Comparing a LLM to a less than ideal way of connecting two subjects is a more accurate model to work from than the popular construction of a "thinking" machine.