🧊 3D AI Ethics Audit Bench: Fairness Metrics, Threshold Gates, Balance Scale and Explainability Monitors
A Blender-built audit bench: candidate tokens stacked by model score for two groups, glass threshold gates, a balance scale that tips with the selection-rate gap, fairness tubes and monitors for ROC, explainability and the threshold trade-off, driven by the same model as the 2D lab.
⚙ Under the hood
A Blender-built audit bench: candidate tokens stacked by model score for two groups, glass threshold gates, a balance scale that tips with the selection-rate gap, fairness tubes and monitors for ROC, explainability and the threshold trade-off, driven by the same model as the 2D lab.
AI ethicsresponsible AIalgorithmic fairnessdemographic parityequalised oddsbias auditexplainabilitylogistic regressionBlenderThree.js
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install