Name the losers before you ship

Personal & Social Development · Ethics & Philosophy · ages 16-18

Name ______________________ Date ____________

Who bears the cost when a tenant screening tool is wrong?
- Flagged tenants who lose homes
- Nobody at all
- Only the programmers
A hiring filter trained on past hires favours men. What should you conclude?
- It proves men are better hires
- Data never carries old choices
- It learned past choices, so calling it objective hides them
A tool trained on old decisions can still carry old biases.
Circle one: True False
Duty thinking refuses the same trade. Why?
- It forbids using one group as a mere means without consent
- It dislikes hospitals
- It never allows any triage
Results thinking might bless a biased triage tool. Why?
- It loves bias
- It counts every survivor equally and totals the good
- It ignores all numbers
What does responsible building add before launch?
- A fancier logo
- Longer adverts
- Measuring errors per group and designing the appeal path
Why is technology is neutral a dangerous line?
- It treats a human choice with losers as mere maths
- Tools never matter at all
- Maths is always wicked
Risky routing software is tested on drivers who never agreed. Results thinking approves since delays fall. A duty thinker objects. Where exactly do they split?
- Speed, since one side is slower
- Consent, since one side requires it while the other totals past it
- Maps, since one side reads them better

Answer key

For grown-ups. Fold this page away before handing over the rest.

Flagged tenants who lose homes · Thresholds pick losers as well as winners. Rejected tenants pay for the error with their housing.
It learned past choices, so calling it objective hides them · The data carries the old favouritism forward. Objective language then covers a human choice.
True · True. Past choices ride inside the data, and the tool learns them like everything else.
It forbids using one group as a mere means without consent · Consent draws the line. A group sacrificed without agreeing to it is used, not treated.
It counts every survivor equally and totals the good · Only the total matters here. If more lives are saved overall, the skew inside the total goes unpunished.
Measuring errors per group and designing the appeal path · Name the harmed, count the skew, and build the way back. That work belongs before shipping, not after.
It treats a human choice with losers as mere maths · Neutral language moves blame from choosers to numbers. The losers stay real while responsibility evaporates.
Consent, since one side requires it while the other totals past it · Both see the shorter delays. Only duty asks whether the test drivers agreed to carry the risk.