Designing explainable artificial intelligence with active inference: A framework for transparent introspection and decision-making