AI Solves 167-Year Math Problem Humans Couldn't Crack

An engineer went for a jog, stopped to type a message into an AI chatbot, and helped push forward one of mathematics' most notorious unsolved puzzles.

Jarred Sumner, working at AI company Anthropic, asked Claude to tackle the Riemann hypothesis, a 167-year-old problem about prime numbers that has defeated the world's brightest mathematical minds. Over the next day and a half, he mostly just cheered the AI on with messages like "You got this."

Claude didn't fully solve the hypothesis, but it got further on a related test than any human ever has. The breakthrough marks a meaningful step toward cracking one of math's biggest mysteries, complete with a million-dollar prize for whoever solves it.

This wasn't a one-off lucky guess. In July, Claude disproved an 87-year-old geometry theory called the Jacobian conjecture. Rival company OpenAI announced its AI made 10 mathematical discoveries in just one week. Several famous problems left by celebrated Hungarian mathematician Paul Erdős have been solved by AI in the past year alone.

Just six months ago, these same AI systems couldn't count correctly. Now they're solving problems that take human mathematicians years to work through.

"I think it's going to accelerate the area massively," says Kevin Buzzard, a math professor at Imperial College London. Henry Yuen, a mathematician at Columbia University, marvels at the rapid progress from basic counting to solving complex proofs.

The Bright Side

This AI revolution could trigger a golden age of mathematical discovery. The technology can absorb and connect vast amounts of mathematical knowledge across different domains simultaneously, something no human has been able to do for over a century.

These breakthroughs could unlock solutions to real-world problems in physics, computer science, and cryptography. Mathematical advances have historically led to innovations in everything from internet security to medical imaging, often in unexpected ways.

The technology also makes advanced mathematics more accessible. People without years of specialized training can now explore complex mathematical territories, democratizing a field that has long been limited to elite experts.

Tim Gowers, a Cambridge professor who won the prestigious Fields Medal, notes that while some mathematicians worry about their role changing, the technology could free researchers to tackle even bigger questions. Rather than spending years on calculations, mathematicians might focus on asking the right questions and interpreting discoveries in meaningful ways.

Young mathematicians like Giorgio Navone at University College London see the potential for AI to handle tedious computations while humans focus on creative problem-solving and building deeper understanding.

The combination of human curiosity and AI's computational power is opening doors to mathematical insights that seemed impossible just months ago, promising breakthroughs that could benefit everyone.