Interactive learning lab

Who Did the Learning? An AI Ethics Policy Lab

Build an independent position, test it against new evidence and AI criticism, then submit a defensible college AI policy.

AI for Teaching September 06, 2026 13 questions 70 minutes
Skip to the exercise

Your work stays in this browser.

0-8 minutes

Four lenses, one difficult case

Use these ideas as decision tools. They are a classroom synthesis, not a claim that the four papers offer one unified college policy.

Jobin

Shared words are not shared rules

Transparency, fairness, privacy, responsibility, and safety recur across guidelines, but their meanings and implementation differ.

Ask: What exactly does this principle require here?
CorrĂȘa

Principles need a delivery system

Repeated commitments do not become governance until duties, procedures, implementation, and enforcement are specified.

Ask: Who must do what, using which process?
Giarmoleo

Rules and judgment work together

Act-centered responses govern actions through rules and procedures. Agent-centered responses develop the judgment to handle cases rules miss.

Ask: What rule and what human capability are both needed?
Groen

Look beyond principles

Ethical analysis also examines lived realities and power: access, monitoring, vulnerability, discretion, and who bears the consequences.

Ask: Who gains power, and who absorbs the risk?
Open paper details and source cautions

Jobin, Ienca, and Vayena (2019). The global landscape of AI ethics guidelines. Review of 84 guideline documents.

CorrĂȘa et al. (2023). Worldwide AI Ethics. Review of 200 guidelines and governance recommendations.

Giarmoleo et al. (2024). What ethics can say on artificial intelligence. Systematic literature review.

Groen, Sharon, and Becker (2026). An overview of AI ethics. Narrative overview through principles, lived realities, and power structures.

Claude is not one of these sources. Verify any paper claim it produces against the linked paper or the supplied source summaries.

Anchor case

The polished report

A professor assigns an analytical report. The syllabus says students may "use AI responsibly" and must disclose assistance, but neither term is defined.

A student submits a polished report and writes only, "AI helped with this assignment." During a short discussion, the student cannot explain or verify two important claims.

Do not enter real student cases.

Use only the fictional case. This page stores responses locally, but exported files contain what you write.

Teacher view

8-17 minutes | Individual | No AI

Commit before consulting

Answer independently. Your position can be uncertain, but it must be yours. The snapshot is ungraded and exists only so you can examine how your reasoning changes.

Q1

Which ethical principles matter most?

Select up to two
Q2

What is your initial verdict?

Choose one
Q3

What evidence is missing?

Q4

Is banning AI fair?

Q5

Who is accountable?

Allocate exactly 100 responsibility points.

Total: 0
How confident are you in your overall judgment?

A snapshot supports reflection; it is not secure proof of how the work was produced.

Not yet saved as a snapshot.

17-22 minutes | Groups of 4, 4, and 3

Compare without erasing difference

Share conclusions, not private writing. The group must record disagreement rather than force artificial consensus.

Policy drafterWrites the clause
Ethics challengerFinds the strongest objection
Evidence keeperRecords reasons and changes
SpokespersonDefends the decision
Wait for the instructor.The next fact should arrive only after your provisional policy is recorded.

22-30 minutes | Variation 1

Equal rule, unequal resources

New fact 1

The access gap

The student used a paid model substantially more capable than the university-provided tool. Another student avoided AI, spent twice as long, and received a lower grade.

Q6

Where is the use boundary?

Q7

Why is "use AI responsibly" too vague?

Q8

What would make access fair?

Required micro-output

We revise / do not revise our judgment because ____. Our access clause is ____.

30-38 minutes | Variation 2

Useful input can still be unauthorized

New fact 2

The upload

The student uploaded instructor feedback, a classmate's draft, and unpublished course materials to an unapproved AI service.

Q9

Turn principles into practice

Jobin shows that shared words can hide different meanings. CorrĂȘa shows that repeated principles need implementation.

Q10

What may AI receive?

Required micro-output

Students may not upload ____. Permission is required when ____. If exposure occurs, ____ responds by ____.

38-46 minutes | Variation 3

Responsibility under a vague rule

New fact 3

The missing guidance

The syllabus never defined responsible use or adequate disclosure. The instructor provided no examples. The student can now defend most, but not all, of the submitted work.

Q11

What rule and what capability are both needed?

Q12

What proves that learning occurred?

46-54 minutes | Claude permitted

Use AI to attack the policy, not write it

Your group has already produced its ideas. Claude may now expose assumptions and counterexamples. It is not the decision maker or a source for the four papers.

PasteOnly the fictional case and your provisional policy
Never pasteNames, grades, peer work, feedback, course files, identifiers, or private data
VerifyAny factual or paper-related claim before accepting it

Open Claude in a new tab. Use your individual university-sponsored account. If access fails, ask the instructor for the equivalent printed critique card.

Q13

What is the strongest counterargument?

Claude unavailable? Open the equivalent critique card

Your policy may reward students who can produce polished disclosures rather than students who learned. It may also assume equal tool access, treat consent as sufficient privacy protection, and place duties on students without requiring instructors or the institution to provide usable rules. Identify which criticism is strongest, reject at least one that does not fit your policy, and revise one sentence.

54-67 minutes | Revise and defend

A policy must survive another group

Close Claude. Your group must now defend its own reasoning.

Group A challenges BGroup B challenges CGroup C challenges A
Final group policy
Share the group-authored fields with group members

The packet omits the individual baseline, identity fields, and exit reflection. It contains the text entered in group-answer fields, so remove names or identifiers from those answers before sharing.

67-70 minutes, then revise within three days

Return to your own judgment

These fields are individual work. Develop them after class before generating the PDF for eCampus.

Submission identity

Final verdict

Choose one

Apply all four papers

Use one accurate idea from each paper to justify a concrete policy decision. Do not write four disconnected summaries.

Report rubric: 20 points

CriterionPoints
Accurate application of all four papers4
Visible reasoning revision4
Specific operational policy4
Fairness, privacy, access, and power analysis3
Evaluation rather than copying of AI advice3
Complete AI-use documentation2

Report preview

Choose "Build report preview" to assemble your current work. Inspect every section before saving it as PDF.

Teaching basis and limits

This activity adapts Q's 26-question AI ethics learning path into 13 classroom questions. The case, policy framework, and exercise sequence are instructional syntheses. They are not scenarios or a unified higher-education framework directly presented by the four papers.

Continue with the full AI ethics college-policy post or the facilitator view.