Mid-20th century problems posed by Paul Erdős are now being solved by AI. Mathematicians are analyzing these specific successes to determine if AI can tackle more complex, non-pattern-based proofs. This trend suggests a shift in how researchers approach formal verification. It provides a blueprint for automating discovery in pure mathematics.