AI scientist Blaise Aguera y Arcas argues that large language models have a great deal to teach us about “the nature of language, understanding, intelligence, sociality, and personhood.”