Switch course ▾
Session 5. Data, Fallbacks and Your AI-Enhanced Lesson Pack
Objective. Apply data-protection habits, prepare an offline fallback for every AI-assisted task and assemble a one-page lesson pack that marks the decisions kept by the teacher.
A lesson pack is only as good as its weakest link, and the weakest links are usually the least exciting ones: a pupil's name pasted into a tool, or a lesson that cannot run when the power fails. This final session turns privacy and fallbacks into simple habits, and shows how an AI-use log keeps your judgement visible. By the end you will have one lesson you can teach tomorrow with AI, without AI, or with a power cut.
Key ideas
What counts as personal data
- Names, photos, voices, marks, attendance, health, family details and home addresses. In a small class even 'the quiet girl who sits by the window' can identify a child.
- Data minimisation: give a tool only what it needs, typed and anonymous. Typed lists of common errors work; photos of exercise books or the class register do not.
- Free tools may store or reuse what you type. Read the tool's settings and your school's rules first; if you are unsure, do not enter it. Ask the school lead which tools are approved.
An offline fallback for every task
- Use AI for preparation, not as something the lesson depends on. Power cuts, slow internet, a tool that is down or a draft you cannot check in time must not stop your lesson.
- For each AI-assisted task write one line: 'If no AI: ...'. It can be the textbook exercise, a hand-copied version, a colleague's sheet or a board task. Print or copy it before the lesson.
- Decide your switch points in advance: no device, no time to check, doubt about privacy. When one is true, go to the fallback without guilt.
Judgement stays with the teacher
- Some decisions are never delegated: marks and grades, long-term grouping or placement, discipline and cheating verdicts, messages to families about a child, and safeguarding or health concerns.
- Keep a two-line AI-use log: date, tool, purpose, what you checked, what you changed. It helps you improve and gives you an answer if a parent or the school lead asks.
- Be open with pupils and colleagues about when you use a tool; it models the honesty you ask of them in session 4.
Try it in your next lessons
- Teach one real lesson from your lesson pack, with the fallback printed and ready. Share your pack and prompt with one colleague and agree a class AI rule with your pupils.
- Evidence: a 5-minute observation note (yours or a colleague's) on what worked, what you checked and whether you needed the fallback.
- Evidence: an exit ticket on the objective, your AI-use log entry and the date you will review the pack.
May it go into a tool?
| Item | Decision |
|---|---|
| Typed list of common errors, no names | Yes |
| The textbook unit and your objective | Yes |
| A pupil's essay with the name on it | No: remove identifiers or do not share |
| Class register, marks, health or family notes | No, never |
The AI-use cycle
A three-question data test
Before anything goes into a tool, ask three questions. Could this identify a child, directly or through details such as class, seat or a family story? Would I be comfortable if the parent read exactly what I typed? Does the tool really need it to do this job? If the answer to the first is yes, or the third is no, remove it.
You can almost always get the same help without data. Type common error types instead of exercise books, invent a sample pupil answer instead of using a real one, or ask a general question and apply the answer yourself. Ask your school lead which tools are approved, and check whether a tool lets you stop it saving your inputs, but never rely on a setting to protect sensitive information.
Making a fallback a real plan
A fallback only helps if it exists on paper before the lesson. Prepare it when you prepare the AI version: copy the task to a sheet or the board, write the answer key by hand, and put both in a folder. Define the trigger in advance, such as 'no power at 8:45' or 'I could not check the draft'. Time the fallback once so you know it fits the 45 minutes.
The fallback does not have to be the same activity; it has to serve the same objective. A folder of three ready generic activities per subject, such as a retrieval quiz, a paired explanation and a short problem set, protects you on bad days. Review the folder each month and replace what you used.
Examples from the classroom
Remarks without names
Scene. Malika needs short messages for parent meetings. Instead of uploading the register, she writes five types of remark on paper, such as 'forgets homework' or 'helps classmates', with no names. The tool drafts neutral wording for each type, she edits them, and she picks which fits which child. She writes every final message herself.
Insight. The tool helps with wording; the knowledge of the child, and the decision, stay with the teacher.
The cold morning
Scene. Aziz teaches Grade 7 in Namangan. On a winter morning the power goes off at 8:40, and his AI-made activity needed a screen. His fallback folder holds a printed version of the same task and a hand-written key. The class starts on time, and he notes in his log that the paper version took 3 minutes longer to hand out.
Insight. A fallback prepared in advance turns a crisis into a small adjustment.
A parent asks
Scene. At a parent meeting in Samarkand, a mother asks Dilnoza whether 'the computer marks my daughter's work'. Dilnoza opens her AI-use log: she used a tool to draft comment ideas from error types without names, edited them herself, and marks all work by hand. The mother is reassured and asks how to talk with her daughter about homework rules.
Insight. A simple log lets you answer questions about AI openly and calmly.
Common pitfalls
- Believing that deleting a name is enough when the class, seat and story still point to the child.
- Treating a tool's privacy setting as a licence to enter sensitive information.
- Preparing a fallback in your head rather than on paper.
- Handing the final wording of a message to parents to the tool.
Build Your AI-Enhanced Lesson Pack
- Alone (15 min): choose a real lesson you will teach soon. Fill the one-page lesson pack from the Activity Kit: objective, prompt, checks, versions and feedback, pupil AI rule, data check, offline fallback and the decisions that stay yours.
- In pairs (10 min): swap packs and apply the seven-point peer checklist. Mark each gap with a question, not a verdict.
- Alone (10 min): revise your pack. Pay special attention to the fallback (could you teach it tomorrow with no power?) and to the data check.
- As a group (10 min): each person reads one thing they will never enter into a tool and one decision that stays theirs. Agree three practices for your school: an AI-use log, a shared prompt bank and a class AI rule.
Your facilitator will guide you through this activity in the session.
Questions to discuss
- What did the peer reviewer's question make you add or change in your pack?
- Which of your tasks has the weakest fallback, and how will you strengthen it?
- Which of the three school practices will you start first, and who will you ask to join you?
Reflect
- What have you entered into a tool in the past that you would now remove first?
- How would your next lesson go if there were no electricity for the whole morning?
- Which decision about a child are you most tempted to hand to a tool, and why should you not?
Try it tomorrow
- Apply the three-question test to the last five things you typed into any tool.
- Make a folder with a printed fallback for your next AI-assisted lesson and time it once.
- Start the AI-use log with two lines for this week's use.
Further reading and links
Keep children's data out of tools, have a no-AI version of every task, and keep the decisions about children in your own hands.