Remembering with AI: From Distributed Memory to AI-Curated and Human-AI Co-Memory

Abstract

The emerging field of distributed memory has called attention to the memory resources spread outside of us – in our environment, social relations, embodied practices, and technological resources. In this paper, I attend to different ways of distributing, curating, and co-constituting one’s memory to and with extraindividual aids ranging from more traditional technologies to algorithmic personalisation and conversational AI companions. I defend a view that in relation to memory, conversational AI agents based on LLMs require their own category that differs both from traditional technologies and from other types of AI systems, such as algorithmic profiling systems. I suggest that they should be understood as quasi-social partners in a shared remembering process and that this can already be accomplished with existing conversational AI agents, such as HereAfter AI or Replika. To conclude, I discuss some of the risks associated with remembering with AI.