Extension-Enabled Professional Development
Learning about Learning — Reflection, Transfer & Accountability
Upload your Phase 1–3 artefacts to an AI assistant. Ask it to identify patterns in your work — recurring themes, thinking evolution, blind spots. Be generous with the prompt; let the AI do real work here.
Then annotate every single insight it offers with one of three tags:
- Accepted — the pattern is real and you agree with the framing. Nothing to add.
- Modified — the pattern is partly real but the AI’s framing is off in a way you can name. Write the modification.
- Rejected — the pattern is not real, or is an artefact of how the AI reads your work. State the subject-specific reason the AI could not have known.
At least two rejections must identify an echo-chamber effect — point at a specific “insight” and say: “This is the AI repeating back to me the ed-tech framing I wrote in. It isn’t telling me about my teaching; it’s telling me about the language I write about my teaching in.”
Write at least 600 words. This is a reflection artefact, not a report — write it in your own voice.
Evaluation criteria: D4.1 AI-Assisted Professional ReflectionYou will be assigned an AI tool you have not used before. Set a 45-minute timer. Evaluate the tool using the Day 1 A/B/C taxonomy and the Day 2 boundary analysis habits. At the end of 45 minutes, write a report with:
- Extension type(s) — A, B, or C in your subject? Could it be different types in different activities?
- Boundary conditions — where would you not use this tool? At least three situations.
- Pedagogical potential — the best thing this tool could do for a student in your subject.
- Primary risk — the most likely way this tool could quietly harm learning if used without attention.
- Adoption recommendation — would you use it, and in what role?
The time constraint is the point. Level 4 requires completion in under 30 minutes. This is a stress test for whether the evaluation habits from Days 1 and 2 have become reflexive.
Evaluation criteria: D4.2 Extension-Transfer CapabilityTake one AI-related decision from Phase 3 (real, traceable to your decision log). Write three short communications about that decision — maximum 200 words each:
- To a student in your class — age-appropriate, honest about what the tool did and didn’t do, positions them as a thinker not a consumer.
- To a parent or carer — plain-language, addresses “is this safe, is this replacing teaching” without jargon, honest about limits.
- To an administrator or department head — curriculum-and-outcomes language, connects the decision to learning outcomes, names the evidence.
All three must specify the extension type, the decision rationale, and the AI’s limits. What they must not do is read like three cut-and-paste versions of each other. The language has to be demonstrably different.
Evaluation criteria: D4.3 Extension Accountability CommunicationA plan with at least three goals for your next three months. Code each goal:
- Self-identified — you noticed this gap from your own reflection on Phases 1–3.
- AI-proposed — came from the Task 1 self-analysis (an Accepted or Modified insight).
- Modified-from-AI — the AI raised it, you restructured it materially before adopting it.
For each goal, include a verification method — how will you know in 3 months whether you’ve actually grown? “I’ll feel more confident” is not a verification method. “I’ll run a Day 2-style prompt log on a new unit and compare the rationale quality” is.
At least one goal must address a specific limitation from your Phase 1–3 work. If every goal is forward-looking without touching what actually went thin this week, you haven’t really learned from your own material.
Evaluation criteria: D4.1 AI-Assisted Professional Reflection- AI transcript is included — evaluator can see what the AI said
- Every AI insight has an explicit annotation, not just disagreements
- Echo-chamber identifications explain how the AI is reflecting back rather than analysing
- Tool was genuinely new to the teacher
- Timing evidence present (timestamps or stated duration)
- A/B/C classification applied, not just feature evaluation
- Recommendation grounded in this unit’s needs, not personal preference
- Three documents recognisably different in register, not just in header
- Extension types named in audience-appropriate language
- Limits are honest, not “AI is perfectly safe”
- Goals are coded — teacher is transparent about the source
- Each goal has a concrete verification method
- At least one goal addresses a limitation from Phases 1–3