Abstract The present paper explores how large language models (LLMs) can function as personal memory systems. Unlike conventional lifelogging technologies, which passively record data, LLMs can actively participate in memory processes through dialog, interpretation, and narrative generation. LLMs, we suggest, mark a new chapter in the history of memory technology, one characterized by the active solicitation of mnemonically-relevant information and the co-production of narrative artifacts. Courtesy of their conversational capabilities, LLMs are poised to influence memory in the manner of a human social companion. This, we suggest, blurs the distinction between technologically- and socially-situated approaches to the shaping of human memory.
Smart et al. (Sat,) studied this question.