Google’s AI Solved Math Olympiad Problems At Medalist Level
Google DeepMind has announced a major achievement in artificial intelligence.
The AI systems AlphaProof and AlphaGeometry 2 successfully solved four out of six problems at the International Mathematical Olympiad (IMO), which is equivalent to a silver medal.
To test the capabilities of the new AI systems, Google DeepMind researchers tasked them with solving six problems from this year’s International Mathematical Olympiad (IMO) and proving the answers were correct. AlphaProof solved two algebra problems and one number theory problem, one of which was the hardest in the Olympiad, while AlphaGeometry 2 solved the geometry problem. Two combinatorics problems remained unsolved.
For a number of reasons, solving mathematical problems that require advanced reasoning abilities is not yet within the capabilities of most AI systems. This is because these types of problems require the formation and use of abstractions. They also require complex hierarchical planning, setting subgoals, backtracking, and finding new paths, which is a difficult issue for AI. Both new AI models have the ability to use advanced mathematical reasoning to solve complex mathematical problems. AlphaProof was built using reinforcement learning, gaining the ability to prove mathematical statements in the formal programming language Lean. It was created using a pre-trained language model of AlphaZero, a reinforcement learning algorithm that previously taught itself to play chess, shogi, and Go.
These systems have achieved such success for the first time at IMO. Google DeepMind Vice President Pushmeet Kohli noted that there have been no systems capable of solving problems with such accuracy and versatility before.
