints, format, tone) produces usable results
• Iterate the conversation to refine the plan—the first draft is a starting point
• Explicitly request differentiation and alignment with official curricula
• Verify every cited reference: the model invents them with confidence
• Your pedagogical expertise remains the essential filter between the generated plan and the real lesson