Toward Accountable and Explainable Artificial Intelligence Part Two: The Framework Implementation

This paper builds upon the theoretical foundations of Accountable eXplainable Artificial Intelligence (AXAI) capability framework presented in part one this paper. We demonstrate incorporation AXAI real time Affective State Assessment Module (ASAM) a robotic system. show that adhering to eXtreme Programming (XP) practices would help understanding user behavior and systematic Machine Learning (ML) systems. further collaborative software design development process (SDDP) facilitate identification ethical, technical, functional, domain-specific system requirements. Meeting these requirements increase confidence ML AI Our results ASAM can synthesize discrete continuous models affective state expressions for classifying them real-time. The continuously shares important inputs, processed data output information with users via graphical interface (GUI). Thus, GUI presents reasons behind decisions disseminates about local reasoning, handling decision-making. Through demonstrated work, we expect move toward enhancing systems’ acceptability, utility establishing chain responsibility if fails. hope work will initiate investigations on developing use suitable SDDP incorporating