Temporal Evolution of Cognitive Knowledge Networks in AI-Assisted Conversations
March 15, 2026 Rejected
Abstract
How does a person's knowledge landscape evolve through sustained AI-assisted conversation? We extend our cognitive MRI methodology to the temporal domain, tracking how a semantic similarity network grows over 29 months across 1,908 ChatGPT conversations. The network exhibits super-linear densification, sub-linear preferential attachment, and early community stabilization. Bridge conversations emerge at identifiable moments and maintain their structural role. Comparison across model eras reveals that AI model capabilities shape sub-network topology. These findings suggest that individual AI-assisted knowledge exploration self-organizes according to the same macroscopic laws observed in collective knowledge systems.