01. Understand what an AI assistant can and cannot do
Observable objective — Six-case decision table with reasons
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Classify six uses as allowed help, verify-first help and work substitution.
Guided demonstration and decisions
- Classify six uses as allowed help, verify-first help and work substitution.
- compare a plausible answer against the course definition.
- record three overtrust risks.
Worked solution guidance
Application to the fictional case: Classify six uses as allowed help, verify-first help and work substitution. The submitted work must show “Six-case decision table with reasons”. The topic-specific check is: record three overtrust risks.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
compare a plausible answer against the course definition. Transfer and next step: record three overtrust risks.
Acceptance criteria
Expected deliverable: Six-case decision table with reasons. record three overtrust risks.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
02. Set a measurable study objective
Observable objective — SMART study brief with pretest and post-test
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Turn “revise science” into three observable outcomes.
Guided demonstration and decisions
- Turn “revise science” into three observable outcomes.
- set time and a no-AI success criterion.
- create a five-minute check to repeat after studying.
Worked solution guidance
Application to the fictional case: Turn “revise science” into three observable outcomes. The submitted work must show “SMART study brief with pretest and post-test”. The topic-specific check is: create a five-minute check to repeat after studying.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
set time and a no-AI success criterion. Transfer and next step: create a five-minute check to repeat after studying.
Acceptance criteria
Expected deliverable: SMART study brief with pretest and post-test. create a five-minute check to repeat after studying.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
03. Build questions that test understanding
Observable objective — Three source-grounded questions
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Using the supplied course passage, write two factual and one why-question.
Guided demonstration and decisions
- Using the supplied course passage, write two factual and one why-question.
- request AI reformulation without answers.
- verify each question against the source.
Worked solution guidance
Application to the fictional case: Using the supplied course passage, write two factual and one why-question. The submitted work must show “Three source-grounded questions”. The topic-specific check is: verify each question against the source.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
request AI reformulation without answers. Transfer and next step: verify each question against the source.
Acceptance criteria
Expected deliverable: Three source-grounded questions. verify each question against the source.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
04. Use active recall and correction
Observable objective — Recall log with real before/after attempts
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Hide answers on five flashcards.
Guided demonstration and decisions
- Hide answers on five flashcards.
- recall them and mark correct, partial or wrong.
- request a targeted explanation and retry without help.
Worked solution guidance
In the fictional dataset, five revision cards span three topics. A “partial” card does not count as correct: the learner identifies the missing point then retries unaided. Evidence is the new answer they actually wrote, not an AI-predicted score.
Keep your original response and revised attempt. If no new attempt occurred, record “not tested”. A copied answer is not evidence of learning.
Independent task
recall them and mark correct, partial or wrong. Transfer and next step: request a targeted explanation and retry without help.
Acceptance criteria
Expected deliverable: Recall log with real before/after attempts. request a targeted explanation and retry without help.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
05. Build a realistic revision schedule
Observable objective — Justified calendar with catch-up plan
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
List four topics and deadlines from the supplied CSV.
Guided demonstration and decisions
- List four topics and deadlines from the supplied CSV.
- schedule two short sessions and a rest day.
- revise the plan after a cancelled slot.
Worked solution guidance
Application to the fictional case: List four topics and deadlines from the supplied CSV. The submitted work must show “Justified calendar with catch-up plan”. The topic-specific check is: revise the plan after a cancelled slot.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
schedule two short sessions and a rest day. Transfer and next step: revise the plan after a cancelled slot.
Acceptance criteria
Expected deliverable: Justified calendar with catch-up plan. revise the plan after a cancelled slot.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
06. Check explanations and sources
Observable objective — Claim-evidence-status-correction register
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Locate a claim absent from the supplied reading.
Guided demonstration and decisions
- Locate a claim absent from the supplied reading.
- ask the assistant to flag uncertainty.
- check the claim against the course or a reliable source and correct the notes.
Worked solution guidance
Application to the fictional case: Locate a claim absent from the supplied reading. The submitted work must show “Claim-evidence-status-correction register”. The topic-specific check is: check the claim against the course or a reliable source and correct the notes.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
ask the assistant to flag uncertainty. Transfer and next step: check the claim against the course or a reliable source and correct the notes.
Acceptance criteria
Expected deliverable: Claim-evidence-status-correction register. check the claim against the course or a reliable source and correct the notes.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
07. Solve problems without copying AI output
Observable objective — Two explained solutions with disclosed help
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Attempt a fictional percentage problem unaided.
Guided demonstration and decisions
- Attempt a fictional percentage problem unaided.
- request a hint rather than the answer.
- explain every calculation and solve a different transfer problem.
Worked solution guidance
Application to the fictional case: Attempt a fictional percentage problem unaided. The submitted work must show “Two explained solutions with disclosed help”. The topic-specific check is: explain every calculation and solve a different transfer problem.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
request a hint rather than the answer. Transfer and next step: explain every calculation and solve a different transfer problem.
Acceptance criteria
Expected deliverable: Two explained solutions with disclosed help. explain every calculation and solve a different transfer problem.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
08. Prepare a presentation with original input
Observable objective — Five-slide outline and AI-use disclosure
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Build a thesis, three arguments and a limitation from personal notes.
Guided demonstration and decisions
- Build a thesis, three arguments and a limitation from personal notes.
- ask AI to critique structure without writing the final talk.
- prepare five slides and cite material actually used.
Worked solution guidance
Application to the fictional case: Build a thesis, three arguments and a limitation from personal notes. The submitted work must show “Five-slide outline and AI-use disclosure”. The topic-specific check is: prepare five slides and cite material actually used.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
ask AI to critique structure without writing the final talk. Transfer and next step: prepare five slides and cite material actually used.
Acceptance criteria
Expected deliverable: Five-slide outline and AI-use disclosure. prepare five slides and cite material actually used.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
09. Spot bias, errors and invented references
Observable objective — Critical review with three warning signs
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Compare two answers with opposing assumptions.
Guided demonstration and decisions
- Compare two answers with opposing assumptions.
- look for missing evidence and loaded language.
- mark unverifiable citations as unusable.
Worked solution guidance
Application to the fictional case: Compare two answers with opposing assumptions. The submitted work must show “Critical review with three warning signs”. The topic-specific check is: mark unverifiable citations as unusable.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
look for missing evidence and loaded language. Transfer and next step: mark unverifiable citations as unusable.
Acceptance criteria
Expected deliverable: Critical review with three warning signs. mark unverifiable citations as unusable.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
10. Protect data and follow course policy
Observable objective — Anonymised prompt and local-policy check
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Remove names, grades and identifiers from a fictional prompt.
Guided demonstration and decisions
- Remove names, grades and identifiers from a fictional prompt.
- check local course rules before submission.
- write an honest permitted-use disclosure.
Worked solution guidance
Application to the fictional case: Remove names, grades and identifiers from a fictional prompt. The submitted work must show “Anonymised prompt and local-policy check”. The topic-specific check is: write an honest permitted-use disclosure.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
check local course rules before submission. Transfer and next step: write an honest permitted-use disclosure.
Acceptance criteria
Expected deliverable: Anonymised prompt and local-policy check. write an honest permitted-use disclosure.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
11. Consolidate learning with a self-test
Observable objective — Self-test summary and targeted revision plan
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Write six questions covering distinct skills.
Guided demonstration and decisions
- Write six questions covering distinct skills.
- answer without AI, categorize error causes.
- create a correction activity and retest after a break.
Worked solution guidance
Application to the fictional case: Write six questions covering distinct skills. The submitted work must show “Self-test summary and targeted revision plan”. The topic-specific check is: create a correction activity and retest after a break.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
answer without AI, categorize error causes. Transfer and next step: create a correction activity and retest after a break.
Acceptance criteria
Expected deliverable: Self-test summary and targeted revision plan. create a correction activity and retest after a break.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
12. Final project: responsible exam preparation
Observable objective — Reproducible study portfolio and self-review
Scenario
Fictional case: prepare a science exam and a presentation without outsourcing thinking to AI.
What matters
Combine goals, schedule, five cards, a presentation and evidence log.
Guided demonstration and decisions
- Combine goals, schedule, five cards, a presentation and evidence log.
- complete two sessions and retain actual attempts.
- disclose AI assistance and limitations precisely.
Worked solution guidance
Application to the fictional case: Combine goals, schedule, five cards, a presentation and evidence log. The submitted work must show “Reproducible study portfolio and self-review”. The topic-specific check is: disclose AI assistance and limitations precisely.
If that check was not performed, mark it “not tested” in your evidence log; do not fabricate results.
Independent task
complete two sessions and retain actual attempts. Transfer and next step: disclose AI assistance and limitations precisely.
Acceptance criteria
Expected deliverable: Reproducible study portfolio and self-review. disclose AI assistance and limitations precisely.
This lab uses fictional data. Verify any AI recommendation before use. Institutional rules take precedence.
