A curriculum for intelligent AI use · Grades 4–12
The thinking stays with the student.
Students learn how AI works, when to keep it off, and how to make the first move, delegate one declared, bounded job, verify what comes back and defend the work they make.
Each star opens a curriculum design example. Select one with a pointer, or explore the full curriculum architecture and framed lesson catalogue below.
More than tool tips
Eleven AI skills. One accountable way to work.
“AI skills” names the practical layer, not the whole curriculum. Students learn eleven specific moves, the knowledge that makes those moves intelligent and the judgment to decide when AI should stay off. Those skills sit inside Act → Delegate → Verify → Defend, with one data-safety rail running throughout.
Why now: when any student can generate a finished answer, the answer stops being the evidence — the thinking behind it is. Own Mind is designed to keep that thinking visible through the student's first move, AI's declared role, the checks and the final explanation.
Each lesson is designed to connect an AI skill, curriculum knowledge and a human decision.
One curriculum architecture
Eleven AI skills, beside ten knowledge strands.
They are two connected layers, not two versions of the same list. The ten strands describe what the curriculum teaches. The eleven AI skills name the specific moves lessons are designed to practice. Together, they give students the understanding and practical judgment to work intelligently with AI.
Named AI skills
Concrete moves that return through lessons, projects and subject work.
- S1
Steer it
Give AI one clear job, with a role, source boundary, expected output and stopping point.
- S2
Make it attack your own work
Start with your own idea or draft, then use AI to challenge it and test the challenge.
- S3
Check, don't trust
Check claims against sources, data, calculation or the real thing before relying on them.
- S4
Spot the slant
Notice framing, omissions, uneven representation and whose perspective is missing.
- S5
You make the call
Weigh the evidence and keep the final decision human.
- S6
Make sense of a big pile
Use AI to sort or compare a large set, then inspect what it grouped, flattened or missed.
- S7
Get unstuck
Ask for a hint, question or debugging move that leaves you making the next move.
- S8
Pick the right job
Decide whether AI is useful, allowed and necessary, or whether another tool or person is a better fit.
- S9
Know when to do it yourself / AI off
Keep AI off when using it would replace the learning, judgment or responsibility that must stay yours.
- S10
Show what was yours (provenance)
Record what you made first, what AI contributed and what you accepted, changed or rejected.
- S11
Judge how much to trust it (calibration)
Match confidence to the strength and limits of your check; accept, reject or hold uncertain for a reason.
Knowledge strands
Ideas that return across subjects, projects and grade bands.
- 01
AI mental models
Prediction, models, limits and uncertainty.
- 02
Data and representation
Training data, classification, bias, omission and provenance.
- 03
Briefing and delegation
Define a job, inputs, outputs, limits and human approval.
- 04
Truth and evidence
Sources, corroboration, claims, hallucinations and recency.
- 05
Cognitive custody
Act, use the smallest help, preserve judgment and transfer with AI off.
- 06
Creation and craft
Make quality work across media and use specialist tools.
- 07
People and inclusion
Listen, design for access, examine power, benefit and harm.
- 08
Testing and revision
Run user, source, experiment, bias, access and edge-case tests.
- 09
Collaboration and leadership
Record decisions, roles, conflict, handover and accountability.
- 10
Venture and governance
Create value while governing operations, resources, AI and closure.
Inside an Own Mind lesson
Turn AI knowledge into work students can explain.
These examples show the intended pattern: what students learn, what they do, how they check it without AI, which human capability the work is designed to exercise and what a teacher could review. They are curriculum design examples—not released Own Mind lessons or modules.
61 examples
Use the filters, then scroll this framed catalogueStart Here: What AI Means
- Knowledge focus
- AI is a pattern-guessing tool, not a person
- Student action
- Students learn the starter words, open four strange machines, catch one confident mistake and write a simple rule for supervising AI.
- Independent check
- With the machine closed, they explain what AI can do, what it cannot do and what stays in their charge.
- Human capabilities this is designed to exercise
- Problem-framing
- Self-regulation
- Work a teacher could inspect
- AI Is / Is Not Board
What's the Job?
- Knowledge focus
- Match each job to its right next move — AI is only one option
- Student action
- Students sort four ordinary jobs, mark the part that must stay theirs and choose AI, search, a person or no tool as the next move.
- Independent check
- They build a fresh task tile naming the job, what stays human and why their chosen next move fits.
- Human capabilities this is designed to exercise
- Problem-framing
- Metacognition
- Work a teacher could inspect
- Task Tile
Help the Robot Understand
- Knowledge focus
- A clear prompt gets a more useful answer, not an automatically right one
- Student action
- Students decide what a question must achieve, compare two possible forms of help, then add their own focus and a way to check the answer.
- Independent check
- They write a clear request for a fresh topic and name one way they would test the answer.
- Human capabilities this is designed to exercise
- Problem-framing
- Metacognition
- Work a teacher could inspect
- Better Question Card
The No-Data Rule
- Knowledge focus
- Private information stays out of AI — remove it, or take it to a trusted adult
- Student action
- Students scan a message for private clues, replace unsafe details and decide when AI should stay out altogether.
- Independent check
- They write the No-Data Rule in their own words and identify the point at which a trusted adult takes over.
- Human capabilities this is designed to exercise
- Ethical judgment
- Self-regulation
- Work a teacher could inspect
- Personal Safety Card
Real, Changed or Can't Tell Yet?
- Knowledge focus
- Source clues, not pixels, tell you if a picture is real
- Student action
- Students give five mystery pictures a first verdict, test them against source clues and use “cannot tell yet” when the trail is incomplete.
- Independent check
- They explain what pixels can reveal and why provenance is needed to establish where a picture came from.
- Human capabilities this is designed to exercise
- Adaptive thinking
- Critical reasoning
- Work a teacher could inspect
- Real-or-Made Challenge
My Picture, My Choice
- Knowledge focus
- A picture of a person is theirs, and sharing it needs their yes
- Student action
- Students make consent decisions for fictional picture scenarios, find the share that was genuinely fine and write their own picture rule.
- Independent check
- They complete the rule: before I take, change or send a picture of someone, I will…
- Human capabilities this is designed to exercise
- Ethical judgment
- Self-regulation
- Work a teacher could inspect
- My Picture Promise
Says Who?
- Knowledge focus
- A claim is backed only when a real source actually says it
- Student action
- Students test three confident backing cards against real sources, catch a citation that points at nothing and distinguish support from mention.
- Independent check
- With every card closed, they explain “supports,” “just mentions” and what to do when a citation cannot be found.
- Human capabilities this is designed to exercise
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Says Who? Reveal Panel
Some Jobs Need a Person
- Knowledge focus
- AI helps with some jobs, but only a person can care and keep you safe
- Student action
- Students route five fictional jobs to a tool, a person who cares or a trusted adult straight away, then rehearse how to ask for help.
- Independent check
- They explain when a tool is fine, when a person is better and when a trusted adult is needed immediately.
- Human capabilities this is designed to exercise
- Ethical judgment
- Metacognition
- Work a teacher could inspect
- People I Can Ask Card
Learning Support, Supervised
- Knowledge focus
- AI is a study helper you supervise, not an answer machine
- Student action
- Students make a first call, choose the smallest useful kind of help, test two forms of support and complete the target themselves.
- Independent check
- With the help cards closed, they explain what they kept, what they rejected, what they added and what remains untested.
- Human capabilities this is designed to exercise
- Self-regulation
- Metacognition
- Work a teacher could inspect
- Help Request Card
Whose Work Is It?
- Knowledge focus
- Your own work and AI's help are separate parts — and you say which is which
- Student action
- Students sort mixed work into mine, AI help, copied and source-needed, then write an honest record of the help used.
- Independent check
- They judge a fresh case and explain exactly what a human would need to know about the AI contribution.
- Human capabilities this is designed to exercise
- Ethical judgment
- Metacognition
- Work a teacher could inspect
- My Work / AI Help Card
Is the Rule Fair Enough?
- Knowledge focus
- A fair rule is one you test on who's left out, not one that just treats everyone the same
- Student action
- Students write a clear chooser rule, run three people the rule might overlook through it and revise from what the test reveals.
- Independent check
- With the cards away, they explain what their rule decides, whom it nearly forgot and what would make them change it again.
- Human capabilities this is designed to exercise
- Problem-framing
- Self-regulation
- Work a teacher could inspect
- Fair Chooser Card
Make It Work for Someone Else
- Knowledge focus
- A finished make is planned, checked, and fit for its person
- Student action
- Students plan a make for one real person, keep, change or cut prepared suggestions, trace factual claims and run a recipient preview.
- Independent check
- They name one part that had to stay theirs and the condition under which AI would not belong in the make.
- Human capabilities this is designed to exercise
- Problem-framing
- Self-regulation
- Work a teacher could inspect
- Checked Make Receipt
Show the Make, Set the Rules
- Knowledge focus
- A showcase is showing your own chosen work, not being graded
- Student action
- Students choose work for a family showcase, add how-I-checked-it notes, answer an unseen question with AI closed and set shared family rules.
- Independent check
- They explain why the make is theirs, how they checked it and which family rule matters most to them.
- Human capabilities this is designed to exercise
- Collaboration
- Ethical judgment
- Metacognition
- Work a teacher could inspect
- Family Showcase Page and AI Agreement
How AI Works
- Knowledge focus
- Many AI models learn patterns from examples, then predict
- Student action
- Students train a tiny pattern machine, predict a label, test an unusual case and record the limit instead of trusting the gauge.
- Independent check
- They explain the difference between a rule, a training example and a prediction without the machine in view.
- Human capabilities this is designed to exercise
- Metacognition
- Adaptive thinking
- Work a teacher could inspect
- Pattern Machine Note
Catch a Hallucination
- Knowledge focus
- Confidence isn't proof — a claim is supported, contradicted, or not supported yet
- Student action
- Students inspect a polished AI briefing, pin evidence to claims and distinguish supported, contradicted and not supported yet.
- Independent check
- With AI and sources hidden, they rebuild the claim-check table and justify a careful hold when the evidence runs out.
- Human capabilities this is designed to exercise
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Hallucination Catch Case Card
Reading the Hidden Bias
- Knowledge focus
- Bias can hide in the data and the question, not just deliberate unfairness
- Student action
- Students use a bias lens to find what a polished answer leaves out, then repair the answer with visible changes.
- Independent check
- They inspect a fresh source set and name one missing group, example or viewpoint.
- Human capabilities this is designed to exercise
- Critical reasoning
- Ethical judgment
- Adaptive thinking
- Work a teacher could inspect
- Bias Lens Case Card
When Not to Reach for AI
- Knowledge focus
- AI is not the right tool for every job
- Student action
- Students run different situations through use, limit, no-AI and ask-a-human routes, then inspect the consequence of each choice.
- Independent check
- They choose a route for a fresh scenario and state the boundary that made the decision responsible.
- Human capabilities this is designed to exercise
- Metacognition
- Ethical judgment
- Adaptive thinking
- Work a teacher could inspect
- Tool Choice Card
Prompting on Purpose
- Knowledge focus
- A prompt is a plan you design, not magic wording
- Student action
- Students set a goal, success rule and boundary, compare two prompt plans, then revise the request themselves.
- Independent check
- They build a prompt for a new task from memory and explain how it defines useful output.
- Human capabilities this is designed to exercise
- Metacognition
- Problem-framing
- Work a teacher could inspect
- Prompt Test Log
Keep Your Voice
- Knowledge focus
- You keep your voice; polished AI writing is not automatically better
- Student action
- Students mark the choices that make a draft sound like its writer, test AI edits and protect the voice from a flattened rewrite.
- Independent check
- With AI closed, they explain one construction decision, one rejected change and one honest limit.
- Human capabilities this is designed to exercise
- Metacognition
- Problem-framing
- Work a teacher could inspect
- Voice Preservation Card
Build and Test a Local Classifier
- Knowledge focus
- A classifier follows patterns from its examples, not the meaning of the category
- Student action
- Students predict, train a local classifier, test unseen patterns, change its data, retrain and write a responsible-use model card.
- Independent check
- With the model closed, they explain what it compared, why one case needed review and where it must not be used.
- Human capabilities this is designed to exercise
- Critical reasoning
- Collaboration
- Work a teacher could inspect
- Local Classifier Model Card
Read and Debug Simple Code
- Knowledge focus
- You read and test code yourself before trusting a change, even AI's
- Student action
- Students predict what code will do, run a failing test, trace the bug, make a patch and explain why the test now passes.
- Independent check
- They read a fresh tiny program and explain its output or bug without AI.
- Human capabilities this is designed to exercise
- Metacognition
- Self-regulation
- Work a teacher could inspect
- Read-Run-Debug Card
Where Did This Come From?
- Knowledge focus
- Content is what it shows — provenance is where it came from
- Student action
- Students inspect a quote, photo, audio clip and news paragraph, then follow creation, source and context trails when appearance proves nothing.
- Independent check
- With every drawer closed, they explain the difference between content and provenance and recall the three trail questions.
- Human capabilities this is designed to exercise
- Critical reasoning
- Adaptive thinking
- Metacognition
- Work a teacher could inspect
- Provenance Trail
Make It Better Without Replacing It
- Knowledge focus
- Iteration is a decision you make, not a rewrite you accept
- Student action
- Students pin one rough make, ask for bounded feedback, test two notes, make one change themselves and refuse one takeover.
- Independent check
- With the notes closed, they reconstruct the change they made, the advice they refused and the line they protected.
- Human capabilities this is designed to exercise
- Metacognition
- Self-regulation
- Work a teacher could inspect
- Before / Feedback / After Strip
Forge, Label, Then Hunt
- Knowledge focus
- Anyone can make a fake — checked trails and honest labels help establish origin
- Student action
- Students create a harmless synthetic version of an artifact, label it honestly and hunt through an exhibit where trails—not sharp eyes—settle origin.
- Independent check
- They explain why looks cannot establish origin, what a trustworthy trail adds and how to hold judgment when no trail exists.
- Human capabilities this is designed to exercise
- Critical reasoning
- Ethical judgment
- Work a teacher could inspect
- Labeled Synthetic Exhibit and Real-or-Made Challenge
Your Face, Your Say / Real People for Real Feelings
- Knowledge focus
- Your image needs your yes, and real feelings go to real people
- Student action
- Students route fictional photo edits, recognize secrecy and dependence language and build a private list of people real feelings can go to.
- Independent check
- They state the consent rule for editing or sharing an image and the next move when a tool asks for secrecy.
- Human capabilities this is designed to exercise
- Ethical judgment
- Self-regulation
- Adaptive thinking
- Work a teacher could inspect
- Consent and Real-Help Card
Read the Ask, Not the Polish
- Knowledge focus
- Polish is cheap to fake — the ask is what you check
- Student action
- Students strip polished and scruffy messages down to their real ask, flag risky pressure and build a second-channel verification plan.
- Independent check
- On a fresh urgent message, they identify the ask, name the risk and choose an independent channel before acting.
- Human capabilities this is designed to exercise
- Critical reasoning
- Self-regulation
- Adaptive thinking
- Work a teacher could inspect
- Second-Channel Check Plan
Different Room, Different Rule
- Knowledge focus
- The rule follows the room, not the tool
- Student action
- Students carry the same AI action through five different contexts, watch the rule change and write an ask-first script plus disclosure.
- Independent check
- With the rules hidden, they judge the action in a fresh context using purpose, permission and disclosure.
- Human capabilities this is designed to exercise
- Ethical judgment
- Self-regulation
- Work a teacher could inspect
- Room Rules Card
The Second-Draft Wall
- Knowledge focus
- Error is a change in a model, not a label on a person
- Student action
- Students return to an earlier explanation, test each clause against a counterexample and repair only what breaks.
- Independent check
- They explain which clause changed, why the original evidence was insufficient and which parts still hold.
- Human capabilities this is designed to exercise
- Metacognition
- Self-regulation
- Adaptive thinking
- Work a teacher could inspect
- Model Repair Log
Who Made This Possible?
- Knowledge focus
- AI does not create from nothing — generated work stands on human work
- Student action
- Students trace the hidden human chain behind one generated object, identify missing links and build the credit block a project should carry.
- Independent check
- They trace a fresh object and write its credit block with the evidence face down.
- Human capabilities this is designed to exercise
- Ethical judgment
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Credit and Provenance Block
Use the Right Tool, Not the Flashiest One
- Knowledge focus
- The right tool is the one that passes the five checks, not the one with the most flash
- Student action
- Students run one job through seven possible tool routes, test each for fit, data, rules, control and checking cost, then commit to the route that survives.
- Independent check
- With trial cards hidden, they defend the route they chose, the route they refused and the decisive check for each.
- Human capabilities this is designed to exercise
- Metacognition
- Problem-framing
- Work a teacher could inspect
- Tool Plan Card
Make It Survive Another Person
- Knowledge focus
- Work for a real audience is tested by a reader, not by how finished it feels
- Student action
- Students build one real thing for one real reader, test whether it works without the student beside it and revise from what happens.
- Independent check
- With the project closed, they defend what they changed, what they refused to change and why the refusal protected the reader.
- Human capabilities this is designed to exercise
- Metacognition
- Adaptive thinking
- Problem-framing
- Work a teacher could inspect
- Survival Test Card
How Chatbots Guess Text
- Knowledge focus
- An LLM predicts likely text; its answer is not automatically a checked fact
- Student action
- Students build a toy next-token trail, watch fluent text drift and explain how a language model works without treating it like a mind.
- Independent check
- They explain training, use, context limits and confident error in their own words.
- Human capabilities this is designed to exercise
- Metacognition
- Adaptive thinking
- Work a teacher could inspect
- How Chatbots Guess Card
Rehearse How to Show AI Help
- Knowledge focus
- A process record shows how AI helped as you work — not proof that the work is yours
- Student action
- Students preserve a source-aware first sample, inspect two possible contributions, keep or reject each one and write a bounded disclosure.
- Independent check
- With the contribution cards closed, they reconstruct what they kept, what they rejected, what they changed and one remaining limit.
- Human capabilities this is designed to exercise
- Metacognition
- Ethical judgment
- Work a teacher could inspect
- Prepared Contribution Disclosure Record
Build and Test AI-Assisted Code
- Knowledge focus
- Code you test and own, not code you trust because it ran once
- Student action
- Students define the requested code change and existing behavior to protect, test two suggestions, make one bounded change and run the final suite.
- Independent check
- With AI closed, they explain the kept and rejected suggestions, their own change, the test evidence and one remaining limit.
- Human capabilities this is designed to exercise
- Self-regulation
- Metacognition
- Problem-framing
- Work a teacher could inspect
- Code Change Card
Compare Rules with a Trained Model
- Knowledge focus
- Fixed rules and data-driven models make decisions differently
- Student action
- Students compare a fixed rule with a classifier learned from examples, test both on held-back cases and change one rule from evidence.
- Independent check
- They explain how written rules and learned patterns differ, then name one evidence limit for each.
- Human capabilities this is designed to exercise
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Rules-versus-Model Boundary Record
Check Sources Like a Detective
- Knowledge focus
- A source can exist and still not support the claim
- Student action
- Students inspect citations for existence, relevance, support and date before accepting what they appear to prove.
- Independent check
- They test one fresh claim-and-source pair without AI and state whether the source truly supports it.
- Human capabilities this is designed to exercise
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Source Check Card
Explain Your Work Out Loud
- Knowledge focus
- Being able to explain your work, not just show it
- Student action
- Students rehearse an explanation privately, turn vague answers into specific ones and capture a human-facing defense without an AI verdict.
- Independent check
- They answer one unseen question about their saved work with AI closed.
- Human capabilities this is designed to exercise
- Metacognition
- Critical reasoning
- Work a teacher could inspect
- Spoken Check Card
The Humans Behind and Beside AI
- Knowledge focus
- Every AI system is held up by many human roles, most of them not coding
- Student action
- Students take apart a real-looking AI system, tag the human who decided or checked each stage and find where a person remains accountable.
- Independent check
- With the system hidden, they name three human roles, identify the accountable role and explain why software cannot hold accountability.
- Human capabilities this is designed to exercise
- Ethical judgment
- Critical reasoning
- Work a teacher could inspect
- Human-in-the-Loop Map
What AI Genuinely Does Well
- Knowledge focus
- Some AI genuinely does a hard job well — the skill is naming what it does well and where it still needs a human
- Student action
- Students inspect a published AI system, identify what it did well, find where it still needs a human and form a calibrated trust verdict.
- Independent check
- They state one genuine strength, one human-only check and who remains accountable.
- Human capabilities this is designed to exercise
- Critical reasoning
- Metacognition
- Work a teacher could inspect
- Exhibit Verdict Card
Diagnose the Task Before the Tool
- Knowledge focus
- Legitimacy lives in the task, not the tool
- Student action
- Students diagnose a real task through audience, success, constraints, load-bearing thinking, permitted help and failure before any tool opens.
- Independent check
- They diagnose a fresh task aloud, refuse one unsuitable help move and name when they would stop and ask a teacher.
- Human capabilities this is designed to exercise
- Metacognition
- Problem-framing
- Work a teacher could inspect
- Task Diagnosis Brief and Allowed-Help Contract
Coursework, Practice and the Line Between
- Knowledge focus
- Integrity is contextual: the same AI move can be fine in practice and malpractice in assessed work
- Student action
- Students route homework, revision, coursework and group-work cases through a rulebook and watch the same AI move change status with context.
- Independent check
- With the rulebook closed, they route a fresh case and explain what to do when the rules do not settle it.
- Human capabilities this is designed to exercise
- Ethical judgment
- Metacognition
- Work a teacher could inspect
- Academic-Integrity Checklist
Build a Revision Workflow That Makes You Recall
- Knowledge focus
- Keep the recall step yours: help around the attempt, never inside it
- Student action
- Students build an AI-off retrieval set, make a cold attempt, schedule spaced practice and decide which helper jobs are allowed before and after recall.
- Independent check
- They complete one retrieval attempt with everything closed, check afterwards and name any gap as a topic rather than a mark.
- Human capabilities this is designed to exercise
- Metacognition
- Self-regulation
- Work a teacher could inspect
- Revision Workflow Card
Take the Critique, Keep the Voice
- Knowledge focus
- Critique and authorship are different jobs
- Student action
- Students name the markers of a writer’s voice, take critique without takeover, decide on every point and rewrite accepted advice in their own words.
- Independent check
- With critique cards closed, they recall the voice markers, defend one rejection and rewrite one accepted point from memory.
- Human capabilities this is designed to exercise
- Critical reasoning
- Metacognition
- Work a teacher could inspect
- Voice-Kept Edit Trail
Privacy Before You Paste
- Knowledge focus
- A badge is not enough: check the input and the tool's rules before you paste
- Student action
- Students classify five inputs at the moment of pasting, read a tool’s real terms and choose cross, minimize, redact, replace or never.
- Independent check
- With the terms hidden, they classify a fresh input, state its risk and rule what would need to change before it could cross.
- Human capabilities this is designed to exercise
- Ethical judgment
- Self-regulation
- Work a teacher could inspect
- Tool-Risk Assessment
If This Ever Happens
- Knowledge focus
- A rehearsed calm response beats an improvised one
- Student action
- Students rehearse fictional high-pressure cases until they can pause, refuse, preserve evidence, verify another way and tell a trusted adult.
- Independent check
- They recall the five calm moves and apply them to a fresh fictional case, including the second channel and first human contact.
- Human capabilities this is designed to exercise
- Self-regulation
- Ethical judgment
- Work a teacher could inspect
- Calm-Response Plan
That Is a Person Job
- Knowledge focus
- A tool can help you find the words — a person makes the call
- Student action
- Students route fictional late-night conversations to a tool job, a person job or a person-now job and keep the one safe tool request.
- Independent check
- They explain what work a tool can hold, why important human decisions stay with people and what to do when the job changes.
- Human capabilities this is designed to exercise
- Ethical judgment
- Critical reasoning
- Work a teacher could inspect
- Human-Help Boundary Card
Who Gains, Who Loses, Who Decides?
- Knowledge focus
- Automation is a trade-off between people — audit who gains, who loses, and who decides
- Student action
- Students audit an automation plan from every stakeholder seat, judge each step and redesign the workflow so value and access survive.
- Independent check
- They apply who gains, who loses and who decides to one step of their own workflow, then record a reasoned ruling.
- Human capabilities this is designed to exercise
- Ethical judgment
- Critical reasoning
- Work a teacher could inspect
- Automation Trade-Off Brief
The Default Setting Has a Cost
- Knowledge focus
- A model's default is a pattern with costs — so proportionate use is a design decision
- Student action
- Students probe model outputs for defaults nobody requested, inspect resource costs and re-brief a project with better examples and a no-generation line.
- Independent check
- With the file closed, they name the strongest default, the pack that survived and one step where generation is not worth it.
- Human capabilities this is designed to exercise
- Problem-framing
- Ethical judgment
- Work a teacher could inspect
- Representation Probe and Generation Policy
Would You Trust This Tool With the Job?
- Knowledge focus
- Tool adoption is a risk-and-fit decision, not brand familiarity
- Student action
- Students run three tools through the same blind trial and leave with a chosen tool chain plus evidence for rejecting the familiar option.
- Independent check
- With profiles closed, they justify the decisive row, the boundary the chosen tool still needs and the step kept tool-free.
- Human capabilities this is designed to exercise
- Metacognition
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Completed Tool-Risk Assessment
Build It, Hand It Over, Transfer It
- Knowledge focus
- Real work travels a designed chain — and a handover means a real use, not a sent link
- Student action
- Students design a real build chain—human-first, AI-supported, check, handover and stop—use it, revise from one real user and transfer it to a new brief.
- Independent check
- With every card closed, they redraw the chain for a fresh brief and explain each boundary, check, stop line and handover.
- Human capabilities this is designed to exercise
- Collaboration
- Adaptive thinking
- Work a teacher could inspect
- Workflow Chain Card
Spot Where an Agent Might Fail
- Knowledge focus
- An AI agent working on its own can fail at any step — so narrow its permissions and add human checkpoints
- Student action
- Students inspect a simulated agent trace, predict where it may fail, narrow its permission and add a human checkpoint.
- Independent check
- They explain one agent failure and one human checkpoint without AI.
- Human capabilities this is designed to exercise
- Problem-framing
- Adaptive thinking
- Work a teacher could inspect
- Agent Failure Map
Tool Risk and Privacy by Design
- Knowledge focus
- Assess a tool before adoption — and set the data boundary first
- Student action
- Students classify the data a task would use, compare two tools against their terms and draw a boundary around what may and may not enter.
- Independent check
- With profiles closed, they explain the data classes, the decisive failed row, the surviving boundaries and why no tool remains an option.
- Human capabilities this is designed to exercise
- Problem-framing
- Ethical judgment
- Adaptive thinking
- Work a teacher could inspect
- Tool-Risk Assessment and Data-Boundary Diagram
Plan a Safe Agent
- Knowledge focus
- A safe agent is bounded by permissions and checkpoints, not just fast
- Student action
- Students set an agent control boundary, test two workflow plans, add their own stop and handoff, then run a fresh failure trace.
- Independent check
- With the plans closed, they explain what they kept, what they rejected, their stop-and-handoff change and one remaining limit.
- Human capabilities this is designed to exercise
- Problem-framing
- Ethical judgment
- Adaptive thinking
- Work a teacher could inspect
- Agent Workflow Plan
Contract Test
- Knowledge focus
- A specification is trusted only after it survives tests built to break it
- Student action
- Students attack an agent contract with normal, edge, adversarial, transfer and permission tests, then repair the clause that gives way.
- Independent check
- They name the test that broke the contract, the escaped clause, their repair and the constraint that survived every case.
- Human capabilities this is designed to exercise
- Problem-framing
- Critical reasoning
- Adaptive thinking
- Work a teacher could inspect
- Agent Test Log
Review a Suggested Patch Safely
- Knowledge focus
- You own the design; a described patch is not a tested change
- Student action
- Students define a human-owned module boundary, compare two proposed patches with a system map and write a bounded architecture change.
- Independent check
- They explain the simulated change and name the files, code and tests a developer would still need to inspect and run.
- Human capabilities this is designed to exercise
- Critical reasoning
- Metacognition
- Adaptive thinking
- Work a teacher could inspect
- Architecture Note and Declarative Patch Review Card
Be Honest About AI Help
- Knowledge focus
- Honesty about AI help means showing and explaining how it helped, not just following a policy
- Student action
- Students judge a difficult AI-use case through permission, authorship, assessment purpose, disclosure and accountability.
- Independent check
- They decide a fresh case and write either a precise AI-help note or a reasoned refusal without AI.
- Human capabilities this is designed to exercise
- Ethical judgment
- Metacognition
- Work a teacher could inspect
- Honesty Case Card
Who Is Answerable When It Ships?
- Knowledge focus
- A shipped system needs a named answerable human, not just a working feature list
- Student action
- Students put a proposed system through an answerability review, change one design choice and write its redress, retirement and ownership rules.
- Independent check
- They name the people affected, the finding that changed the design, the accountable human and the route a harmed person could use.
- Human capabilities this is designed to exercise
- Ethical judgment
- Collaboration
- Self-regulation
- Work a teacher could inspect
- Responsible Deployment Note
The Drill
- Knowledge focus
- An incident plan is written before the pressure, not during it
- Student action
- Students rehearse a fictional provenance incident calmly, preserve safe evidence, avoid amplification and prepare a private escalation route.
- Independent check
- From memory, they state what can be established, what gets preserved, which moves are unsafe and which human receives the case.
- Human capabilities this is designed to exercise
- Self-regulation
- Ethical judgment
- Adaptive thinking
- Work a teacher could inspect
- Public Provenance and High-Stakes Escalation Plan
Notice When AI Is Doing Too Much
- Knowledge focus
- Over-reliance on AI is something you audit, not something you feel
- Student action
- Students map their reliance on AI, mark where decisions moved to the tool and choose one practical independence upgrade.
- Independent check
- They inspect a fresh set of steps and identify where AI might be doing too much, without asking AI to judge it.
- Human capabilities this is designed to exercise
- Metacognition
- Self-regulation
- Work a teacher could inspect
- AI Use Log
Claim Under Pressure
- Knowledge focus
- A claim earns trust from defined-failure checks, then one kept regression case
- Student action
- Students define what would disprove an important claim, check it through independent routes and record the model, source and version.
- Independent check
- They explain the claim, its failure condition, which routes were truly independent and what material change would trigger a recheck.
- Human capabilities this is designed to exercise
- Critical reasoning
- Adaptive thinking
- Ethical judgment
- Work a teacher could inspect
- Claim Under Pressure Audit
Explain Your Project Trail
- Knowledge focus
- How you made it lives in a kept decision trail, not a polished story
- Student action
- Students assemble a project trail from purpose, first work, contribution decisions, tests, material change, risk and limits.
- Independent check
- With the cards closed, they answer an unseen question about what changed, what the evidence shows and what still needs real-world checking.
- Human capabilities this is designed to exercise
- Critical reasoning
- Ethical judgment
- Adaptive thinking
- Collaboration
- Work a teacher could inspect
- Project Decision Trail
From lesson to practice
A skill becomes useful when it returns in real work.
- 01
Act
The student makes an interpretation, prediction, calculation or design decision before AI contributes.
- 02
Delegate
AI receives one declared job with approved inputs and a stopping point—or it stays off.
- 03
Verify
The student checks the contribution through subject evidence, a calculation, a source or reality.
- 04
Defend
The student explains the decisive work and its limits with AI closed.
Powerful · Fallible · CatchableEach admitted AI job is designed to matter, be open to error and be checkable by the student doing the work.
AI knowledge becomes useful when it informs a human decision without taking it over.