‘Design-for-responsible’ algorithmic decision-making systems: a question of ethical judgement and human meaningful control

Abstract

There has been an intensified drive towards the ‘responsible adoption’ of automated decision-making systems. Such initiatives are questionable in that they fail to properly take into account the phenomenological complexities of human judgement. Their representational systems suffer from the frame regress limitation, thereby making them unable to match the human ability to discern cues and actions that are of ethical significance within a given moral dilemma. We argue that a socio-technical system can only meet its moral responsibilities by attributing this directly onto the human decision maker’s shoulders. Yet, prior to establishing such responsibilities, socio-technical design configurations must not thwart the human ability to expertly deliberate and judge in a meaningful and relevant manner in the first place.