Chinese AI Cracks 30-Year Math Problem in Research First

For the first time ever, Chinese researchers have used a homegrown AI model to crack a mathematical problem that had remained unsolved for over 30 years.

The team, led by Professor Xu Disheng from Greater Bay University, used the Intern-S2 AI model to prove the Ghys conjecture. This complex mathematical puzzle, proposed by French mathematician Etienne Ghys in 1995, explores how chaotic systems behave and has long been considered a key test of mathematical reasoning.

The breakthrough wasn't about replacing human mathematicians. Instead, it showcased a powerful new partnership between people and machines working together toward discovery.

The AI handled the heavy lifting: combing through mountains of research papers, breaking down complex propositions, exploring multiple proof strategies at once, and rapidly testing thousands of possible solutions. Meanwhile, the human researchers guided the overall direction, evaluated which approaches showed the most promise, and provided critical feedback to steer the model's next steps.

Two research teams at Peking University formally verified the proof using their own framework, confirming the solution was mathematically sound.

The Ripple Effect

The Shanghai AI Laboratory is already packaging this breakthrough into a reusable tool that other scientists can access. They've built a scientific discovery platform that connects AI models with real lab equipment, allowing researchers across different fields to use AI as a tireless research assistant.

The impact is already showing up in other areas of science. A chemistry team at Wuhan University used the platform to compress their research timeline from up to 10 years down to just one or two years for developing new electrochemical processes.

This human-plus-AI collaboration model is now expanding into physics, chemistry, and life sciences. Researchers see AI taking over the tedious, time-consuming tasks like literature reviews, testing countless variations, and checking logical steps, while scientists focus on the creative and strategic thinking that drives innovation forward.

The team believes this approach could fundamentally change how research gets done. By handling massive trial-and-error cycles and broadening the search space for human thinking, AI research assistants may help solve problems faster and explore possibilities that would take human teams decades to investigate.

Scientific discovery just got a powerful new partner, and the possibilities are opening up across every field of research.