⚖️ Algorithmic Bias & Fairness Metrics — 3D

This is the same fairness model as the 2D bias simulator: six sliders — Demographic Parity, Equalized Odds, Calibration, Bias Level, Fairness Score and Accuracy — drive the identical per-group bias formula (Group A = Bias Level, Group B = Bias Level × 0.8, Group C = Bias Level × 1.2) and the identical fairness-indicator count (Fairness Score ÷ 10). In 3D, the three demographic groups become real orbiting nodes arranged around a central fairness hub, each wrapped in a bias ring whose radius scales with its own bias value exactly as the 2D stroke-ring did, connected to the hub by real 3D beams. The fairness indicators become a drifting shell of particles around the hub, and the bias-score formula |P(Y=1|A=a) − P(Y=1|A=b)| from the theory section is computed live between Group A and Group C. Demographic Parity, Equalized Odds, Calibration and Accuracy additionally drive the hub scale, node size, rotation speed and beam brightness — giving every one of the six original metrics a real, inspectable effect from any camera angle via free orbit.

3D · WebGL (Three.js r128) · 60 FPS target · runs fully client-side, no install