Erdos Problem 1196: Can AI now solve maths that no human can?

Artificial intelligence has achieved a significant milestone in pure mathematics by solving Erdos Problem 1196, a challenge involving primitive sets that had eluded human thinkers for decades. While professional mathematicians like Stanford’s Jared Duker Lichtman spent years researching related conjectures, amateur mathematician Liam Price utilized a sophisticated AI prompt to generate a valid proof in under 80 minutes. This breakthrough highlights the legacy of Paul Erdos, the itinerant mathematician who posed over 1,200 problems, and signals a shift in the academic landscape. Although the AI produced raw and unstructured output, its core logic was verified as correct and elegant by experts. This event suggests that AI is transitioning into a collaborative role, acting as a digital colleague that provides rapid intuition, though human expertise remains essential for verifying complex proofs and understanding their broader implications within number theory.