To create a lesson plan with AI, give the model a verified curriculum goal, a de-identified learner profile, lesson constraints, and approved source material. Ask it to align observable objectives, success evidence, timed activities, formative checks, differentiation, accessibility, and a backup path. A teacher must verify facts, approve support choices, teach the lesson, and revise the plan from classroom evidence.

This guide designs one lesson. It does not replace a long-term AI study plan or a training manual. The general guide to using AI explains safe context and review; this workflow turns that discipline into a teacher-owned lesson document.

Key Takeaways

Begin with learners, time, standards, and approved materials.

Write observable objectives before selecting activities.

Decide what evidence will show learning before drafting instruction.

Plan checks during the lesson, not only a final task.

Build differentiation and accessibility into the sequence.

Never provide identifiable student, IEP, medical, or disability records.

Require teacher approval and a post-lesson reflection.

How do you describe the class without exposing students?

Describe the class at group level: subject, approximate age or stage, prior instruction, language of teaching, duration, room and technology constraints, and known access needs expressed generically. Do not enter names, student identifiers, identifiable work, individual education plans, medical information, disability records, behavior records, or family circumstances.

Use a context card:

Field	Example of safe detail
Lesson	One 50-minute introduction to proportional reasoning
Learners	Mixed prior knowledge; group-level description only
Prior learning	Fractions and equivalent ratios were taught
Constraints	Shared materials; no individual devices required
Approved sources	Curriculum document and teacher-checked examples
Access	Multiple representation and response modes needed
Authority	Named teacher approves the plan

UNESCO's guidance on generative AI in education calls for a human-centered approach, data privacy protection, age-appropriate use, and human agency.［1］ A useful plan therefore starts with a privacy boundary and named professional responsibility.

Verify standards and source material

If the lesson must address a curriculum standard, copy the exact identifier and wording from the official source. Do not ask AI to recall the standard. Attach the source link or document location and record the version if it can change.

Prepare a small evidence pack: the verified standard, teacher-approved textbook pages or readings, key facts, vocabulary, worked examples, and prohibited misconceptions. Check licenses before reproducing material. When AI proposes a quotation, fact, historical claim, formula, or standard number not in the pack, verify it from the primary source before use.

UNESCO's AI Competency Framework for Teachers treats AI use as part of broader human-centered practice, ethics, foundations, pedagogy, and professional learning.［2］ The model is not the curriculum authority; it is a drafting aid inside a teacher's process.

What makes a learning objective observable?

An objective should describe what learners will do and under what conditions. “Understand photosynthesis” is hard to observe. “Given a labeled diagram, learners will explain how light, water, and carbon dioxide relate to glucose and oxygen, using the target vocabulary” identifies visible evidence.

Use one to three objectives for a single lesson. Separate knowledge, reasoning, and skill when they require different evidence. Avoid claiming mastery that cannot be demonstrated in the available time.

Ask AI to test each objective:

Is the action observable?
Does it match the verified standard?
Can the planned evidence demonstrate it?
Is it feasible in the lesson duration?
Does it assume prior knowledge not established?
Does it require a mode of expression unrelated to the objective?

The final wording belongs to the teacher.

Why define success evidence before activities?

Plan backward from evidence. For every objective, define what the teacher will see or hear and what would count as partial, sufficient, or incorrect evidence. This is not yet a grading rubric; it is a compact success criterion for instructional decisions.

Objective	Observable evidence	Success criterion	Likely misconception	Teacher response
Compare two strategies	Learner explains one similarity and one difference	Uses evidence from both examples	Names steps without comparing	Model a comparison stem and recheck

Choose low-stakes evidence: oral explanation, worked example, sorting task, annotated diagram, short response, demonstration, or exit prompt. Ensure the response mode does not create an irrelevant barrier.

Sequence the lesson with realistic timing

A reliable sequence usually includes an opening, teacher modeling, guided practice, independent or collaborative application, formative checks, and a close. The exact structure depends on the subject and learners.

Phase	Purpose	Teacher action	Learner action	Time	Evidence/check
Opening	Activate prior knowledge	Present a short prompt	Recall and connect	5 min	Responses reveal starting point
Model	Make thinking visible	Demonstrate and narrate	Observe and question	10 min	Targeted questions
Guided practice	Rehearse with support	Prompt and diagnose	Solve with peers	12 min	Sample work and explanations
Independent practice	Apply the objective	Monitor	Produce evidence	15 min	Objective-aligned task
Close	Consolidate and decide next step	Run exit check	Explain or submit	5 min	Exit evidence

Add transition time, distribution of materials, and a buffer. If the minutes do not fit, remove or narrow an activity rather than pretending every phase will happen.

How do formative checks guide teaching decisions?

Formative assessment helps the teacher decide what to do next during learning. Write checks at the moments where misunderstanding is likely. Avoid “Does everyone understand?”; use a prompt that produces observable evidence from more than the most confident volunteers.

For each check, define a response:

If most evidence meets the criterion, continue or deepen.
If a common misconception appears, pause and model another example.
If evidence is mixed, regroup or offer targeted practice.
If the check itself was confusing, revise the prompt before judging learning.

AI can suggest likely misconceptions, but the teacher must verify them against subject knowledge and actual learner evidence. It must not invent diagnostic conclusions about individuals.

How do you build differentiation and accessibility into the core?

CAST's UDL Guidelines 3.0 organize considerations around multiple means of engagement, representation, and action and expression.［3］ Use them as design prompts, not as a claim that one checklist serves every learner.

Plan access without lowering the objective unnecessarily:

provide vocabulary, examples, diagrams, audio, or manipulatives where relevant;
offer a clear model and chunk multi-step directions;
allow more than one valid way to show learning when mode is not the target;
include extension through depth or transfer, not simply extra volume;
anticipate sensory, language, motor, attention, and technology barriers;
prepare an offline or low-technology path.

AI must not decide a named student's formal accommodation or support. Those decisions belong to educators and established processes using information that should not be placed in the prompt.

Prepare materials and a contingency path

List every handout, slide, object, link, device, safety item, and setup action. Verify that links open, examples are correct, reading levels are suitable, and materials are licensed. If AI drafts practice questions, solve them independently and check that all answers and distractors are defensible.

Write a contingency for likely failures: less time, absent technology, fewer materials, a misconception taking longer, or learners already knowing the content. A backup should preserve the objective, not merely fill time.

The U.S. Department of Education's report on AI and the future of teaching and learning emphasizes keeping humans in the loop and using AI to support, rather than displace, educator judgment.［4］ A contingency plan is another place where teacher knowledge of the classroom matters more than generated fluency.

Create a lesson plan with AI using a bounded prompt

Draft one teacher-reviewed lesson plan from the supplied evidence only.
Return: learning context; verified standards; observable objectives;
success criteria; prior knowledge; likely misconceptions; timed sequence;
materials; formative checks and decision rules; differentiation;
accessibility; contingency; and teacher-approval field.
Mark unsupported facts or standards as VERIFY.
Do not create student profiles, accommodations, diagnoses, or final grades.
Do not claim the lesson is suitable until a teacher approves it.

Ask for a traceability pass: every activity should support an objective, every objective should have evidence, and every cited fact should point to an approved source.

What should the teacher review before teaching?

The teacher reviews content accuracy, curriculum alignment, cultural and linguistic assumptions, age appropriateness, cognitive load, timing, classroom safety, accessibility, and assessment validity. Read every instruction from the learner's perspective. Complete the tasks yourself.

Use a simple approval block:

Review	Result
Facts and sources checked	Name/date
Standards verified	Name/date
Timing rehearsed	Notes
Access and differentiation reviewed	Notes
Materials ready	Location
Teacher approval	Approved / revise

Do not teach directly from an unreviewed model response.

Reflect after the lesson

After teaching, record evidence rather than a vague impression. Which objectives were demonstrated? Where did timing drift? Which instruction or example caused confusion? Which access option helped? What should be retained, revised, or removed?

Keep the reflection at class level unless the authorized school process requires protected individual records. Update the lesson plan with a version note and evidence. Do not send identifiable student responses back to a general AI tool.

If learners use AI during the lesson, set a separate policy for attribution, verification, privacy, and acceptable assistance. The guide to using AI for homework responsibly can help frame that boundary. For retrieval practice, adapt the verified content through the flashcard and quiz workflow.

Copyable lesson plan template

# Lesson plan
- Learning context:
- Duration and constraints:
- Verified standard/source:
- Prior knowledge:
- Common misconceptions:

## Objective and evidence
- Observable objective:
- Success criteria:
- Evidence learners will produce:

## Sequence
| Phase | Teacher action | Learner action | Time | Formative check |

## Access and response
- Differentiation:
- Accessibility:
- Extension:
- Contingency:

## Review
- Facts and citations checked by:
- Teacher approval:
- Post-lesson evidence and revision:

Final quality checklist

Context is useful but contains no identifiable student information.
Standards, facts, quotations, and examples were checked against primary sources.
Objectives are observable, feasible, and linked to evidence.
Activities fit the available time and include transitions.
Formative checks have explicit teacher decision rules.
Differentiation and accessibility preserve the intended learning goal.
Materials and backup paths were verified.
A teacher approved the plan and owns all formal support decisions.
Post-lesson evidence will inform the next revision.

FAQ

Can AI write a complete lesson plan?

It can draft the structure and alternatives, but a teacher must verify content, standards, timing, accessibility, safety, and suitability.

Can I include student names or IEP details?

No. Do not provide names, identifiers, IEPs, medical or disability records, or identifiable work to a general AI tool.

How many objectives should one lesson have?

Usually a small number that can be observed in the available time. Feasibility and alignment matter more than a fixed count.

Should activities come before objectives?

No. Define objectives and success evidence first, then choose activities that produce that evidence.

Can AI choose formal accommodations?

No. Educators and established support processes make individual decisions; AI may suggest general access questions only.

How do I verify a standard number?

Open the official curriculum source and check the exact identifier and wording. Do not rely on model memory.

What if the planned technology fails?

Prepare an offline or low-technology contingency that still addresses the same objective and evidence.

How should I improve the plan after teaching?

Record class-level evidence about learning, timing, misconceptions, and access, then version the plan with specific revisions.

Sources

UNESCO, “Guidance for Generative AI in Education and Research” — https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research
UNESCO, “AI Competency Framework for Teachers” — https://www.unesco.org/en/articles/ai-competency-framework-teachers
CAST, “UDL Guidelines 3.0” — https://udlguidelines.cast.org/
U.S. Department of Education, “Artificial Intelligence and the Future of Teaching and Learning” — https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf

Sources checked 31 August 2026.

How to Create a Lesson Plan with AI