Fairness in AI is a design principle aimed at preventing the systematic disadvantage of individuals or groups through AI systems. Formal fairness notions include "demographic parity" (equal acceptance rates across groups), "equal opportunity" (equal hit rates among those actually qualified) or "calibration" (equal accuracy across groups). Some of these are mathematically incompatible – which makes the debate non-trivial.
In recruiting, fairness is anchored through balanced training data, context-aware evaluation metrics, human review of sensitive decisions, clear explainability ("Why was this application scored lower?") and an open complaint process. Classic anti-discrimination rules such as the German AGG remain valid.
For candidates, fairness above all means: a right to transparent, traceable decisions and to human review. Anyone who suspects they have been disadvantaged by AI should approach HR or counselling services.
Lunigi aligns to these principles: AI sorts roles, the final decision rests with the person.