Ramanujan is a good analogy for this situation. Theories could be right/wrong, until there's a proof. Same with anything AI produces. There's always a "told you so" baked in with it's response.