A model that gives an answer without any reason is hard to trust, especially where decisions affect people. Explainability is the effort to show which inputs mattered and why, turning an opaque verdict into something a person can examine and challenge.
Complex models resist simple explanation, and an explanation that is merely plausible can mislead as much as no explanation at all. The goal is honest insight into the model's behaviour, enough to catch mistakes and unfair patterns rather than to rationalise them after the fact.