Free classroom guide
The best classroom use of AI begins with a learning goal, not a tool. Below are four workflows you can try, copyable starting prompts, and a complete example you can adapt tomorrow.
First principles
Use it to generate options you can inspect: lesson structures, examples, practice questions, explanations, and feedback questions. Do not treat a fluent answer as a correct one. You know the curriculum, the children, and the purpose of the task; the tool does not.
Before using any student-facing tool, follow your school’s approval and privacy rules. Keep identifiable student information out of general-purpose prompts unless your school has specifically approved that use. Check facts, representations, reading level, and accessibility before sharing an output.
Copy and adapt
Replace the bracketed details. These prompts request drafts and questions, leaving the instructional decision with you.
Workflow 1
I teach [grade/subject]. My objective is [objective]. Suggest three different ways to teach it in [time] minutes using [available materials]. For each, show what students will do, how I will check understanding, and one likely misconception. Do not invent standards or cite sources you have not checked.
Your check: Choose the approach that fits your actual students; confirm any standard and test the examples yourself.
Workflow 2
Create four short practice items for [skill] at [grade level]: one straightforward, two with common misconceptions, and one transfer task. Include a separate answer key explaining the reasoning. Avoid personal student information.
Your check: Solve every item before students see it. Revise difficulty and wording for your class.
Workflow 3
For this learning goal [goal], suggest a vocabulary preview, a worked example, and three guiding questions. Keep the same core thinking task for all students. Explain what each support helps a learner do.
Your check: Check accessibility, reading load, and whether the support preserves the intended thinking.
Workflow 4
Here is a fictional sample response to [assignment]: [invented response]. Identify one strength, one misconception, and two questions that would help the student revise. Do not write the revised answer for the student.
Your check: Use a fictional or de-identified example. Read the student’s real work yourself before giving feedback.
A classroom example
Learning goal: Students support a claim with accurate evidence and explain the link between the two. This example can work in science, history, or reading with a teacher-selected source.
Students do not need their own AI accounts for this lesson. The assessment is their evidence-based explanation, not the polished text the model produced.
Quick decision
Skip it when the task is meant to reveal a student’s independent thinking, when a direct conversation would serve a child better, or when you cannot verify the material in time. A blank page, a productive struggle, and a teacher’s observation can all be more valuable than a generated answer.
UNESCO’s AI competency framework for teachers organizes the knowledge, skills, and values educators need. The U.S. Department of Education’s leadership toolkit offers a broader school planning framework.
For a closer look at planning, read how to use AI for differentiated instruction and how to turn an existing lesson into inquiry learning. Both include a classroom example and a copyable teacher prompt.
Read AI Is Here. How Should Teaching Change? for three teacher-directed prompts and a way to make student thinking visible. Explore the companion Teacher Directed AI Prompts Across Subjects for six templates and subject adaptations.
Help children build literacy and a reading life with AI for Reading. Teach students to question AI output with the AI literacy lesson, or return to the AI in Education guide. School leaders can use the school AI guidance starter to set clear boundaries for staff and students.
If your team wants to practice these decisions together, see keynotes and scheduled workshops.