P1-S6 across subjects: observing an AI-integrated lesson and giving peer feedback (a one-page six-column record and a 30-minute feedback meeting)

StageP1-S6
SubjectAcross subjects
Learning strandfundamental operating principles of AI / human-AI communication and collaboration / reflection on the application of AI
Official documentsBlueprint for Digital Education Development in Primary and Secondary Schools · AI Literacy Learning Framework for Primary and Secondary Schools · Guide to Using AI in Teaching in Primary and Secondary Schools · EDBCM No. 156/2026 · EDBCM No. 107/2026
Time to prepare, observe and meetabout 15 minutes of preparation (briefing, choosing how to record, fixing a time); one lesson observed; a 30-minute feedback meeting

A peer-observation guide: three things to settle before the lesson, a one-page six-column record (activity, use of AI, teacher's check, pupil verification, pupil's own work, one thing to improve), a 30-minute feedback meeting, three tests, and how the output joins the school's self-evaluation cycle. Nothing is graded and nothing goes into staff appraisal.

Two teachers with no facial detail sit side by side at a pale table, one holding up a blank sheet and the other a deep red pen, with a stack of blank cards and a royal blue folder on the table.

AudienceTeachers
ThemeSchool planning

EquipmentThe one-page six-column record (photocopiable on its own); a pen; the one-page agenda for the feedback meeting
Version1.0
Last updated2026-10-28

What an observation is for is seeing where AI was used in this lesson, who checked it, and what the pupils did themselves — not whether this teacher taught well.

What this solves

Once a school starts using AI in class, two kinds of observation appear, and neither helps the next lesson:

The old observation form. It asks whether the explanation was clear, whether the questioning had layers, whether the timing was sound. Those still matter, but not one column can show what AI did in the lesson — so the record and the lesson do not match, and the feedback meeting can only talk about technique.
The tool review. The opposite: the whole lesson is spent on whether the tool was easy, whether the interface was smooth, whether pupils were excited. A usable tool is not the same as learning, and the next lesson uses a different tool, so the conclusion is worthless.

This resource offers a third way: a one-page six-column record that captures only what is visible in this lesson, with a 30-minute feedback meeting and three tests. The attachment is an editable Word document; the record photocopies onto one page and goes into the room with you.

Three boundaries must be stated first.

Nothing is graded and nothing goes into staff appraisal. The record has no score column, no grade, no scale. Its only purpose is the next lesson. The moment a record can reach a personnel file, an appraisal or a promotion discussion, the lesson becomes a performance and the record stops working.
This is this resource’s suggested record, not an Education Bureau observation instrument. The official documents this resource draws on contain no “AI lesson observation indicators” and prescribe no columns for an observation record. The six columns, the three tests and the meeting below are this resource’s suggestion, and a school may change them.
This page looks at one lesson. School-wide or cross-subject coverage counting is a different tool’s job; bringing a coverage question into a single lesson’s observation ruins both.

Purpose

With this resource, the observer and the teacher being observed can:

Spend 15 minutes before the lesson settling why AI is being used, instead of guessing on the way in.
Use a one-page six-column record to capture visible facts rather than impressions and judgements.
Use a 30-minute meeting to turn the record into one thing to improve and one change that can be made next time.
Use three tests to judge whether the AI use in a lesson holds up — not whether the teacher does.
Accumulate the “one thing to improve” from several observations into the panel’s self-evaluation evidence, joining the school’s existing self-evaluation cycle.

How to use it

Step 1: The three things to settle before the lesson (about 15 minutes)

All three happen before the lesson. Miss one and the record turns into guesswork halfway through.

A 15-minute briefing: the teacher explains the learning objective and why this lesson uses AI. The observer is listening for which learning objective the tool serves, and what the class would learn less of without it. There is no need to discuss method here; this is about purpose.
Agree how to record: one page, six columns, words only. No video, no photographs, no pupil names (use seat numbers or codes), no collection of pupil work. The record goes back to the teacher the same day; the observer keeps no copy.
Fix the meeting: within three working days of the lesson, 30 minutes. Open it by saying that the record is not graded, does not go into appraisal and will not be seen by a third party. Say it once before the lesson and again on the day.

Step 2: The one-page six-column record (during the lesson)

The order of the columns is deliberate: from the teacher’s design, to the pupils, and only then to one thing to improve. Reverse it and the record becomes a conclusion looking for evidence.

Column	What goes in it	How it reads (example)
1 · Activity	How the lesson divides up, and how long each part took	“Opening 5 min / group task 20 min / reporting 10 min / close 5 min”
2 · Use of AI	What AI was used for in each part; write “none” where it was not used	“Group task: rewrite three openings with the tool, pupils choose one and explain why”
3 · Teacher’s check	Where the teacher made a judgement (choosing the task, removing data, limiting the output, rejecting an answer, requiring a redo)	“Removed the school name from the model text before demonstrating; rejected one of the tool’s examples and used their own”
4 · Pupil verification	How pupils checked the tool’s output; whether anyone actually stopped to look	“Two groups checked a date against the textbook; one group adopted the answer without checking”
5 · Pupil’s own work	Where the pupil wrote or revised something themselves; how much of the product shows their own judgement	“Each pupil wrote 50 characters first, then compared with the tool; three pieces kept the original wording”
6 · One thing to improve	Facts and location only: which activity, and what happened	“Activity 2: three of four groups adopted the tool’s example without any checking step”

Column 6 is the easiest to get wrong and does most to set the tone of the meeting. Facts and location, no evaluation:

Write “Activity 2: three groups adopted the tool’s example without any checking step”, not “pupils are too dependent on the tool”.
Write “Activity 1: the opening ran 12 minutes, leaving 13 for the group task”, not “poor time management”.
Write “Activity 3: no one mentioned what they had changed when reporting”, not “weak presentation skills”.

A fact with a location gives the meeting something to work on; a judgement turns the meeting into a defence.

Step 3: The 30-minute feedback meeting (within three working days)

Five parts, in this order. The first belongs to the teacher, and it is the most important step in the process.

Time	Content	Who speaks
0-5 min	The teacher speaks first: what I meant to achieve, what actually happened, and where it differed from what I expected	The teacher
5-13 min	The observer reads only the facts on the record, column by column, adding no evaluation. Then asks: “have I recorded anything wrongly?”	The observer
13-20 min	Choose one thing to improve together — one only. Start from column 6, but the teacher chooses, not the observer	Both
20-27 min	Write one change that can be made next time: which activity, changed how, and what is needed	Both
27-30 min	Say where the record goes: back to the teacher, no copy kept; the “one thing to improve” goes into the panel’s self-evaluation record	The observer

Three rules:

The observer does not interrupt the first part. If the teacher hears an evaluation before finishing the design intent, the whole meeting becomes explanation and defence.
One thing only. A list of three strengths and three weaknesses changes nothing next time; one change that is actually made earns a second observation.
The change must be small enough to make next time. “Redesign the whole unit” is not a change; “add a three-step checking card to the desks in activity 2” is.

Step 4: The three tests (used in the meeting, never as a score)

The three tests judge whether the AI use in a lesson holds up. Each is either visible or not visible; there is no middle grade.

The learning objective comes first. The lesson’s use of AI can state which learning objective it serves. If it cannot, the activity came before the objective. The question is not “was AI used” but “why was it used”.
The check sits at a key point. Of the three points — input, output, adoption — the teacher made a visible judgement at least once (changed the task, removed data, limited the output, rejected an answer, required a redo). With no judgement at any of the three, the tool is effectively teaching the lesson.
The pupils left their own judgement in it. Pupils can say what they checked and what they changed, and the product shows their own choices. A product with no trace of the pupil’s hand does not show that learning happened.

Three cautions:

The three tests are not a marking scheme. Do not total them into a score or write “all three met”. Use them to decide what the meeting discusses.
Do not apply them during the lesson. Record facts while observing; the tests belong to the meeting, otherwise the record is led by the conclusion.
They can be used to reject a design. If the briefing already shows that none of the three would be met, the design is worth changing before the lesson is taught and observed.

Step 5: Joining the school’s self-evaluation cycle (once a term)

The observation is not the point; its output has to answer “what has our subject changed about teaching with AI”.

Two or three a term, no more. More observations with hasty meetings are worth less than two done properly.
Copy the “one thing to improve” into the panel’s annual self-evaluation record, one line each, with the date, the class and the activity. By the end of the year that page is the evidence of what was changed, and the starting point for the next observation.
Join the three school-level nodes: planning, implementation, evaluation. The observation and its meeting are evidence for implementation; the panel’s record is used at evaluation; at planning it decides which subject and which kind of lesson to look at next year.
Do not create a second system. If the school already has an observation arrangement, put the six columns inside it rather than running both.

Review checklist

Check	Done properly	Needs work
Before the lesson	All three settled, including why AI is used, at the briefing	Asking on the way in which tool the lesson uses
The record	One page, six columns, every column filled; “none” where AI was not used	Comments on technique only, with no column for the use of AI
Column 6	Facts and location only (which activity, what happened)	“Pupils are too dependent” or “poor time management”
Scores	No score, grade or scale anywhere on the record	A score column or a total
Appraisal	Said at the start that it goes into no appraisal, and the record goes back to the teacher	The observer keeps the record, or appraisal is mentioned
The meeting	30 minutes, teacher first, observer reads facts only	The observer evaluates first, or it becomes an inquisition
The change	One only, naming the activity and the action	Five recommendations, none with an action
The tests	Used to choose what to discuss, never totalled	The three tests added up as “how far met”
Pupils	No video, photographs, names or collected work	Pupil names or photographs in the record

The test: when the meeting ends, the teacher is holding a next action, not a set of comments. Without one change that can be made next time, the observation was only an inspection.

Files

Observing an AI-integrated lesson and giving peer feedback (editable Word document)

Four parts: ① the checklist of three things to settle before the lesson ② the one-page six-column record, photocopiable on its own to take into the room, with columns for activity, use of AI, the teacher’s check, pupil verification, the pupil’s own work, and one thing to improve — column 6 has space for facts and location only ③ the 30-minute feedback meeting, with prompts for whoever chairs it ④ the three tests and how the output joins the self-evaluation cycle. The record has no score column, no grade and no scale.

Official documents

Every resource in this library is written to the AI Literacy Learning Framework for Primary and Secondary Schools and falls within the policy scope of the Blueprint for Digital Education Development in Primary and Secondary Schools.

This resource maps to:

Blueprint for Digital Education Development in Primary and Secondary Schools (three-fold leaflet, p.2): requires the delivery of digital and AI education to be planned in the School Development Plan, with a clear implementation timetable; peer observation is one kind of evidence at the implementation stage.
Supplement I: AI Literacy Learning Framework for Primary and Secondary Schools (p.1-4): sets out the three learning strands; this resource covers the fundamental operating principles of AI, human-AI communication and collaboration, and reflection on the application of AI.
Supplement I: AI Literacy Learning Framework for Primary and Secondary Schools (p.1-13): sets the direction for assessing AI literacy; column 5 and the third test are designed to it.
Supplement II: Guide to Using AI in Teaching in Primary and Secondary Schools (pp.2-2 to 2-5): the seven principles, including putting people first with technology as an aid and the teacher’s professional gatekeeping; column 3 is designed to it.
Supplement II: Guide to Using AI in Teaching in Primary and Secondary Schools (pp.2-7 to 2-8): sets out requirements on learning and teaching; the meeting is about learning and teaching, not about what the tool can do.
EDBCM No. 156/2026 (2026/27 School Year Holistic Planning of School-based Digital Education and Related Teacher Training): paragraph 3 sets out the requirements on peer observation and professional gatekeeping; the observation and meeting here follow it.
EDBCM No. 107/2026 (Training on Implementing Artificial Intelligence Education Planning in Schools and Teacher Professional Training in Artificial Intelligence): paragraph 4 describes the school-level self-evaluation cycle through the three nodes of planning, implementation and evaluation; step 5 says how the output of an observation joins those three.

Learning strand: fundamental operating principles of AI / human-AI communication and collaboration / reflection on the application of AI

Official documents page: https://www.edb.gov.hk/en/edu-system/primary-secondary/applicable-to-primary-secondary/it-in-edu/debp.html

This page is an independent Beets teaching resource, not an official EDB document. The official documents it draws on contain no AI lesson observation indicators and prescribe no columns for an observation record; the six-column record, the three tests and the meeting are this resource’s suggestion for schools to adopt, not an Education Bureau requirement. Nothing is graded and nothing goes into staff appraisal. This page covers what happens in one lesson; school-wide or cross-subject coverage counting is outside its scope.

Feedback

If you have any comments or questions about this resource, or an idea for the next one, tell us on the feedback page.

Related resources

Reuse

Download and adapt these resources freely for teaching; please keep the attribution.