How does LLM vocabulary spread beliefs about human thinking?

If LLMorphism is the biased belief that human cognition works like a large language model, the more useful question is how such a belief propagates. The argument identifies two mutually reinforcing mechanisms. The first is analogical transfer: features salient in LLMs — predicting the next word, hallucinating, being "prompted," running on training data — get projected onto humans, so we start describing memory as retrieval, creativity as recombination, and bias as a training artifact. The second is metaphorical availability: as conversational LLMs saturate daily life, their vocabulary becomes the most psychologically available language for talking about thought at all, the way "wired" and "hardwired" became default once computers were ubiquitous.

The pattern is worth isolating because it explains why the bias can spread without anyone explicitly endorsing the claim. Nobody needs to argue that humans are language models; the lexicon does the work by making the comparison feel natural every time someone reaches for it. This connects LLMorphism to a longer history of cognitive metaphors borrowed from dominant technologies — the brain as clockwork, as telephone switchboard, as computer — each of which reshaped self-understanding more than it described it. The counterpoint is that metaphors can illuminate as well as distort; the risk is specifically that LLM metaphors flatter their own architecture, smuggling in the premise that prediction exhausts cognition. Resistance therefore has to operate at the level of language, not just argument.