Google DeepMind takes step closer to cracking top-level maths

Researchers at Google DeepMind have developed two new AI systems, AlphaProof and AlphaGeometry 2, which have made significant progress in tackling complex mathematical problems. These systems worked together to solve questions from the International Mathematical Olympiad, a global competition for gifted students. Although they did not win a gold medal, their performance was close to that of top human mathematicians. The AI systems demonstrated their capabilities by solving problems in algebra, geometry, and number theory. In one problem, they had to prove a geometric relationship involving angles and midpoints. Another challenge involved navigating a game board with hidden monsters, requiring strategic movement and memory. Overall, the AI systems' breakthrough brings them closer to surpassing human mathematicians in complex mathematical domains.