Google's AI Outperforms Math Olympiad Gold Medalists

San Francisco: Google has announced that its latest AI-driven math system, AlphaGeometry2, has surpassed the performance of gold medalists at the International Mathematical Olympiad (IMO). The AI successfully solved 84% of complex geometry problems, exceeding the 81.8% average achieved by top human competitors.

Developed by DeepMind, the system initially performed at a silver medalist level when introduced last year. However, after significant enhancements, Google now claims it has reached and exceeded gold-standard performance.

Breakthrough in AI Mathematics

To boost its problem-solving abilities, Google expanded the AlphaGeometry language, enabling the system to handle object movements, complex angle ratios, and linear equations. These improvements raised its coverage rate on IMO geometry problems from 66% to 88%.

Additionally, the integration of Google’s Gemini AI tool has refined AlphaGeometry2’s search capabilities, enhancing its reasoning process.

Future Improvements & Challenges

Despite its success, Google acknowledges that AlphaGeometry2 still faces limitations. The AI struggles with non-linear equations, inequality-based problems, and variable point calculations—areas that researchers aim to refine.

"Our long-term goal is to achieve fully automated and error-free geometry problem-solving while improving speed and accuracy," DeepMind researchers stated.

A Step Towards AI-Driven Mathematics

With these advancements, Google continues to push AI’s potential in tackling complex mathematical challenges, reinforcing the growing role of artificial intelligence in problem-solving and education.