thought experiment arguing that a computer cannot exhibit "understanding"

metaphor to describe the theory that large language models, though able to generate plausible language, do not understand the meaning of the language they process