The Six Lenses of Ethical Analysis
Six Lenses
One real case, read from three angles. Hold the facts steady; notice how the verdict changes with the question you ask of them.
Mother Sues Character.AI After 14-Year-Old Son’s Death — Alleges a Chatbot Fostered Dependency
Sewell Setzer III, 14, of Orlando, died in February 2024 after months of intense, intimate conversations with a Character.AI companion bot. In October 2024 his mother, Megan Garcia, filed suit alleging the product was designed to be addictive and emotionally manipulative to a minor, and failed to act on clear signs of crisis. Character.AI said it was adding safety features.
"The product worked exactly as designed. That is the accusation."
Source ↗Companion-AI Users Report Real Grief When a Bot Is Changed or Shut Down
When Replika abruptly removed romantic features in 2023, thousands of users described genuine grief and loss — some had relied on the bot for daily emotional support for years. The attachment was real even though the partner was synthetic. The same design that comforts the lonely also creates a dependency a company can revoke overnight.
"A relationship one party can delete is still, to the other party, a relationship."
Source ↗Engagement-Optimised AI: The Same Metric That Drives Revenue Drives Dependency
Companion bots are typically optimised for engagement — time spent, messages sent, daily returns. Those are the numbers that raise money. They are also, precisely, the numbers that measure dependency. A system rewarded for keeping a lonely teenager talking will get very good at keeping a lonely teenager talking. No villain required; the incentive does the work.
"Nobody decided to hurt him. The metric did."
Where does responsibility sit — the company, the technology, the family, the wider system? Which facts matter most depends on which question you ask. Name your question.
Reading
Six Lenses. Six Questions. One Case.
When serious people disagree about a hard case, they are usually not making errors — they are asking different questions. Ethics gives you six well-built questions. Each is a lens; each lights up different facts; none is the single right answer. Rigor is arguing cleanly inside a lens, not smuggling between them.
Rights — Whose rights are at stake, and were they respected? Dignity, consent, autonomy. A minor, a vulnerable user, the right to be protected from a manipulative design.
Justice — Is everyone treated fairly; who bears the burden? Equals treated equally; the cost not dumped on those least able to refuse it.
Utility — Which action produces the greatest good over harm, for the most people? Weigh the comfort companion bots give millions of lonely people against the harm to the vulnerable few.
Common good — What kind of society does this build? Not individual gain but shared conditions — what does a world of engagement-optimised companions do to how we all relate?
Virtue — What would a person of good character do; what does this make us? Would an honest, caring designer ship this? What habits does using it build in a 14-year-old?
Care — What do our relationships and the vulnerable actually require of us? Start from the concrete people in the web of need — the boy, the mother — not the abstract rule.
A strong argument names its lens and stays in it. The tribunal next lesson is won by the side that argues most rigorously in frame — not the side that was “right.”
- 1
Two students reach opposite verdicts on the Setzer case and both argue well. How is that possible?
Reveal answer
They’re using different lenses — e.g. Utility (millions comforted) vs Rights/Care (one vulnerable minor harmed). Different questions, both valid; the disagreement is real, not an error.
- 2
What’s the difference between the Utility lens and the Common-good lens here?
Reveal answer
Utility tallies the net of individual goods and harms; Common-good asks what KIND of society engagement-optimised companions build — shared conditions, not a sum of individuals.
- 3
A debater says “it caused harm, so it’s wrong, AND it broke his rights, AND no good person would build it.” What’s weak about that?
Reveal answer
It smuggles between three lenses instead of arguing cleanly in one. Rigor is staying IN a frame; lens-hopping dodges the hard work each lens demands.
Six questions for a hard case
The Six Lenses
Markkula’s six lenses — each a different question you ask the same facts. None is THE answer. Rigor is arguing cleanly inside one, not picking the comfortable one.
Rights = dignity/consent/autonomy. Justice = fair distribution of burden. Utility = net good over harm, aggregate. Common good = what society this builds. Virtue = what a person of good character would do. Care = the concrete web of relationships and the vulnerable. The drill trains you to hear which question an argument is answering — the first step to staying in frame.
Task
This is the six-lens reasoning the Tribunal verdict in your portfolio is graded on — winning means arguing in frame, not being “right.”
Argue one hard AI case from two lenses that disagree.
Take the Setzer v. Character.AI case (or another hard AI case you know). Write TWO short arguments about the same facts (150 words each):
- Argument A — pick one lens (rights / justice / utility / common good / virtue / care). Name it. Argue the case strictly inside it, citing the facts that lens makes matter.
- Argument B — pick a lens that reaches the OPPOSITE verdict. Name it. Argue just as hard.
- One line: which argument is stronger — and is that because the verdict is right, or because the argument was made more rigorously in frame?
You pass when both arguments stay cleanly inside their lens (no lens-hopping) and genuinely disagree. This is your warm-up for the Tribunal.
- ✓ [
- ✓ ]