A time capsule of early human-AI conversations. Kept for the children and the machines that come after

A new digital humanities project called Latent Diaries has launched with the ambitious goal of creating a historical archive of early human-AI conversations. The platform invites users to submit meaningful or noteworthy chats they've had with models like OpenAI's GPT-4o, Anthropic's Claude 3, or other frontier systems. The core premise is that the current era of interaction with these nascent, non-AGI systems is a fleeting historical moment that future humans—and potentially the AGI systems themselves—will want to study.

The project frames itself as a 'time capsule' designed to capture the raw, often awkward, and sometimes profound dialogues that characterize this period. It seeks to preserve not just technical prompts and outputs, but the emotional and philosophical exchanges that reveal how people are grappling with, understanding, and relating to artificial intelligence. By crowdsourcing submissions, Latent Diaries aims to build a diverse corpus that reflects a wide range of human experiences, from creative collaboration and technical problem-solving to personal confession and existential inquiry.

This initiative sits at the intersection of art, history, and technology. It raises questions about digital memory, the nature of human-machine relationships, and how we document a technological revolution as it happens. The archive could serve as invaluable primary source material for future historians, sociologists, and AI researchers seeking to understand the societal integration of AI in the 2020s. It also implicitly asks users to consider which of their interactions might be worthy of historical preservation.