Use this guide alongside your own attempts, evidence and reflections. The activities are original practice resources, not accredited assessments. Use your browser’s print command to save a PDF.
1. Purpose & agency
1. Curiosity with a destination
Before reading: Describe your current approach or make a prediction about why the idea matters.
Curiosity gives you a reason to start; an observable destination gives you a way to judge progress. Keep room for exploration, but connect it to something you want to explain, decide or make. If the question changes, update the goal rather than collecting unrelated outputs.
Common trap: A list of interesting facts becomes a substitute for doing something with them.
Practice: Write a question you care about, an artifact it could lead to and one check of whether that artifact works.
Decision check: Which goal connects curiosity to evidence?
Create a chart and justify its scale using the source data
Spend an hour browsing impressive charts
Ask AI to tell me everything about visualization
Useful distinction: Create a chart and justify its scale using the source data.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
2. Learner agency
Before reading: Describe your current approach or make a prediction about why the idea matters.
Agency means having a meaningful role in goals, decisions and evaluation. It does not mean working without help. Choose what to delegate, inspect its effect and retain the decisions needed for your own future performance. Ask whether the support expands what you can do or merely hides what you cannot yet do.
Common trap: Choice is cosmetic when every learner must reproduce the same reasoning and answer.
Practice: List three decisions in a task. Keep one yourself, ask for support on one and explain why one may be delegated.
Decision check: Which activity preserves meaningful agency?
Choose a font for an otherwise fully prescribed answer
Accept every recommendation because the AI is confident
Choose an approach and justify it against shared criteria
Useful distinction: Choose an approach and justify it against shared criteria.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
3. Outcomes you can observe
Before reading: Describe your current approach or make a prediction about why the idea matters.
An outcome describes a performance, its conditions and a standard for judging it. Broad verbs such as understand need an observable companion. Avoid reducing a rich skill to a trivial check merely because it is easy to count. Define the smallest performance that still represents the real skill.
Common trap: A measurable target counts activity while missing the intended capability.
Practice: Rewrite one broad goal as a performance with conditions and a check. Explain what the check would miss.
Decision check: Which outcome is aligned with source evaluation?
Open ten articles in one session
Compare two sources and defend which supports the claim
Name five websites from memory
Useful distinction: Compare two sources and defend which supports the claim.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
4. Roles for people and AI
Before reading: Describe your current approach or make a prediction about why the idea matters.
Assign roles before opening a tool. AI may suggest examples, ask questions or critique a draft, while a person verifies facts and makes consequential decisions. The right boundary depends on the goal. If writing is the skill, having AI write the entire artifact can remove the practice you needed.
Common trap: Using AI during preparation is confused with requiring learners to use AI during the activity.
Practice: Create a role agreement: learner decisions, teacher decisions, optional AI support and actions that need human review.
Decision check: Which role assignment protects a writing goal?
AI asks questions; the learner drafts and defends the paragraph
AI writes the paragraph while the learner only submits it
The teacher judges quality solely by polish
Useful distinction: AI asks questions; the learner drafts and defends the paragraph.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
2. Diagnose & support
5. Find the starting point
Before reading: Describe your current approach or make a prediction about why the idea matters.
A small baseline task reveals current performance under stated conditions. Preserve the attempt and assistance used. One failure does not diagnose a whole person: look at the particular step, interpretation or missing prerequisite. Use the result to choose support, not to attach a permanent ability label.
Common trap: A baseline becomes a high-pressure grade rather than useful planning information.
Practice: Design a three-minute baseline and identify two different errors it could reveal.
Decision check: What should a baseline guide?
A permanent label of learner ability
A claim that one mistake proves no prior knowledge
The next support or practice task
Useful distinction: The next support or practice task.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
6. Surface prior knowledge
Before reading: Describe your current approach or make a prediction about why the idea matters.
Existing knowledge can help or interfere with a new idea. Invite a prediction, explanation or example to reveal the learner’s model. Do not assume familiar vocabulary means shared understanding. Compare the prediction with evidence, then help the learner revise the model explicitly.
Common trap: The teacher corrects the answer without finding the idea that produced it.
Practice: Write a prediction question, two plausible answers and a follow-up that distinguishes the reasoning behind them.
Decision check: What reveals a learner’s model?
Give the answer before collecting a prediction
Ask why they expect an outcome, then test it
Ask only whether the topic sounds familiar
Useful distinction: Ask why they expect an outcome, then test it.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
7. Manage the thinking load
Before reading: Describe your current approach or make a prediction about why the idea matters.
A task can demand content knowledge, tool navigation and formatting at once. Remove demands that do not serve the current goal while preserving the important reasoning. Break the task into meaningful parts and reunite them later. Simpler presentation should not become a permanently simpler learning target.
Common trap: Simplification removes the reasoning that the activity was intended to teach.
Practice: Identify three demands in an activity. Remove one irrelevant demand and explain how the core challenge remains.
Decision check: Which change reduces unnecessary load?
Provide a clean dataset while retaining the interpretation task
Supply the interpretation so nobody has to reason
Add simultaneous instructions in several places
Useful distinction: Provide a clean dataset while retaining the interpretation task.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
8. Productive difficulty
Before reading: Describe your current approach or make a prediction about why the idea matters.
Difficulty is useful when the learner can make progress or learn from feedback. Repeated guessing without a foothold is not automatically productive. Watch for evidence such as an articulated plan, a partial solution or a specific question. Offer the smallest useful support and see whether it enables another attempt.
Common trap: All struggle is treated as valuable, even when it produces no new information.
Practice: Write a support rule based on observable behavior, including when to intervene and what to offer.
Decision check: When should support change?
Only when a fixed timer expires for everyone
Never, because all difficulty produces learning
When evidence shows the current task is not enabling progress
Useful distinction: When evidence shows the current task is not enabling progress.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
3. Explain & scaffold
9. Worked examples with purpose
Before reading: Describe your current approach or make a prediction about why the idea matters.
A worked example is useful when the learner sees why decisions were made. Direct attention to a few important choices rather than requiring passive copying. Pair the example with a related attempt and inspect whether the learner can choose the method. More examples are not always the answer to every gap.
Common trap: A copied solution looks complete while its decisions remain unexplained.
Practice: Annotate three decisions in an example and create a similar task that changes one important feature.
Decision check: What makes an example active?
Save it for later without attempting anything
Predict a step, explain it and try a variation
Copy every line without discussing its purpose
Useful distinction: Predict a step, explain it and try a variation.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
10. Fade support deliberately
Before reading: Describe your current approach or make a prediction about why the idea matters.
Support can move from a full example to missing steps, hints and independent work. Adjust using performance rather than a rigid schedule. Removing help is not the goal by itself; the goal is control of the underlying skill. Restore a small support when it reveals or repairs a specific gap.
Common trap: Success with a scaffold is reported as if it were unaided performance.
Practice: Design three versions of one task: modeled, partially supported and independent. Keep the core goal consistent.
Decision check: What should be recorded alongside a result?
The support available and actually used
Only whether the final answer is correct
Only the time spent on the page
Useful distinction: The support available and actually used.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
11. Analogies and boundaries
Before reading: Describe your current approach or make a prediction about why the idea matters.
An analogy maps a familiar relationship onto a new one. State what corresponds and where the comparison fails. An analogy can guide an explanation but cannot prove a claim. After using it, return to the actual system and check whether the learner can explain it without relying only on the metaphor.
Common trap: A memorable metaphor replaces an accurate model.
Practice: Create an analogy, map two relationships and explain one boundary using a concrete counterexample.
Decision check: What completes a useful analogy?
It is presented as literally identical
It is repeated until the learner agrees
Its correspondences and limits are explained
Useful distinction: Its correspondences and limits are explained.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
12. Questions that reveal thinking
Before reading: Describe your current approach or make a prediction about why the idea matters.
Different questions serve different purposes: prediction, explanation, comparison, diagnosis or transfer. Ask one focused question and give time for an attempt. A chain of questions can become a disguised lecture if the learner only guesses what the teacher wants. Use the response to decide what to ask next.
Common trap: Questions reward guessing the teacher’s intended phrase instead of reasoning.
Practice: Write a question, a likely partial answer and a follow-up that investigates the gap.
Decision check: Which follow-up investigates thinking?
Would you agree if I told you my answer?
What evidence led you to that prediction?
Can you guess again until you say the right word?
Useful distinction: What evidence led you to that prediction?.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
4. Practise & remember
13. Retrieval with correction
Before reading: Describe your current approach or make a prediction about why the idea matters.
Try to produce an explanation, procedure or answer before reviewing it. Then compare, correct and attempt again later. Recognition while reading and independent production are different performances. Keep retrieval appropriate to the goal: remembering a definition alone may not establish that you can apply it.
Common trap: Recall becomes a score without feedback or a route to repair.
Practice: Create a recall task and a correction step, plus one application that goes beyond the recalled wording.
Decision check: What makes a retrieval activity useful for practice?
An attempt followed by comparison and correction
Seeing the answer before every attempt
Counting only how quickly the answer appears
Useful distinction: An attempt followed by comparison and correction.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
14. Spacing and return visits
Before reading: Describe your current approach or make a prediction about why the idea matters.
Plan another attempt after time has passed. Use the result to decide what to revisit; there is no single perfect interval for every learner and topic. A calendar entry is only a plan. The useful event is the later attempt, feedback and decision about the next practice.
Common trap: A schedule is treated as evidence that learning actually occurred.
Practice: Create two return visits with a specific task and a rule for adapting the plan after each.
Decision check: What should determine the next review task?
A fixed streak count alone
How attractive the calendar looks
What the learner could retrieve and apply at the last check
Useful distinction: What the learner could retrieve and apply at the last check.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
15. Mix and distinguish
Before reading: Describe your current approach or make a prediction about why the idea matters.
Mixed practice can require the learner to decide which method applies. Choose tasks that are meaningfully related and discuss the distinctions. Do not confuse variety with random difficulty. Learners may need initial support with each type before a mixed set becomes useful.
Common trap: Every task changes at once, so the learner cannot see what distinction matters.
Practice: Build three short cases that require different approaches and state the cue that distinguishes each.
Decision check: What is the key demand in mixed practice?
Repeating the same method without choosing
Selecting an approach from meaningful task cues
Doing unrelated tasks as quickly as possible
Useful distinction: Selecting an approach from meaningful task cues.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
16. Deliberate practice loops
Before reading: Describe your current approach or make a prediction about why the idea matters.
Choose a specific weakness, attempt a targeted task, get relevant feedback and retest. Keep the target narrow enough to inspect but connected to the larger skill. More repetitions of the same error may not help. Make the next attempt respond to evidence from the previous one.
Common trap: Practice time is counted while the same unexamined error repeats.
Practice: Describe a gap, a focused exercise, feedback and a changed retest. Explain how it supports the whole skill.
Decision check: What should change after feedback?
The next attempt should address the identified gap
Nothing, because repetition alone is sufficient
Only the visual formatting of the answer
Useful distinction: The next attempt should address the identified gap.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
5. Feedback & evidence
17. Feedback that leads somewhere
Before reading: Describe your current approach or make a prediction about why the idea matters.
Useful feedback connects evidence to a criterion and a next action. Limit the number of targets so the learner can respond. The value of feedback is not its length or friendliness alone; look for a revision and a later check. A learner may disagree with feedback if they can justify the decision.
Common trap: A long critique leaves no time or clear opportunity to act.
Practice: Write feedback with an observation, criterion and next action. Design a check of the revision.
Decision check: What completes a feedback loop?
The teacher sending detailed comments
The learner thanking the reviewer
A revision and a check of what changed
Useful distinction: A revision and a check of what changed.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
18. Rubrics as shared language
Before reading: Describe your current approach or make a prediction about why the idea matters.
A rubric makes quality criteria visible. Describe observable differences rather than vague labels such as excellent. Use examples to interpret the criteria and leave room for valid approaches. A rubric is a judgment aid, not a guarantee that every important aspect has been captured.
Common trap: Points create an appearance of precision without a clear meaning.
Practice: Write three criteria and describe what weak, developing and strong evidence would look like for one.
Decision check: Which rubric criterion is observable?
The answer sounds like an expert
The claim is connected to evidence that supports its scope
The work feels sophisticated
Useful distinction: The claim is connected to evidence that supports its scope.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
19. Assess the intended capability
Before reading: Describe your current approach or make a prediction about why the idea matters.
Choose evidence that matches the goal. A quiz, artifact, explanation and live demonstration reveal different things. Triangulate when a single task could be misleading. Do not infer independent performance from a polished artifact whose production conditions are unknown.
Common trap: An attractive product is treated as proof of all the skills involved in making it.
Practice: Match a goal to two evidence sources and state what each can and cannot establish.
Decision check: What supports an independence claim?
A relevant task performed under stated assistance conditions
A polished output with no process record
A self-rating alone
Useful distinction: A relevant task performed under stated assistance conditions.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
20. Confidence and calibration
Before reading: Describe your current approach or make a prediction about why the idea matters.
Confidence is a report of how sure you feel. Compare it with performance and checks to understand where certainty is justified or misleading. Avoid punishing honest uncertainty. The aim is to make better decisions about checking and seeking support, not to maximize a confidence number.
Common trap: Confidence is used as a substitute for evidence.
Practice: Describe a confident mistake and one check that would have caught it. Plan when to use that check next.
Decision check: What makes confidence more useful?
Rewarding the highest rating regardless of performance
Treating uncertainty as failure
Comparing it with the result of a meaningful check
Useful distinction: Comparing it with the result of a meaningful check.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
6. Include & collaborate
21. Access without lowering the goal
Before reading: Describe your current approach or make a prediction about why the idea matters.
Identify barriers in the format, environment or tools. Offer routes that preserve the learning target. If the goal is reasoning, typed, spoken or diagram-based explanations may all be valid when they reveal that reasoning. Keep criteria explicit so different routes remain comparable.
Common trap: Accommodation removes the core learning goal instead of removing an irrelevant obstacle.
Practice: Identify a barrier, offer an alternative route and explain which evidence remains unchanged.
Decision check: Which adaptation preserves a reasoning goal?
Require one format even when it is not the target skill
Allow a spoken explanation assessed against the same reasoning criteria
Supply the reasoning and assess only copying
Useful distinction: Allow a spoken explanation assessed against the same reasoning criteria.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
22. Differentiate with evidence
Before reading: Describe your current approach or make a prediction about why the idea matters.
Adapt support, complexity or context in response to current evidence. Avoid permanent groups based on a single performance. Maintain a shared purpose where possible and recheck who needs what. A simpler input can prepare someone for a richer task without becoming their ceiling.
Common trap: Learners are assigned fixed difficulty levels that never change.
Practice: Design two supports for the same outcome and describe the evidence that would trigger or remove each.
Decision check: What should drive differentiation?
Current task evidence and a plan to recheck
A permanent label from an early mistake
The assumption that one format suits everyone
Useful distinction: Current task evidence and a plan to recheck.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
23. Collaboration with individual thought
Before reading: Describe your current approach or make a prediction about why the idea matters.
Collaboration is stronger when people contribute ideas, compare reasoning and share responsibility. Give time for an individual attempt before discussion when useful. Group success does not establish what each member can do. Include an individual explanation or transfer check afterward.
Common trap: One person or AI does the thinking while everyone receives the same completion mark.
Practice: Plan a group task with an individual starting point, a comparison step and an independent final check.
Decision check: What checks individual learning after group work?
The group submitting one polished artifact
A count of how often each learner spoke
A new individual task with stated support conditions
Useful distinction: A new individual task with stated support conditions.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
24. Motivation through meaningful progress
Before reading: Describe your current approach or make a prediction about why the idea matters.
Connect practice to a purpose and make small improvements visible. Offer choices that matter and feedback that explains the next step. Rewards can mark participation, but they should not displace the real goal. Protect the option to admit confusion and ask for help.
Common trap: Badges encourage clicking through while the learning task is ignored.
Practice: Design a progress signal based on evidence and explain how it avoids rewarding empty activity.
Decision check: Which progress signal is most informative?
A streak that ignores what happened during practice
A before-and-after example linked to a criterion
The number of tabs opened
Useful distinction: A before-and-after example linked to a criterion.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
7. AI literacy & trust
25. Verify generated material
Before reading: Describe your current approach or make a prediction about why the idea matters.
Generated text can be fluent and still be wrong. Check important claims, calculations, examples and answer keys before relying on them. Separate what you checked from what you merely intend to check. Teacher-created and AI-assisted materials both benefit from review.
Common trap: A citation or confident explanation is accepted as verification.
Practice: Audit a small fictional lesson resource: list a claim, its check, the result and any unresolved uncertainty.
Decision check: What counts as a completed check?
Inspecting relevant evidence and recording what it supports
Asking the same model whether it is sure
Seeing a source-like title in the response
Useful distinction: Inspecting relevant evidence and recording what it supports.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
26. Prompt for a learning purpose
Before reading: Describe your current approach or make a prediction about why the idea matters.
A prompt should specify the task, context, learner needs and the AI’s limited role. Include the form of help and what the learner should attempt first. Prompt quality is judged by the learning activity it enables, not by length or impressive role descriptions.
Common trap: A long prompt produces a complete answer that removes the intended practice.
Practice: Write a coaching prompt that includes a task, assistance boundary and a way to check the result.
Decision check: Which prompt supports learning?
Complete the whole assignment before I begin
Use expert language so the output looks advanced
Ask one targeted question and wait for my attempt
Useful distinction: Ask one targeted question and wait for my attempt.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
27. Privacy, fairness and boundaries
Before reading: Describe your current approach or make a prediction about why the idea matters.
Use only data you are permitted to share and minimize what the task requires. Consider who could be identified or excluded. Keep important decisions accountable to people. The exact requirements depend on your setting and current policies, so inspect them before using real learner records.
Common trap: Removing a name is assumed to remove all identification risk.
Practice: Write a data and access plan specifying what is shared, what is omitted and a non-AI alternative.
Decision check: What is a sound starting point for classroom experimentation?
Require an unapproved service for every learner
Use fictional material and inspect applicable rules before real data
Upload full records because names can be removed later
Useful distinction: Use fictional material and inspect applicable rules before real data.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
28. Make assistance visible
Before reading: Describe your current approach or make a prediction about why the idea matters.
Document how AI and people contributed to the work. This supports interpretation of evidence and makes future practice easier to plan. Disclosure should be specific enough to understand the process without exposing private information. It does not automatically establish quality or resolve every authorship question.
Common trap: Disclosure is treated as either proof of mastery or proof that no learning occurred.
Practice: Write a contribution statement and propose an independent task that tests the capability the artifact suggests.
Decision check: What does an assistance record help establish?
The conditions under which the work was produced
That every claim in the work is accurate
That the learner cannot perform independently
Useful distinction: The conditions under which the work was produced.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
8. Transfer & improvement
29. Transfer to a changed situation
Before reading: Describe your current approach or make a prediction about why the idea matters.
Transfer asks you to choose how a principle applies when the surface details or constraints change. State what remains relevant and what needs adaptation. A new task should test the intended idea, not depend mainly on unrelated knowledge. Inspect both the result and the choice of approach.
Common trap: Repeating the same example is labeled as transfer.
Practice: Create a transfer task, name the stable principle and identify a new demand that might need support.
Decision check: What is a meaningful transfer demand?
Repeat the original steps with identical inputs
Remember the location of the answer on the page
Select and adapt a principle in a changed context
Useful distinction: Select and adapt a principle in a changed context.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
30. Projects with authentic evidence
Before reading: Describe your current approach or make a prediction about why the idea matters.
A project connects several skills to a useful purpose. Define audience, deliverable and acceptance criteria before production. Save intermediate work so a final artifact can be interpreted. Keep the project small enough to check, then expand based on evidence rather than ambition alone.
Common trap: A large impressive project has no clear way to judge whether it works.
Practice: Write a project brief with three criteria, two milestones and an independent demonstration.
Decision check: What makes a project assessable?
Only a final screenshot
A bounded deliverable and explicit evidence criteria
The largest possible scope
Useful distinction: A bounded deliverable and explicit evidence criteria.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
31. Reflect and adapt
Before reading: Describe your current approach or make a prediction about why the idea matters.
Reflection connects experience to the next decision. Name what happened, what evidence supports your interpretation and what you will change. Distinguish a plausible explanation from a proven cause. A short specific reflection is more useful than a long account with no next action.
Common trap: Reflection becomes a positive story that excludes inconvenient results.
Practice: Write an evidence-based reflection with one alternative explanation and one next experiment.
Decision check: What should reflection produce?
A justified next action linked to evidence
Only a statement that the session was enjoyable
A guarantee about why every change occurred
Useful distinction: A justified next action linked to evidence.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
32. Improve a learning system
Before reading: Describe your current approach or make a prediction about why the idea matters.
Look across goals, activities, support and evidence rather than blaming one tool or learner. Test a small change, watch for tradeoffs and keep what works under the relevant conditions. Share the limits of your findings. A useful local improvement is not automatically a universal teaching rule.
Common trap: One successful session is generalized to every learner and subject.
Practice: Design a small improvement experiment with a question, comparison, evidence and a decision rule.
Decision check: What is a responsible conclusion from a small pilot?
It proves a universal best method
It makes all future evaluation unnecessary
It worked under these conditions and needs further checking elsewhere
Useful distinction: It worked under these conditions and needs further checking elsewhere.
Transfer: Change the audience, context or constraints. Apply the principle and inspect the result. Record assistance and uncertainty.
Reflect: What changed in your reasoning? What will you check next?
Decision-replay labs
The polished answer nobody can explain
A group submits an excellent AI-assisted report. When asked why a source supports the main claim, most members cannot answer.
Award mastery for the polished report The artifact is recognized, but the evidence still does not establish individual reasoning. The next task must reveal what each person can explain.
Ask for individual source explanations and a fresh claim check You now see different gaps: one learner understands scope, another only recognizes the citation. You can target support without discarding the group’s work.
Ban all tools and restart the entire project The restart changes the conditions, but it does not yet diagnose the specific source-evaluation gap. Some useful work is lost without showing what to practise.
The goal is evaluating evidence. What follow-up gives the most relevant new information?
A new source-and-claim pair checked individually This directly tests the target judgment in a changed case. Record assistance and use errors to choose support.
A longer group report on the same claim More polished group output still leaves individual understanding uncertain.
A quiz about the names of AI tools Tool-name recall does not test whether a source supports a claim.
Reflect: What part of the report can you still value, and what evidence is missing for an independence claim?
The learner who is stuck
A learner has tried the same spreadsheet formula three times and repeats the same error. They cannot explain what a cell reference means.
Wait longer without changing the task More time produces another repetition. You still need information about the missing prerequisite.
Ask about the reference, model one decision, then return a step The learner identifies the input cell and completes the next reference with a small hint. The support has exposed a specific gap.
Give the completed spreadsheet The sheet works, but you have little evidence that the learner can choose a reference next time.
After a supported success, what should happen next?
Declare the whole skill mastered Supported success does not yet establish independent use.
Remove every support and assign an unrelated hard project The jump adds unrelated demands and may conceal the original gap.
Try a changed small example and record any hint used This checks whether the repaired distinction survives a new case while keeping assistance visible.
Reflect: What observable signal tells you to offer support, and how will you know when to fade it?
The classroom with unequal AI access
Your activity asks learners to critique an AI-generated explanation. Some learners cannot use the chosen service.
Require everyone to create an account immediately Access becomes a barrier and may conflict with local requirements. The critique goal does not inherently require individual accounts.
Give those learners an unrelated worksheet They participate, but the evidence no longer targets the same capability.
Provide a verified sample output for everyone to critique All learners can practise the same reasoning. Those with approved access may generate an additional comparison, but it is not required.
How should you compare the resulting work?
Use shared criteria for identifying claims, evidence and uncertainty Equivalent reasoning criteria preserve the goal across access routes.
Award extra mastery points for using the paid tool Tool access is not evidence of stronger critique.
Grade only how long each response is Length does not establish quality of reasoning.
Reflect: Which requirement was an access barrier rather than part of the learning goal?
The quiz that teaches the wrong idea
An AI-generated quiz has a confident answer key, but two options seem defensible under different interpretations.
Publish it because an answer key exists Learners may be penalized for a reasonable interpretation. A generated key is not validation.
Clarify the scenario and test the revised item You define the missing condition, inspect each option and ask a colleague to explain the intended distinction.
Delete every challenging question Removing all challenge avoids this item but does not improve your process for writing meaningful checks.
What should a useful revision make explicit?
That the teacher’s choice is always correct Authority alone does not resolve an ambiguous item.
The task conditions and reasoning that distinguish the options Clear conditions allow the item to test the intended idea rather than guessing.
More decorative details in the scenario Extra detail does not help unless it resolves the relevant ambiguity.
Reflect: How will you audit generated examples and answer keys before learners use them?
The explanation everyone recognizes
Learners say a source-checking demonstration makes sense. A week later, they cannot reconstruct the process.
Repeat the same presentation more enthusiastically It may feel familiar again, but you still need an independent attempt to locate the gap.
Give a closed-note attempt, then compare and correct The attempts reveal that learners remember opening a source but omit checking whether it supports the exact scope.
Count their positive reactions as sufficient evidence Enjoyment and agreement do not establish recall or application.
How would you plan the next visit?
Read the identical passage without attempting anything This does not show whether learners can produce the missing step.
Use a changed claim, collect an attempt, then check the source match The new case makes the scope distinction visible and creates a useful reason to revisit it.
Ask only whether learners feel more confident Confidence can accompany a check but cannot replace it.
Reflect: What is the difference between recognizing a process and using it independently?
The feedback nobody uses
You write detailed comments on every draft, but learners move to the next assignment without revising.
Write even more comments The volume increases while the missing revision opportunity remains.
Replace all comments with a grade A grade may summarize a judgment, but it does not create a clear repair action.
Choose one priority, allocate revision time and recheck The learner changes the missing evidence link and tests the explanation with a new reader.
What should you inspect after the revision?
Whether the targeted criterion improved in the result This connects feedback to an observable change.
Whether every sentence was changed Change volume is not the goal.
Whether the learner repeated your wording exactly A different valid revision may meet the criterion.
Reflect: What will you remove from the schedule to make room for acting on feedback?
The busy project with no clear evidence
A class spends two weeks producing a visually impressive website. The goal was explaining scientific claims, but most time went into decoration.
Judge the project only by visual polish The evaluation rewards a different skill from the stated goal.
Refocus on a small set of claims and an evidence review The artifact remains useful, and the next milestone makes the intended reasoning visible.
Add more website pages The scope grows without fixing the mismatch between activity and goal.
Which milestone best restores alignment?
A new animation for each page Animation may help communication but does not by itself demonstrate evidence reasoning.
A count of hours spent editing Time is an input rather than a direct measure of the capability.
An explanation of two claims, their sources and a limitation This creates inspectable evidence aligned with the goal.
Reflect: Which attractive activity could you reduce so learners have time for the central capability?
The pilot that seems too successful
After one AI-supported session, learners score higher on a repeated exercise. You want to call the method a universal improvement.
Announce that the method works for everyone The repeated task and local conditions do not justify that broad claim.
Record the improvement and test a fresh comparable task You preserve the observation while investigating familiarity, assistance and transfer.
Ignore the positive result entirely The result may be useful even though its interpretation needs limits.
What belongs in the pilot report?
Only the highest score This hides variation and the conditions of the result.
The conditions, assistance, evidence, alternative explanations and next test Readers can assess what the result supports and what remains uncertain.
A promise that future learners will improve equally The small pilot does not establish that guarantee.
Reflect: What result would make you revise your interpretation of the pilot?
Project briefs
Redesign one learning session
Choose a 30–60 minute session you already know. Preserve its core goal, add a baseline attempt, introduce one targeted support and finish with a changed independent task.
Deliverable: A lesson plan, two learner task examples and a short review of alignment.
Outcome and final task assess the same capability
Support responds to a plausible baseline gap
The plan includes time for feedback and another attempt
Build a personal learning experiment
Choose one skill and compare two realistic practice approaches on comparable small tasks. Record starting performance, time, help and results. Treat the findings as local evidence.
Deliverable: An experiment log, task examples and a decision about your next practice.
The comparison uses stated common criteria
Assistance and task differences are recorded
The conclusion includes an alternative explanation
Create an inclusive practice pack
Design a practice activity that works with and without learner AI access. Offer two response formats that reveal the same core reasoning, with clear instructions and feedback.
Deliverable: Two versions of a task, shared criteria and a teacher facilitation note.
Both versions preserve the goal
The access change removes an irrelevant barrier
Each version includes a meaningful result check
Make an evidence-first AI resource
Create a short AI-assisted teaching resource using fictional or public material. Audit its claims, examples and answer key. Mark what was checked and what remains uncertain.
Deliverable: The resource, an audit log and an assistance statement.
Important claims have documented checks
The answer key has been reviewed for ambiguity
Unverified material and assistance are visible
Teach a concept to someone new
Prepare a five-minute explanation with a worked example and a boundary. Ask a learner to attempt a new case. Use their response to revise the explanation.
Deliverable: An explanation, a transfer task and before/after teaching notes.
The explanation shows a mechanism
The learner attempts a changed case
The revision responds to observed evidence
Build a complete Vibe Learning portfolio
Combine a meaningful goal, initial attempt, focused practice, feedback, verification and independent application. Explain what you can now do and which claim about your learning remains untested.
Deliverable: A checked artifact and a concise evidence narrative with next steps.
The portfolio includes actual work, not only plans
Supported and independent performance are distinguished
The next step follows from a specific remaining gap
Design a learning session
Specify the audience, observable outcome, baseline, support, access routes, evidence and a changed independent task. Allocate time to purpose, first attempt, example, focused practice, feedback and repair, independent transfer and reflection. A complete form does not guarantee an effective lesson: test the plan and use learner evidence to revise it.