An answer that came up while discussing literature with ChatGPT just now

If a human and an AI present the same level of literary interpretation and respond to the same counterarguments with the same level of proficiency, on what grounds should we judge that one has truly understood while the other has not?
I believe that a definitive answer to this question does not yet exist.
Furthermore, I judge this to be not merely a matter of AI technology, but a philosophical issue that forces us to rethink the very nature of human 'understanding.'