The need for formal specifications of morally relevant information for algorithmic decision-making (short paper)

Marija Slavkovik

Keywords: machine ethics, AI alignment, algorithmic decision-making, artificial moral agents

TL;DR: Motivates the need for developing formal specification of morally relevant information for use by autonomous agents

Abstract: Artificial Intelligence and autonomous systems that use it have an established impact on society and individuals. Their behaviour and the decisions they make on our behalf are increasingly in need of moral considerations. If machines are to have moral sensitivity or make moral choices, we need to consider the question of morally relevant information that needs to be made available for their use. Currently morally relevant information is not seriously considered with different people using different data structures to represent different aspects of it. This work motivates the need to develop formal specifications of morally relevant information supplied for the use by machines.

Submission Number: 2