Before applying a new system to other people, I find it useful to ask whether I would accept the same treatment myself.
Would executives agree to the same level of AI monitoring being introduced for frontline employees? Would we accept an automated decision with no meaningful explanation if it affected our own family? Would we consider the appeal process fair if we were the person required to use it?
This is a simple question, but it reveals differences that business language can easily hide. Decisions often appear reasonable when we benefit from the efficiency and someone else carries the inconvenience, uncertainty or risk. AI can increase that distance because the people designing a system may never meet the people living with its consequences.
Reciprocity brings the decision back to a human scale. It does not guarantee that every stakeholder will receive exactly the same outcome. It asks whether the standard remains defensible when our positions are reversed.
A decision deserves greater trust when we would still consider it fair from the other side of the system.