Advocating an ethical memory model for artificial companions from a human-centred perspective

This paper considers the ethical implications of applying three major ethical theories to the memory structure of an artificial companion that might have different embodiments such as a physical robot or a graphical character on a hand-held device. We start by proposing an ethical memory model and then make use of an actioncentric framework to evaluate its ethical implications. The case that we discuss is that of digital artefacts that autonomously record and store user data, where this data are used as a resource for future interaction with users.