Gains:
- Ability to analyze an ethical case from multiple perspectives, from consequentialist, deontological and virtue ethics frameworks
- Ability to clarify facts and stakeholders and name conflicts of principle
- Being able to make ethical judgments and decisions in a justified manner with bias control, without leaving it to artificial intelligence
The heart of applied ethics is the question "what to do?" in a real situation. is to produce a reasoned answer to the question. A hospital ethics committee discusses a treatment decision, a company evaluates whether a use of data is ethical, a researcher weighs the harm of an experiment to participants. The tools used in these works are ethical frameworks (approaches that offer a coherent set of principles and reasoning for evaluating a situation)—for example, consequentialism (approach that evaluates the rightness of an action according to its consequences, the benefits and harms it produces), deontology (approach that evaluates the rightness of an action by its compliance with duties, rights, and rules, independent of its consequences), and virtue ethics (approach that frames the question as “what would a virtuous person do?” rather than “what action?”). AI can be a quick tool to analyze a case from different frameworks, see overlooked dimensions and multiply options. But the final ethical judgment—which principle prevails, which decision to make, and the rationale for it—belongs to the human being, especially the responsible committee and expert.
In this unit, we will see how to analyze an ethical case in an AI-powered but human-judged way.
Multi-frame analysis: The real power of AI
The most common mistake in an ethical case is to stick to a single point of view. A good analysis sees the same situation from multiple frameworks and reveals tensions. This is where AI is powerful: it can analyze a case from consequentialist, deontological, and virtue-based perspectives simultaneously and quickly list what conclusion each framework leads to. This maps out the debate.
Steps:
1. Clarify the facts. Who, what, when is involved; distinguish between uncertain and assumed facts. Ethical analysis cannot be based on false facts.
2. Identify stakeholders and interests. Who is affected by the decision, what are the rights and interests of each?
3. Analyze from different frameworks. How does each framework view this situation and what action does it recommend?
4. Show tensions and conflicts of principle. If the frameworks reach different conclusions, where exactly is the conflict? (For example, between the autonomy of the individual and the benefit of society.)
5. Come to a reasoned decision. Justify which principle outweighs why by accepting the strongest objection of the other side.
Tip: Don't ask the AI to "tell me the right answer"; Say, “analyze this case from three frameworks and show the conclusion and strongest objection of each.” The AI's job is to create the map; You produce the decision and the justification. In ethics, "the answer given by the machine" is not a justification.
Why can't AI make ethical decisions?
Let's put this clearly. Ethical judgment; It requires context knowledge, prioritization between values, taking responsibility and the will to sign that decision. AI has none of these: it has no values, no accountability, limited knowledge of context, and can replicate cultural/ideological biases in training data. Moreover, AI often gives the “average” or most common opinion on a controversial topic; whereas in ethics the correct answer may be mostly in the minority or may not exist at all. AI opens options; You weigh and decide.
Caution: In high-stakes ethical decisions (medical, legal, institutional) AI output should never replace the judgment of a competent board or expert. Use AI as an “advisory drafter”; The final decision and responsibility lies with the responsible person. Also, if the actual case data contains personal/confidential information, do not enter it into public tools.
three mini cases
Case 1 — Multi-frame map. An ethics committee was discussing the allocation of a limited resource (e.g., a limited number of treatment options). They had the AI analyze the case from consequentialist, deontological, and virtue-based perspectives; Three different justifications and their conflict points were mapped in 20 minutes. Using this map as a starting point, the board conducted its own discussion and made its own reasoned decision.
Case 2 — The overlooked stakeholder. A company was evaluating a data usage decision. In its analysis, YZ marked a third group (people whose data was used but not involved in the process) who were indirectly affected, apart from the direct parties of the decision. The team had overlooked this stakeholder. AI's input made the decision more complete — but the team did the prioritization.
Case 3 — Biased suggestion corrected. An investigator had an AI analyze a culturally sensitive case. AI presented a single cultural view as the “universal truth” and ignored alternative value systems. The researcher noticed this one-sidedness and diversified the analysis. The dominant view of the AI's training data is balanced by human supervision.
Weak prompt / Strong prompt
Weak prompt:
What should I do ethically in this situation?
This prompt delegates ethical judgment and responsibility to the AI; You get one "answer", you don't see the rationale and framework, the bias remains hidden.
Powerful prompt:
Your role: applied ethics analyst (not decision maker). Analyze the following case. DON'T GIVE ME THE DECISION.1) List the ambiguous facts that need to be clarified.2) Remove the stakeholders and the interests/rights of each.3) Analyze the case separately from consequentialist, deontological and virtue ethics perspectives; Give the conclusion and strongest objection of each. 4) Show the basic conflict of principles between the frameworks. I will produce the decision and justification. Warn if there is personal/confidential data. Case: [write case]
This prompt strips out the multiframe map, leaves the decision up to you, and makes bias visible.
Four copyable templates
1) Fact and stakeholder clarification:
In the following case, list (a) certain known facts, (b) uncertain/assumed facts, (c) all stakeholders and the interests and rights of each. Add comments and decisions. Case: [case]
2) Three-frame analysis:
Analyze this case separately from the perspectives of consequentialism, deontology and virtue ethics. For each frame: how he sees it, the action he recommends, and the strongest objection to that suggestion. Declaring a framework "correct". Case: [case]
3) Policy conflict diagnosis:
Name the key principles on which the frameworks conflict in the analysis below (e.g., autonomy vs. the common good, justice vs. compassion). Show exactly where the conflict centers. Decision making.Analysis: [paste text]
4) Prejudice and cultural sensitivity audit:
Check whether this ethical analysis does not present a single cultural/ideological view as universal truth. Are there alternative value systems or perspectives that are being ignored? Highlight where there needs to be more pluralism. Analysis: [paste text]
Frame comparison chart
Frame
basic question
Focus
typical weakness
consequentialism
What's the best outcome?
Benefit/harm balance
Can crush minority rights
deontology
What does the assignment/rule say?
right, duty, principle
Strict, can ignore the outcome
virtue ethics
What does a virtuous person do?
character, intention
Vague action guide
Common mistakes
- Making AI decide. "What should I do?" The question delegates ethical judgment; AI creates a map, you decide.
- Being stuck in a single frame. Seeing the case from only one perspective; multi-frame analysis reveals tension.
- Not clarifying the facts. Ethical analysis based on false or assumed facts collapses.
- Not recognizing bias. AI may present the dominant cultural view as “universal”; Pluralism is provided by humans.
- Entering confidential/personal case into public tool. Risk of privacy violation in real files; Anonymize or use secure system.
In summary
The real power of AI in applied ethics case analysis is to quickly map a situation from multiple frames, extract stakeholders, and make conflicts of principle visible. But the ethical judgment—which principle prevails, the decision and the justification—belongs to the human being, especially to the responsible committee and expert; because AI has no values, no responsibility, no context knowledge, and carries bias. Clarify the facts, analyze multiple frameworks, name the conflict, check for bias, and make the decision yourself with justification. In high-stakes decisions, AI is placed alongside, rather than replacing, expert judgment.
Application task
- Choose a real or fictional ethical case (anonymize if it involves personal data).
- Extract facts and stakeholders with the “Fact and stakeholder clarification” template.
- Have the case analyzed from three perspectives with the “three-frame analysis” template.
- Name the underlying tension with the “principle conflict diagnosis” template.
- Write your own reasoned decision — accepting the opposing party's strongest objection.
checklist
- [ ] I clarified the facts and uncertainties.
- [ ] I identified all stakeholders and their interests.
- [ ] I had the case analyzed from at least three ethical frameworks.
- [ ] I have clearly named the principle conflict.
- [ ] I checked for bias and one-sidedness.
- [ ] I left the decision and justification to myself/the board, not to the AI.
- [ ] I protected confidential/personal data.