Let the dead talk: How deepfake resurrection narratives influence audience response in prosocial contexts

Abstract

With the unceasing advancement of novel digital technologies, artificial intelligence (AI) is projected to change and challenge established social dynamics. As such, a deeper understanding of AI and its impacts will prove to be crucial in an AI-enriched world. In the current study, we focus on an emerging AI-based technology, deepfakes, and investigate the effects of a lesser-known use of deepfakes (i.e., using deepfakes for prosocial purposes). Specifically, we examine the impacts of exposure to a “deepfake resurrection” narrative on individuals' policy support and activism intentions in two contexts (i.e., drunk driving and domestic violence). We conducted a between-subjects experiment (N = 1990) that varied in the use (vs. non-use) of deepfakes and tested four psychological processes related to the effects of deepfake resurrection narratives. While the moderating role of point of view and the mediating role of perceived desecration of the dead did not receive much support, the narrative-based and the surprise-focused mediational pathways were shown to be important mechanisms of deepfake resurrection narratives. Finally, we reflect on the potential of prosocial deepfakes and deepfake resurrection narratives and discuss the findings’ relevance to both theory and practice.