1
Think first
Students record what they already know, what they need to learn, their audience and the evidence a good answer would need.
ASEFedu ClassNet19 Innovative Teaching Practice · Open educational resource
Seed prompts for human agency
Help students design a learning task, question what AI produces, and make their own decisions visible. Share the teaching intention; adapt the details to your classroom.
The core idea
Think first. Prompt second. Judge third.
A seed prompt captures a learning goal, task structure and criteria that another teacher or class can adapt. The AI output is material to investigate. Students remain responsible for what they use and communicate.
2
Prompt second
Co-design a reusable seed with a clear role, context, task, form, tone, constraints and evaluation criteria (RCT–FTC–E).
3
Judge third
Check the output against disciplinary evidence and local knowledge. Accept, adapt or reject specific claims and explain each choice.
The learning evidence is the decision.
A polished prompt alone does not show careful thinking. Look for the student’s original position, the evidence they checked, the revision they made, and the reason for their final choice.
The pilot also used PEIR: Position → Examine → Interrogate → Reconstruct. It is a useful deeper reasoning lens alongside the three-part classroom sequence.
Optional digital companion
Build the seed in Athena Hub
The Neural Nomads team used Athena Hub and its Prompt Workshop to practise structured prompting and revision. It can support the Prompt second stage after students have recorded their own starting ideas.
Open Athena Hub Prompt Workshop ↗Opens a separate website at its sign-in page. Check your school’s account and data policies before use.
A classroom route through the workshop
- Begin on paper. Students identify the goal, what they already know and how they will verify a response.
- Structure and revise. Use the workshop to develop a seed with Role, Context, Task, Form, Tone, Constraints and Evaluation criteria. Keep a copy of the first and revised prompts.
- Return to human judgement. Compare the AI response with evidence and record what students accept, adapt or reject using the decision sheet.
A practical lesson
Run one complete cycle
Use a familiar curriculum topic. Allow about 50–70 minutes, or split the stages across two lessons. Pair or group work is fine; one shared device can serve a class.
Start with one seed, one AI response and one justified decision.
- 5–10 min · Before AI
Set a question and a human baseline
Choose a topic students can check against course materials or local knowledge. Ask each group to write what it already knows, what it is unsure of, and what would count as evidence.
- 10–15 min · Design
Build a transferable seed prompt
Students fill in RCT–FTC–E. Ask: which parts carry the teaching intention, and which details should another class change? Review the prompt before it goes to any tool. An optional digital route is the Athena Hub Prompt Workshop.
- 10 min · Generate
Get one AI response
Use a school-approved tool to create a practice activity, explanation, dialogue or inquiry plan. Where access is limited, the teacher can generate and print one response for groups.
- 15–20 min · Verify
Make an Accept–Adapt–Reject record
Students select particular statements or suggestions, compare them with sources or community expertise, and record a decision and reason for each. Teacher and peer checks matter here.
- 10–15 min · Transfer
Adapt and explain
Another group or partner class changes the prompt to fit its own curriculum, language, local examples or resources. Compare what stayed, what changed and why.
Ready to use
Classroom templates
Print these sections or download ready-to-print PDFs. The prompts are guides for thinking, not scripts students must copy unchanged.
1 · Seed prompt canvas
Download PDF ↓Before prompting: What do I already know? What am I trying to learn? What will I check independently?
Role What perspective or function should the AI take?
Context Who are the learners? Which curriculum, place, language and prior knowledge matter?
Task What learning material or inquiry support should it create?
Form What structure would make the output usable?
Tone What language level and voice fit the learners?
Constraints What should be included, avoided or left for students to decide?
Evaluation How will students check accuracy, relevance and contextual fit?
2 · Human decision record
Download PDF ↓Record the original idea, the AI suggestion, the evidence or person consulted, and the final choice.
Accept What is well supported and useful? Why?
Adapt What needs correction, local detail or a different perspective? Why?
Reject What is inaccurate, unsupported or unsuitable? Why?
Reflect What did you decide yourself? What changed in your understanding?
Adaptable example · freshwater and ocean chemistry
“Create a short investigation plan for learners aged 15–17 about how rising CO₂ changes water chemistry. Include a simple mechanism, a local ecosystem question and three claims students can verify. Leave the conclusion open. Mark any claims that need checking against a named scientific source or local data. Use language appropriate for our class.”
Transfer: a freshwater class might use a local river or lake; a coastal class might consider seawater, shell-forming organisms and community livelihoods. Both retain the learning goal and check the chemistry independently.
From the team’s implementation
One structure, different contexts
These short accounts describe the team’s reported practice. They are examples to adapt, not evidence that every classroom will obtain the same result.
Myanmar · Cultural heritage
Whose knowledge counts?
Thirty students in six teams established a human baseline, checked AI descriptions against local and community knowledge, then used a Heritage Decision Record. The team reported that 28 students wrote a decision with a reason. Printed sheets and shared phones supported participation.
Teacher move: ask whose perspective is missing before students decide how to represent a tradition.
Philippines + Austria · Science
From chemistry to communication
The CHEM-ANIMATE collaboration connects freshwater and ocean acidification. Philippine students used structured inquiry, checked AI-assisted research against scientific sources, and developed storyboards and animation about ocean acidification. Austrian classroom exchange and further chemistry work were planned in stages.
Teacher move: transfer the investigation’s intention, then choose local water systems and evidence.
Switzerland · Prompt workshop
Make the task design visible
Johann Sievering developed and tested a workshop using RCT–FTC–E, manual prompt revision, AI generation and group reflection. It informed the team’s structured authoring approach and Athena Hub.
Teacher move: compare the first and revised prompts, then ask whether the final judgement improved too.
Across classrooms · What to notice
Collect traces of judgement
Keep an initial position, one prompt revision, a checked claim, a decision record and a short final reflection. Compare student work before and after the cycle; ask who or what helped them change their mind.
Teacher move: assess reasoning, subject understanding and justified choices alongside any final product.
Responsible use: Follow your school’s AI and privacy rules. Avoid entering identifiable student information or sensitive cultural material into a public tool. Verify factual claims with relevant sources and consult people with authority over local knowledge.
Reuse and adapt
Take the resource into your classroom
The printable pack can be used without this website or a specialist app. Print it, translate it, change the examples, and share your adapted version with attribution. The source download includes the editable text files.
Teacher guide (PDF) ↓Lesson sequence, facilitation, inclusion and assessment.Seed prompt canvas (PDF) ↓Pre-AI thinking and RCT–FTC–E fields.Human decision record (PDF) ↓Evidence-backed Accept–Adapt–Reject reflection.Partner adaptation sheet (PDF) ↓Compare intentions, changed details and learning.Website source code ↓HTML and licensing files for remixing or self-hosting.
Athena Hub is an external companion, not part of this website’s open-source download. No AI tool is required to use the planning and reflection activities. If using an AI tool, the class can share a teacher-generated response on paper.