Plato’s Allegory of the Cave offers a metaphor for the limits of human perception. Plato describes a group of prisoners in a cave who know the outside world only through shadows projected on a wall in front of them. Similarly, LLMs don't learn about the world from direct experience, but from “shadows” in its training data: billions of sentences written by people, describing things, events, and our conversations about them. It is trained to predict the next word in those texts, not to build an accurate picture of reality. Jan Kulveit suggested that an LLM is like a ‘blind oracle in yet another cave, who only hears the prisoners conversations about what they see. I thought It would be interesting to simulate a version