Responsible AI Policy Memo
Topic 3 written homework: principles, implementation, fairness, privacy, and assessment.
Deliverable
One integrated policy memo
Length: 800–900 words, excluding the AI-use disclosure and references.
Structure: Use headings A–D from this sheet. Submit as a PDF or in the format your instructor specifies.
Submission: Upload through the course LMS as Lastname-Topic03-PolicyMemo.pdf. The syllabus and LMS govern late work and resubmission.
Use at least two assigned reviews
Use brief parenthetical references. Example: (Jobin et al., 2019).
No outside sources are required. Do not invent page numbers, quotations, findings, or citations.
Assigned Reading Path
Read all four reviews at the assigned depth below and complete all four rows of the reading matrix. Cite at least two reviews in the memo. Then deepen either Corrêa et al. or Groen et al. through the default no-purchase route: read the full targeted discussion or conclusion, and record one finding, one limitation, and one claim the paper does not establish. Do not purchase an article; if a route fails, notify the instructor and use the posted library route, excerpt, or equivalent.
Jobin, Ienca, and Vayena (2019)
The global landscape of AI ethics guidelines
Read: abstract; full results/discussion only if library access is available. Focus on recurring principles and divergence in interpretation or implementation.
Corrêa et al. (2023)
Worldwide AI ethics: A review of 200 guidelines and recommendations for AI governance
Read: abstract, discussion of earlier guideline reviews, and closing limitations/conclusions. Focus on the broader principle map and the distance between principles and governance mechanisms.
Giarmoleo et al. (2024)
What ethics can say on artificial intelligence: Insights from a systematic literature review
Read: abstract and Section 4, “Discussion on the findings and conclusions.” Focus on ethical concerns, proposed responses, and act- versus agent-centered approaches.
Groen, Sharon, and Becker (2026)
An overview of AI ethics: moral concerns through the lens of principles, lived realities and power structures
Read: abstract, Table 1, and Section 6 conclusion. Focus on what each lens reveals and overlooks.
Reading Matrix
Use short phrases. The last column matters: a review can provide a framework without proving every prediction or outcome discussed in class.
Example of the level of specificity expected
Do not submit this example as your own Jobin row. Read the assigned material and complete every row in the blank matrix below in your own words.
| Reading | Main question | One framework or finding | One limitation | One claim it does not establish |
|---|---|---|---|---|
| Jobin et al. — model | Are shared global AI-ethics principles emerging? | Five principles recur, but their meanings and implementation differ. | Maps guideline documents rather than testing their real-world effects. | Does not establish whether an AI detector is accurate. |
| Reading | Main question | One framework or finding | One limitation | One claim it does not establish |
|---|---|---|---|---|
| Jobin et al. | ||||
| Corrêa et al. | ||||
| Giarmoleo et al. | ||||
| Groen et al. |
In-Class Baseline: Meeting 2, Case 3 (Student A and Student B)
Policy-test boundary: The permission belongs to the assignment in Case 3, not to this memo. This Topic 3 memo does not permit AI-generated or AI-rewritten submitted prose.
After Case 3 in Meeting 2, answer the prompt below without AI and keep it privately as your baseline for section D. It is not a separate submission and the instructor will not collect the original note. You may be asked to explain what changed between your baseline and final memo. If you are absent or need an accommodation, complete the same prompt independently before using AI, or use an instructor-approved equivalent.
State your judgment, one condition that could change it, and what the grade should represent.
Memo Prompt
A. Diagnose the invented-citations case (Meeting 1, Angle 3) 150–175 words
Separate verified case facts, reading-supported claims, assumptions, and inferences. Explain how transparency, fairness, privacy, accountability, and safety appear in the case. Verified case facts are only facts stated in the assigned scenario; reading-supported claims are grounded in assigned readings; label your own reasonable extensions as inferences.
B. Challenge the ban-and-detector response 200–225 words
Give the strongest ethical defense of a university ban and the strongest criticism of it. Apply at least two assigned reviews. Do not claim that these reviews establish detector accuracy, disparate impacts, classroom mistrust, concealment, or employment outcomes unless you have separately approved evidence.
C. Write an operational policy 175–200 words
Write one policy paragraph that specifies permitted AI use, required disclosure, prohibited use, privacy and upload limits, how instructors verify learning, and the default when assignment instructions are silent. Avoid phrases such as “use AI responsibly” unless you define the required behavior.
D. Test your policy against Case 3 (Student A and Student B) 275–300 words
Explain why allowing AI can be fair and why it may still be unfair. Address equal rules, equitable access or capability, and assessment validity. State what the grade should measure, reach a final judgment, and identify one condition that would change your conclusion. If rewriting was explicitly permitted, explain where responsibility for any mismatch belongs.
AI Use for This Homework
The purpose of this assignment is to assess your ethical reasoning. The course therefore permits AI support for learning and feedback while reserving your claims, selected argument structure, submitted prose, and judgment for you.
Allowed with disclosure
- Clarify a concept from an assigned reading.
- Brainstorm alternative arguments or counterarguments.
- Compare possible ways to organize claims you already developed; select and substantially revise the outline yourself.
- Ask for feedback that identifies a problem, asks a question, or suggests a revision strategy.
Not allowed
- Generate or rewrite sentences or paragraphs for submission.
- Replace your analysis, judgment, or reading of the sources.
- Invent, repair, or confirm a citation without checking the source.
- Upload peer work, private course materials, personal data, or confidential information.
References not included in word count
List every review you cite. A complete entry includes the authors, year, article title, journal, and DOI or stable publisher link. Example:
The brief reference list and approximately 50-word AI-use disclosure are excluded from the 800–900-word memo total.
AI-Use Disclosure not included in word count
Write approximately 50 words naming the tool, the purposes for which you used it, and the stages it affected. If you used no AI, write: “I did not use AI for this assignment.”
Grading Rubric
| Criterion | Points |
|---|---|
| Ethical reasoning and use of lenses: distinguishes facts, assumptions, principles, lived effects, and institutional responsibility. | 25 |
| Evidence and claim calibration: uses assigned readings accurately, cites at least two, and does not present inferences as findings. | 20 |
| Counterarguments: represents both sides fairly before reaching a judgment. | 20 |
| Policy quality: provides specific, internally consistent, and usable rules rather than slogans. | 20 |
| Judgment and reflection: explains what the grade should represent and identifies a condition that could change the conclusion. | 10 |
| AI-use disclosure: complete and consistent with the submitted work. | 5 |
| Total | 100 |
Important: You are not graded on choosing “fair” or “unfair.” You are graded on the quality, honesty, and consistency of your reasoning.