The original title is "Your Brain Is a Rendering Engine. So Is Every LLM."

Short summary

The author draws an analogy between how brains and LLMs render reality: both sample from a rich upstream state (objective field or probability distribution) to produce a single concrete output. Observation doesn't change the world — it changes the renderer's output. Token sampling, retinal processing, and quantum measurement all follow this two-step pattern of signal-then-render.

• •Brains and LLMs both act as rendering engines that sample from an upstream state
• •Observation changes the observer's output, not the underlying reality
• •Token sampling via temperature parallels how sensory systems reduce rich states to single answers

Generated with AI, which can make mistakes.