The problems solved by AI with Lean proof assistants often exhibit patterns (like ellipses, Platonic solids, or geometric constructions) that a sufficiently clever mathematician could recognize as significant, but if latent in Lean code, these patterns might be invisible because constructions that later prove to have unifying power often appear unassuming in formal notation.