Ready to put this into action?
Get the complete AI Integration Playbook β Practical AI implementation guide β prompt engineering, workflow automation, and ROI frameworks.
Article 073 Β· Part 8
AI for Teachers, Trainers, and Students
Design the learning objective, the practice, and the evidence of understanding together.
By Randy Salars Β· Published
On this page
- Start with an observable outcome
- Draft a short aligned lesson
- Check every example and answer
- Preserve the objective when adapting access
- Set AI-use expectations before the activity
- Assess the skill directly
- Use feedback to plan the next lesson
- For students: make the tutor ask you to think
- Practice: build an aligned lesson package
Design the learning objective, the practice, and the evidence of understanding together.
An AI assistant creates a lively lesson about percentages. Students watch a video, discuss advertisements, and design colorful posters. The exit task asks whether they enjoyed the activity.
The lesson may be engaging, but it does not show whether students can calculate or explain a percentage change. The teacher still needs evidence that the intended learning occurred.
AI can help prepare examples, adapt explanations, and draft feedback. Its usefulness depends on alignment: what the learner should do, what the practice requires, and what the assessment actually measures.
Start with an observable outcome
Replace βStudents will understand percentagesβ with a task a student can demonstrate. For this lesson: Given an original value and a new value, calculate the percentage change and explain why the original value is the denominator.
A second objective might be to distinguish a percentage-point change from a relative percentage change. Keep the objectives few enough to teach and assess within the available time.
Identify prerequisites. Learners should be able to subtract, divide, and convert a decimal to a percentage, with any approved calculation support. If those skills are uncertain, add a short diagnostic and a scaffold rather than assuming the class is ready.
UNESCOβs teacher framework treats AI pedagogy, ethical use, human agency, and professional learning as parts of teacher competence. It offers a framework for development, not evidence that any particular generated lesson works. UNESCO: AI Competency Framework for Teachers.
Draft a short aligned lesson
Here is an original 25-minute lesson plan for older students or adult learners. The numbers and situations are invented for instruction.
| Time | Activity | Evidence sought |
|---|---|---|
| 3 minutes | Ask learners to compare an increase from 40 to 50 with one from 80 to 90 | Whether they notice equal absolute changes can differ proportionally |
| 5 minutes | Model percentage change using the original value | Correct denominator and explanation |
| 7 minutes | Guided practice with one increase and one decrease | Calculation and reasoning with support |
| 6 minutes | Compare a percentage-point statement with a relative-change statement | Ability to distinguish the two measures |
| 4 minutes | Independent exit task | Transfer to a new example |
The timing is a plan, not an observed classroom result. A teacher should adjust it to the learnersβ needs while preserving the objective.
Use AI to produce an aligned draft:
Design a 25-minute lesson for learners who can perform basic division. Outcome: calculate percentage change using the original value and explain the denominator. Include a worked example, a common-error example, guided practice, an accessible response option, and an independent exit task using new numbers. Show the answer key separately. Do not claim the lesson has been tested or assign learner proficiency from the prompt.
Ask the assistant to map each activity to the outcome. If an activity does not contribute, revise or remove it.
Check every example and answer
For a change from 40 to 50, the increase is 10. Relative to the original 40, that is 10 Γ· 40 = 0.25, or 25%. For a change from 80 to 90, the increase is also 10, but 10 Γ· 80 = 12.5%.
The denominator matters because percentage change describes the change relative to the starting amount. Dividing 10 by the new value of 50 gives 20%, which answers a different ratio question.
For guided practice, use a decrease from 60 to 45. The signed change is β15; dividing by 60 gives β25%, a 25% decrease. Have learners explain both the magnitude and direction instead of memorizing a sign.
Then compare rates. If a fictional completion rate rises from 40% to 50%, it rises by 10 percentage points. Relative to the original rate, the increase is 25%. These statements describe the same change using different units.
The teacher should verify the arithmetic and language before presenting the materials. AI-generated answer keys can be wrong even when the explanation sounds confident. Include an intentional incorrect answer only when it is clearly labeled as an error for students to analyze.
Preserve the objective when adapting access
Access can change without weakening the learning goal. Provide readable text, a spoken explanation, a table, or a worked diagram. Let a learner explain the denominator orally if writing fluency is not the skill being assessed. Use larger text or an accessible digital format when needed.
A scaffold might show: original value, new value, difference, difference divided by original, percentage. Remove parts of the scaffold gradually if independent setup is part of the objective.
Do not equate simpler language with simpler thinking. βWhat amount are we comparing the change with?β can support the same mathematical reasoning as more technical wording. Conversely, an adaptation that supplies the entire denominator and calculation may prevent the assessment from showing whether the learner can choose them.
Follow the learnerβs applicable accommodations and institutional requirements. AI should not diagnose a learner, decide an accommodation, or infer ability from writing style. It can draft alternative formats for a teacher to review.
Set AI-use expectations before the activity
Tell learners where assistance is allowed and what they must demonstrate independently. A clear example policy for this fictional lesson is:
During practice, you may use an approved AI tutor to request a hint or an alternative explanation. Save the prompt and explain which suggestion you used. During the exit task, do not use AI. A calculator and approved accessibility supports remain available because the task measures setup and interpretation rather than arithmetic speed.
This is an example classroom rule, not a universal policy. Real learners must follow their actual course requirements. Define the permitted resource conditions before comparing performance.
Teachers should also use approved tools and data handling. A generated explanation usually does not need a studentβs full name, grade history, disability record, or private circumstances. Fictional examples can often accomplish the instructional task without personal records.
Assess the skill directly
The independent exit task uses a new scenario: βA clubβs attendance rises from 50 to 65. Calculate the percentage increase and explain the denominator. Separately, a rate rises from 20% to 25%. State its percentage-point change and relative percentage increase.β
The answer key is: attendance rises by 15; 15 Γ· 50 = 30%. The rate rises by five percentage points; 5 Γ· 20 = 25% relative increase.
An original scoring guide could use four criteria:
| Criterion | Evidence of success |
|---|---|
| Identifies change | Correctly finds 15 and five percentage points |
| Chooses reference value | Uses 50 and 20, respectively |
| Calculates relative change | Obtains 30% and 25% |
| Explains meaning | Distinguishes relative percentage from percentage points |
Keep the criteria separate when giving feedback. A student who selects the correct denominator but makes an arithmetic slip needs different support from one who divides by the new value. A single total can hide that difference.
The exit task is one piece of evidence, not a complete judgment of a learner. A later independent problem can test retention and transfer. An enjoyable activity or a correct AI-assisted answer does not by itself establish independent understanding.
Use feedback to plan the next lesson
Ask AI to help draft feedback from verified work samples, using approved handling and minimal data. Supply the rubric and the actual error. A useful comment is: βYou correctly found the increase of 15. Use the original attendance of 50 as the reference, then explain what the resulting percentage describes.β
Avoid broad personal judgments such as βYou are weak at math.β The evidence concerns a particular response, not a fixed identity.
If several learners choose the new value as denominator, revisit the comparison concept with another example. If the class calculates correctly but confuses percentage points with relative change, focus the next lesson there. AI can generate practice variations after the teacher identifies the actual need.
For students: make the tutor ask you to think
A student can use a prompt such as:
Give me one percentage-change problem at a time. Wait for my setup before giving feedback. If I choose the wrong denominator, ask a question that helps me identify the original amount. Do not reveal the full solution unless I request it after an attempt. End with a new problem I solve without help.
Check AI feedback against a trusted answer or teacher explanation when something conflicts. Keep an error log showing what went wrong and how you corrected it. Follow your courseβs rules, especially for assessed work.
Practice: build an aligned lesson package
Create a short lesson, checked examples, a rubric, and an independent exit task for one skill. Map every activity to the objective. Include an access adaptation and explicit AI-use expectations. Ask AI to identify places where its own help could hide a learnerβs misunderstanding.
Completion check: The assessment measures the stated skill under known resource conditions, the answers are verified, and the plan distinguishes participation from evidence of understanding. Learner information is handled under institutional rules.
For a stretch exercise, adapt the lesson for workplace training using synthetic sales or service data while preserving the same mathematical objective.
Get the AI Dispatch
Weekly insights on ai & technology β delivered to your inbox. No spam, unsubscribe any time.
Want to choose specific topics? Customize your interests