Digital Replacement of the Dead: A Legitimate Worry?

Abstract Digital “resurrection” of the dead has become more of a reality in the last few years due to the rapid advancement of machine learning and large language models. So-called “interactive personality constructs of the dead” (IPCDs) are AI either trained on decedents’ “digital remains” (e.g. emails, texts, and social media content) after the fact or intentionally developed in cooperation with future decedents pre-mortem to simulate the way they would have interacted with others via some combination of text, video, and audio modalities. While some literature on IPCDs speculates about beneficial uses of this technology, much of it raises the alarm about its possible dangers. For example, some scholars wonder if a tool like this could be used to help people navigate grief, but many also worry it might make grieving worse. Beyond the realm of mere grieving, two significant concerns about IPCDs found in the literature involve the potential for commercial abuses and disrespectful treatment of the dead. The latter will be the focus of the present paper. More specifically, I want to talk about the problem of replacement, which suggests that, to a greater degree than older technologies aimed primarily at aiding remembrance of deceased loved ones, IPCDs seem to make possible, or even likely, an ethically troubling replacement of the deceased in people’s lives. Whether or not this is true has become a matter of recent debate, and this paper argues replacement is indeed a danger that should be taken seriously when developing and using this new technology.