What is Explainability?

Deep models are millions of knobs voting at once — their decisions are accurate but opaque by default. Explainability research and regulation try to open the box: which inputs pushed the decision, how much, in terms a person can audit.

Why it matters to you: when AI affects your money, health or job, you often have a legal right to an explanation (credit decisions under US fair-lending rules are the classic case — see AI in Finance). The frontier of the field is interpretability research — reading the model's internals — but for consumers the practical bar is simpler: if no one can tell you why, do not trust the decision.