How to Use AI for Lesson Planning: A Teacher Review Checklist

AI can help you get past the blank page. Give it a topic, year group, time limit, and learning objective, and it can draft a sequence of activities. That draft is a starting point. It does not know your students, your curriculum, what happened in yesterday's class, or which examples will land badly.

The safest useful workflow is simple: ask AI for a rough plan, then review every part as you would material from an unfamiliar colleague. Check the facts. Check the standard. Make sure the assessment tests the objective you actually set. Remove anything that depends on student information you should not have shared.

This guide gives you a practical workflow and a review checklist. It also shows where a short quiz can fit without turning the whole lesson into a test.

What should AI do in a lesson-planning workflow?

AI is best used for drafting options, not making final instructional decisions. It can suggest an opening question, put activities into a plausible order, or turn a topic into possible checks for understanding. You decide whether those suggestions suit the class.

That distinction matters. UNESCO's guidance for generative AI in education calls for human-centred use, data protection, and ethical and pedagogical validation. The U.S. Department of Education's report on AI and the future of teaching makes a related point: once software starts automating decisions about instruction, bias and unfairness become governance problems, not minor editing errors.

Use AI to widen your options. Keep professional judgment with the teacher.

What information should you give an AI lesson planner?

A vague prompt produces a generic plan. Before asking for a draft, write down the constraints that a substitute teacher would need:

• the learning objective in one observable sentence;
• the age or year group;
• what students should already know;
• the lesson length and available materials;
• one likely misconception;
• any accessibility requirements that can be described without identifying a student;
• the curriculum or standard code you intend to address;
• how you will know whether students met the objective.

Do not paste names, individual education records, disability information, behaviour notes, or other identifiable student data into a public AI tool. Follow your school's approved-tool policy and the law that applies where you teach. In the United States, the Department of Education's student-data security guidance is a useful starting point, but local and state requirements may add more.

Here is a prompt structure you can adapt:

Draft a 50-minute lesson for Year 8 science. Objective: students will explain how the structure of a plant cell relates to the function of the nucleus, cell membrane, and chloroplasts. Students already know the difference between plant and animal cells. Include one diagnostic opening question, one teacher-modelled example, guided practice, a five-question low-stakes check, and an exit ticket. Flag any factual or curriculum details that require teacher verification. Do not invent student information.

The last two sentences are worth keeping. They do not guarantee a flawless output, but they make the expected boundaries clear.

How should you review an AI-generated lesson plan?

Review the draft in this order:

1. Objective: Can you observe or assess the intended learning by the end of the lesson? “Understand cells” is too vague. “Explain how three cell structures support their functions” gives you something to check.
2. Accuracy: Verify every definition, worked example, date, quotation, formula, and answer. Use your curriculum materials and primary sources, not a second AI response as the fact-check.
3. Alignment: Do the activities practise the objective, or are they merely related to the topic? Check any claimed standard against the official publisher and confirm the code and wording.
4. Sequence: Does the lesson surface prior knowledge before introducing new material? Is there enough modelling before independent work? Cut activities that exist only to make the plan look busy.
5. Assessment: Does each question reveal something useful about the objective? Remove trivia, ambiguous distractors, and questions whose answer is obvious from their wording.
6. Accessibility: Check reading load, instructions, colour dependence, captions, assistive-technology compatibility, and alternatives for tasks that create barriers. “Differentiate for learning styles” is not a substitute for this work.
7. Privacy and safety: Remove identifiable student data. For practical work, verify equipment, supervision, allergies, safeguarding, and subject-specific safety requirements yourself.

NIST's Generative AI Profile is written for organisations rather than classroom teachers, but its risk categories translate well: validity, reliability, privacy, harmful bias, transparency, and downstream errors all deserve a check before generated material reaches students.

Where should a quiz fit in the lesson plan?

A quiz is most useful when it has a job. It might diagnose prior knowledge at the start, prompt retrieval after instruction, or show which idea needs reteaching before students leave. “Add a fun quiz” is not a job.

An applied review by Moreira and colleagues examined 23 articles conducted in real classrooms. Retrieval practice was generally more helpful than rereading or doing nothing, but the result depended on factors such as question format, age, feedback, and the comparison activity (Moreira et al., 2019). That supports a measured conclusion: low-stakes questions can help students retrieve taught material, but quiz quality and classroom context still matter.

Feedback needs the same care. Wisniewski and colleagues analysed 435 studies with more than 61,000 learners. Feedback had a positive average effect, but results varied substantially. Information about the task, the process, or what to do next was more useful than praise or a score alone (Wisniewski et al., 2020).

For the plant-cell example, a useful check might include:

• one multiple-choice question that distinguishes the cell membrane from the cell wall;
• one matching question connecting each named structure to its function;
• one short-answer question asking a student to explain why chloroplasts are found in many plant cells but not animal cells;
• feedback that explains the misconception behind each wrong answer.

If the objective is explanation, a quiz made entirely of label-recognition questions is misaligned, even if every question is factually correct.

How can Quizgecko support the assessment step?

Quizgecko does not replace the review above. It can shorten the mechanical part of turning your own lesson material into a question draft.

Upload the handout, paste your notes, or add source material to the AI quiz generator. It can draft multiple-choice, select-all, true/false, short-answer, fill-in-the-blank, and matching questions. You can edit the wording and answers before sharing the result.

If the lesson is already in a document, the PDF-to-quiz tool keeps the questions tied to that source. You can also turn the same material into AI-generated study notes or an AI flashcard set for later review.

Before using any generated question with students, check five things:

1. The keyed answer is correct.
2. The question can be answered from the material taught.
3. The reading level does not accidentally test something else.
4. Wrong answers are plausible but unambiguously wrong.
5. The feedback tells the student what to reconsider.

That final review is the teacher's work. It is also the part that makes the quiz belong in the lesson rather than sit beside it.

What does the evidence not show?

The evidence for retrieval practice and informative feedback is not evidence that an AI-generated lesson plan improves learning. Those are separate claims.

Research can inform the components you choose, such as low-stakes retrieval and feedback that points to a next step. It does not certify a generated plan, guarantee that it matches a particular curriculum, or show that a student will learn more because AI drafted it. Direct evidence about generative AI lesson planning is still developing, and tools change faster than long-term classroom studies can evaluate them.

Treat the output as a draft. A good plan still depends on a clear objective, accurate content, purposeful practice, and a teacher who knows what this class needs next.

A final two-minute check

Before you teach from an AI-assisted plan, ask:

• Can I state what students should learn in one sentence?
• Does every activity support that objective?
• Have I verified the facts, examples, and answers?
• Will the assessment reveal a misconception I can act on?
• Is the material accessible to the students in this class?
• Have I protected student data and followed school policy?
• What will I change if the opening check shows the class is not ready?

If those answers are clear, the AI has done something useful: it helped you reach a plan worth teaching. If they are not, generate fewer ideas and spend the saved attention on the decisions only you can make.

Want to build the low-stakes check? Generate a quiz from your lesson material, then review each question before it reaches your class.

References

Moreira, B. F. T., Pinto, T. S. S., Starling, D. S. V., & Jaeger, A. (2019). Retrieval practice in classroom settings: A review of applied research. Frontiers in Education, 4, Article 5. https://doi.org/10.3389/feduc.2019.00005

U.S. Department of Education, Office of Educational Technology. (2023). Artificial intelligence and the future of teaching and learning: Insights and recommendations. https://eric.ed.gov/?id=ED631097

UNESCO. (2023). Guidance for generative AI in education and research. https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research

Wisniewski, B., Zierer, K., & Hattie, J. (2020). The power of feedback revisited: A meta-analysis of educational feedback research. Frontiers in Psychology, 10, Article 3087. https://doi.org/10.3389/fpsyg.2019.03087