Since LLMs are basically linear algebra all the way down, this is vaguely reminiscent of how human brains also have a very hard time understanding neural circuitry despite literally being made from it.