MAS methodology for responsible AI use