“A common misconception is that large language models represent the entirety of AI, when they are only one specific form of it.”