Sort of, but the benefit is more along the lines of regression testing to make sure when things change they don't break. Also it does random fuzzing which can generate weird, edge-case inputs, like bizarre Unicode strings or extremely large numbers. These might technically pass declared schema validation but end up breaking your actual handler or database implementation.