CS-215 · Topic 3 · Required course map

Four reviews. Five principles. One decision routine.

Use this map to prepare for class and to check your memo. It is a course synthesis of four reviews—not a framework the four author teams jointly proposed.

1 · What each review contributes—and what it cannot establish alone

Jobin, Ienca & Vayena (2019)

Map: 84 AI ethics guidelines; recurring principle language and differences in interpretation.

Best question: Which values recur across guidelines?

Limit: The review does not prescribe one university policy.

Corrêa et al. (2023)

Map: 200 guidelines and governance documents; recommendations and implementation gaps.

Best question: What turns an ethical promise into an operational commitment?

Limit: Published guidance is not the same as enforcement or demonstrated impact.

Giarmoleo et al. (2024)

Map: 309 articles through December 2021; system-design and human–AI interaction concerns, plus act- and agent-centered responses.

Best question: Is the problem in the system, its use, the rule, or human judgment?

Limit: The literature window ends in 2021.

Groen, Sharon & Becker (2026)

Map: moral concerns through principles, lived realities, and power structures.

Best question: Who experiences the rule, who decides, and who bears the risk?

Limit: This conceptual lens does not prove that a particular classroom policy works.

2 · Translate principle language into a governance question

Five recurring AI ethics principle families, plain-language course definitions, and governance questions
Principle family	Plain-language course definition	Governance question for a university
Transparency	Make the AI role, relevant limits, and uncertainty inspectable.	What must be disclosed, to whom, at what time, and in what form?
Fairness / justice	Inspect how access, benefits, burdens, errors, treatment, and remedies are distributed.	Which groups, access differences, error patterns, remedies, and appeal paths matter?
Responsibility / accountability	Name the people and institutions responsible for checking, deciding, explaining, and remedying harm.	Who checks, decides, answers for harm, and supplies a remedy?
Privacy	Constrain data collection, use, retention, inference, and sharing.	What data is collected, retained, inferred, shared, protected, or made contestable?
Safety / non-maleficence	Anticipate, test, reduce, monitor, and stop foreseeable harm.	What failure modes must be tested, and who has authority to stop the use?