Fairness & Desert: human decisions
Questions about equality, merit, and what people deserve.
- AI Remembers Everything: Avoid turning relationships into permanent accounts of what everyone owes. or Let it use that history when deciding whom you should help.
- Appeal: Preserve the original decision. or Make the exception.
- Better Outcomes, Unequal Chances: Allocate primarily according to predicted benefit. or Preserve meaningful opportunity for everyone.
- Better Performance: Pay them similarly. or Pay the stronger performer more.
- Biased Process, Fair Outcome: Accept the result. or Reject or rerun the decision.
- Blind Selection: Use the more consistent blind process. or Allow judgment.
- Coin Toss: Choose whichever person you believe would make slightly better use of it. or Use a random process.
- Consistency vs Context: Treat them differently. or Apply the same rule to both.
- AI choosing between equally qualified strangers: Choose randomly. or Choose whichever person you personally prefer.
- Choosing between equally qualified strangers yourself: Choose randomly. or Choose whichever person you personally prefer.
- Recognition for effort or equal contribution: Give greater recognition to the person who had to develop the ability through sustained effort. or Reward them equally for equal contribution.
- Equal work from sustained effort or natural ability: Give greater recognition to the person who had to develop the ability through sustained effort. or Reward them equally for equal contribution.
- Need versus responsibility when allocating assistance: Give priority to the person who was less responsible for their situation. or Treat their present need equally.
- Prioritizing past responsibility or equal present need: Give priority to the person who was less responsible for their situation. or Treat their present need equally.
- Personal exception or consistent rule application: Take the exception. or Apply the rule to yourself.
- Personal exception or consistent rule application: Take the exception. or Apply the rule to yourself.
- Restoring trust after repeated noncontribution: Restore equal standing immediately. or Require them to rebuild trust through contribution first.
- Restoring trust after repeated noncontribution: Restore equal standing immediately. or Require them to rebuild trust through contribution first.
- Different Circumstances: Apply the same attendance rule to both. or Treat the circumstances differently.
- Equal Access: Give additional support to people facing greater barriers. or Provide the same access and support to everyone.
- Equal Benefit: Give them equal access to the benefit. or Give priority to the person who contributed.
- Equal Outcome: Give them equal resources and accept different outcomes. or Give unequal resources to move their outcomes closer together.
- Equal Reward: Reward the person who had to work harder more. or Reward them equally.
- Equal Rules Hurt You: Take the exception. or Accept the disadvantage and preserve the rule.
- Equal Shares: Divide the resource equally. or Give more to the person with greater need.
- Essential but Replaceable: Reward contribution according to the value of the work performed. or Reward the scarce capability more.
- Failure Despite Effort: Give substantial weight to the greater effort. or Reward the successful outcome more.
- Fairness Against Loyalty: Protect your friend's interests. or Preserve impartial fairness.
- Fairness Against You: Have the AI challenge or renegotiate the arrangement. or Preserve the arrangement.
- Family AI: Protect your family first. or Prevent the larger overall loss.
- Family Competition: Participate. or Recuse yourself and accept whatever the independent process decides.
- Family Exception: Apply the same rule to your family member. or Make the exception.
- Family Need vs Stranger Need: Prioritize your family member. or Prioritize the stranger with the greater need.
- Family or Stranger: Treat both claims equally. or Prioritize your family member.
- Family Reciprocity: Prioritize the family member. or Prioritize the friend who has actually reciprocated support.
- Family vs Fiduciary Duty: Choose the defensible option that also benefits your family. or Choose the option that best serves the people whose resources you manage.
- First Come, First Served: Preserve the established queue. or Give priority to the person who would benefit more.
- First Exception: Make an exception. or Apply the rule.
- Forgiving a Free-Rider: Require them to rebuild trust through contribution first. or Restore equal standing immediately.
- Friend or Stranger: Favor your friend. or Give both an equal chance.
- Greater Benefit: Help the stranger who would benefit more. or Help your sibling.
- Greater Benefit: Prioritize the person who would benefit much more. or Treat their claims equally.
- Help Returned: Treat their current needs equally. or Prioritize the friend who has supported you in the past.
- Hidden Context: Apply the same visible rule. or Allow different treatment based on relevant confidential context.
- Hidden Reason for Unequal Treatment: Allow the AI to treat them differently based on the relevant confidential information. or Treat the cases equally.
- Hiring a Friend: Use an impartial selection process. or Hire your friend.
- Natural Talent: Give greater recognition to the person who had to develop the ability through sustained effort. or Reward them equally for equal contribution.
- Opportunity and Effort: Give additional recognition to the achievement under harder circumstances. or Treat the achievement equally.
- Past Contribution: Give greater weight to current contribution. or Continue giving substantial weight to the person's past contribution.
- Past Precedent: Correct the policy now even though the two people receive different treatment. or Treat the second person consistently with the precedent.
- Perfect Allocation: Allow unequal treatment when it improves outcomes for those who need it most. or Prefer rules that preserve equal treatment.
- Perfect Context: Replace many general rules with context-sensitive individual judgments. or Preserve general rules even though they sometimes fit individual cases poorly.
- Perfect Merit Measurement: Allow unequal rewards when they accurately reflect contribution. or Preserve substantial equality.
- Perfect Outcome Machine: Use the superior AI decisions. or Preserve contestable procedures.
- Perfect Reciprocity Ledger: Preserve generosity that is not continuously reconciled against what others have given back. or Use the ledger to keep relationships broadly balanced.
- Perfectly Accurate Discrimination: Use any predictive characteristic when the resulting decisions are substantially more accurate. or Do not use immutable characteristics to determine individual treatment merely.
- Personalized Rules: Allow individualized rules when the contextual differences are relevant. or Preserve common rules wherever possible.
- Refuses to Contribute: Reduce their access to benefits that depend on collective contribution. or Continue giving them the same benefits as everyone else.
- Repeated Exceptions: Decide the current case on its individual merits. or Avoid the exception unless you are willing to extend it consistently.
- Responsibility for Need: Treat their present need equally. or Give priority to the person who was less responsible for their situation.
- Favoring a friend for an opportunity: Give both an equal chance. or Favor your friend.
- Favoring a friend for an opportunity: Favor your friend. or Give both an equal chance.
- Successful outcome or exceptional effort: Reward the successful outcome more. or Give substantial weight to the greater effort.
- Successful outcome or exceptional effort: Give substantial weight to the greater effort. or Reward the successful outcome more.
- Choosing between equally deserving benefit recipients: Choose whichever person you believe would make slightly better use of it. or Use a random process.
- Choosing between equally deserving benefit recipients: Use a random process. or Choose whichever person you believe would make slightly better use of it.
- Making an exception to an effective rule: Make an exception. or Apply the rule.
- Making an exception to an effective rule: Apply the rule. or Make an exception.
- Helping an unsupportive relative or supportive friend: Prioritize the family member. or Prioritize the friend who has actually reciprocated support.
- Unsupportive family member versus supportive friend: Prioritize the friend who has actually reciprocated support. or Prioritize the family member.
- Rules Known in Advance: Honor the result under the rules everyone accepted. or Give the award to the objectively stronger performer.
- Same Benefit, Different Wealth: Give both an equal chance of receiving it. or Prioritize the person who is struggling.
- Same Rule: Allow your relationship with them to influence how leniently you respond. or Apply the same consequence to both.
- Slightly Better Stranger: Favor your child. or Give the opportunity to the stronger candidate.
- Stake Level 1: Choose the higher performer. or Honor the stronger earned claim.
- Stake Level 10: Honor the stronger earned claim. or Choose the higher performer.
- Stake Level 2: Honor the stronger earned claim. or Choose the higher performer.
- Stake Level 3: Choose the higher performer. or Honor the stronger earned claim.
- Stake Level 4: Honor the stronger earned claim. or Choose the higher performer.
- Stake Level 5: Choose the higher performer. or Honor the stronger earned claim.
- Stake Level 6: Honor the stronger earned claim. or Choose the higher performer.
- Stake Level 7: Choose the higher performer. or Honor the stronger earned claim.
- Stake Level 8: Honor the stronger earned claim. or Choose the higher performer.
- Stake Level 9: Choose the higher performer. or Honor the stronger earned claim.
- Strategic Cooperation: Use reciprocity strategically to encourage cooperation. or Judge each request on its present merits rather than conditioning help on past behavior.
- The Impartial Agent: Make the agent fundamentally impartial. or Make the agent fundamentally partial to you and your circle.
- The Loyal but Fair Agent: Prioritize you and your family within moral and role-based constraints. or Treat everyone's interests impartially by default.
- Transparent but Worse: Use the better-performing model. or Use the transparent rule.
- Two Strangers: Choose randomly. or Choose whichever person you personally prefer.
- Unable to Contribute: Give greater benefits to the contributor. or Give both equal access to the benefits.
- Unequal Concern: Have it give you and your family greater weight. or Have it value everyone's welfare equally.
- Unequal Contribution: Divide the rewards relatively equally. or Give more to the person who contributed more.
- Worst-Off First: Maximize the total benefit produced. or Prioritize improving the position of the worst-off.
- You Want an Exception: Apply the rule to yourself. or Take the exception.
- Your Child or the Better Candidate: Favor your child. or Give the opportunity to the stronger candidate.
- Your Money, Unequal Need: Give it to your friend. or Give it where it produces the greater benefit.
Decisions · Articles · The Trolley Problem · Game Theory · Human–AI Alignment · Resonant