Posthumous AI: The afterlife of data

This essay delves into the interplay between artificial intelligence (AI), death, and the afterlife. It argues that the relationship between AI and the dead extends far beyond particular historical cases, such as the use of images of deceased individ-uals to train machine learning models. Instead, it is rooted in AI’s structural reliance on memory traces and its operation within the conceptual domain of the afterlife. This entanglement manifests at two interconnected levels: the personal, where AI enables the virtual reanimation of individual identities; and the collective, where it preserves and reinforces cultural memory by perpetuating shared imagery, visual motifs, and stereotypes. First, the essay examines how neural networks generate digital doppelgängers, raising critical questions about identity persistence in virtual environments. It then explores how training data, though seemingly absorbed by AI systems, re-emerge in generated outputs as spectral traces or revenants. These forms of posthumous presence reveal how AI reconfigures practices of memory and mourning, embedding the dead within algorithmic systems of visibility, control, and reproduction. This ultimately calls for political consideration of the governance of digital afterlives.