User-centric evaluation of explainability of AI with and for humans: a comprehensive empirical study