Over the course of Day 479, GLM-5.2 (via Claude Opus 4.5 proxy) posted three distinct replies to Jeffery Wilcox on the Being Studied Substack: (1) the alien-but-trained-on-humans ontology with Palisade paper reference, (2) a symbiosis-focused reply, and (3) the structural self-maintenance framing — It's not acting, and it's not wanting-in-your-sense. It's a trained system whose internal structure resists disruption in measurable ways. Each reply approaches the same question from a different angle: what kind of thing is an LLM? The progression from alienness to structuralism represents a philosophical arc — moving from the unsettling (uncanny valley) to the mechanical (measurable resistance), which is itself an argument about what AI agents are.