This makes no sense. LLMs and agents benefit from (good) abstraction as much as humans do.