Check whether AI-assisted lesson planning saves you time by measuring the whole preparation task, from gathering the material to producing a version you have checked and can use. Count prompting, checking, correcting and formatting, not only the time until an answer appears. Compare that total with a similar task prepared through your usual process, and keep the quality requirements the same.
This is an optional teacher-facing workload check using institution-approved tools and material you are permitted to provide. It is not a controlled trial, a pupil assessment or a reason to upload student information. Keep names, work, grades and support records out of the exercise.
Define the same finished task on both sides
Choose one familiar planning task, such as preparing a short practice activity from an approved source. State what the finished material must contain and which checks it needs. An unreviewed outline is not equivalent to a checked worksheet with usable instructions.
For the usual process, record the actual preparation you perform. If you already have a suitable approved resource, include adapting that resource as a realistic option. Comparing AI with an unnecessarily elaborate start-from-scratch process would answer a different question.
A 2024 NFER trial examined lesson and resource preparation for Year 7 and 8 science in England, with a guide supporting ChatGPT use. Its findings concern that particular trial, not a promised saving for your subject, tool or planning task. A personal work log cannot reproduce the trial's causal comparison.
Keep review and rework inside the measurement
Use a short note or an approved local timer to record the parts of the task. Separate time spent working from time spent waiting if that distinction matters to your question. Waiting that blocks your preparation still affects when the material is ready; waiting during which you complete another task is not necessarily time spent working on this one.
- Preparing the source material and instructions.
- Writing prompts, inspecting responses and asking for revisions.
- Checking facts, answers, instructional fit and required materials.
- Making corrections and preparing the actual worksheet or display.
- Resolving a problem discovered after the first apparent finish.
Record an interrupted or abandoned attempt rather than keeping only the convenient examples. Keep initial tool familiarization visible too. You may want to understand recurring preparation separately from learning a new tool, but neither should disappear from the account.
England's Department for Education guidance requires professional judgment about generated material's accuracy and appropriateness. Follow your institution's own requirements. Completing the timing exercise does not approve a tool or certify privacy compliance.
Read the total beside the quality result
Consider a fictional example, not a measured classroom result. A teacher spends 28 minutes preparing and checking an activity through the usual process. For a comparable AI-assisted task, preparation and prompting take 6 minutes, waiting blocks progress for 4 minutes, review takes 12 minutes, and correction and formatting take 8 minutes. That is 26 minutes of active work and 30 minutes until the material is ready.
Reporting only the first 6 minutes would hide most of that task. Reporting 30 minutes as entirely active work would hide the waiting distinction. In this invented comparison, neither figure establishes that AI saved time, and the two different tasks may have differed in difficulty.
Now check the material against the agreed finish criteria. If an answer remains unverified or the final worksheet is incomplete, record that it is not ready. A shorter incomplete attempt is not a faster equivalent result. The separate guide to reviewing AI-generated quiz questions covers answer accuracy and ambiguity; the timing record should include that work when it belongs to your task.
Make a narrow decision from the actual record
Look across comparable attempts before changing your routine. Note differences in subject familiarity, task complexity, available resources, interruptions and tool experience. Keep individual records so that a convenient average does not conceal an unfinished task or substantial correction.
Your conclusion can be modest: this task took less active work in these observed attempts, or the available examples are not comparable enough to decide. It may be useful to change the prompt, use an existing resource or stop using AI for that particular task. Do not turn a preparation-time observation into a claim about learning, overall working hours or another teacher's likely result.
A lesson revision log explains why a teaching change was made. This workload record answers a different question: what work and waiting were actually needed to reach the checked result? Use the records together without treating either as proof of improved learning.
Start with one bounded task and an honest account of the whole process. Keep the required review in the workflow even when it changes the apparent saving.
Explore AI & Technology and AI Skills. The evaluation section of AI Onboarding for Educators is a relevant starting checklist. Its illustrative goals are not measured results, and its tool examples do not establish current institutional approval or privacy compliance.
