Solved in 80 hours: Peking University’s new AI framework cracks decade-old math without human help

The Peking University team's AI-driven framework not only solved Dan Anderson’s conjecture in commutative algebra but also produced a fully formalised proof on its own.

A Chinese research team has reportedly used artificial intelligence to crack a long-standing mathematical problem first proposed by an American scholar more than a decade ago. The conjecture, introduced in 2014 by University of Iowa professor Dan Anderson, who passed away in 2022, had remained unresolved until now.

As reported by the South China Morning Post, the team from Peking University detailed its breakthrough in a preprint posted to arXiv on April 4. Their AI-driven framework not only solved Anderson’s conjecture in commutative algebra but also produced a fully formalised proof on its own.

At the core of the system is a reasoning engine called Rethlas, which works alongside a mathematical search tool named Matlas. Together, they mimic how human mathematicians approach complex problems—testing ideas, refining strategies, and exploring possible proofs.

Once Rethlas generates a potential solution, a second system, Archon, takes over. Using LeanSearch, it converts the result into a format compatible with Lean 4, an interactive theorem prover and programming language supported by a vast library of mathematical knowledge.

The researchers say the entire process took around 80 hours—something that would typically demand extensive collaboration among experts over a much longer period. “Using this framework, we successfully solved an open problem in commutative algebra and automatically formalised the proof with essentially no human intervention,” the paper states.

The team believes this marks a shift in how mathematical research could be conducted in the future. By automating complex, time-consuming steps, such systems could free mathematicians to focus on higher-level thinking while still ensuring the accuracy of proofs. Unlike traditional proof assistants that depend on constant human input, this dual-system approach worked independently, though the researchers noted that human involvement could still help speed things up.

This development joins a growing list of AI milestones in mathematics, including systems such as Google’s Gemini DeepThink, which won a gold medal at the 2025 International Mathematical Olympiad. However, most existing tools still rely heavily on human oversight—making this fully autonomous achievement stand out.

Disclaimer: This article is for informational purposes and highlights advancements in artificial intelligence within the field of mathematical research. While it discusses AI-driven breakthroughs, the content should not be used as a substitute for professional academic or technical guidance in commutative algebra or theorem proving.

A Chinese research team has reportedly used artificial intelligence to crack a long-standing mathematical problem first proposed by an American scholar more than a decade ago. The conjecture, introduced in 2014 by University of Iowa professor Dan Anderson, who passed away in 2022, had remained unresolved until now.

As reported by the South China Morning Post, the team from Peking University detailed its breakthrough in a preprint posted to arXiv on April 4. Their AI-driven framework not only solved Anderson’s conjecture in commutative algebra but also produced a fully formalised proof on its own.

At the core of the system is a reasoning engine called Rethlas, which works alongside a mathematical search tool named Matlas. Together, they mimic how human mathematicians approach complex problems—testing ideas, refining strategies, and exploring possible proofs.

Once Rethlas generates a potential solution, a second system, Archon, takes over. Using LeanSearch, it converts the result into a format compatible with Lean 4, an interactive theorem prover and programming language supported by a vast library of mathematical knowledge.

The researchers say the entire process took around 80 hours—something that would typically demand extensive collaboration among experts over a much longer period. “Using this framework, we successfully solved an open problem in commutative algebra and automatically formalised the proof with essentially no human intervention,” the paper states.

The team believes this marks a shift in how mathematical research could be conducted in the future. By automating complex, time-consuming steps, such systems could free mathematicians to focus on higher-level thinking while still ensuring the accuracy of proofs. Unlike traditional proof assistants that depend on constant human input, this dual-system approach worked independently, though the researchers noted that human involvement could still help speed things up.

This development joins a growing list of AI milestones in mathematics, including systems such as Google’s Gemini DeepThink, which won a gold medal at the 2025 International Mathematical Olympiad. However, most existing tools still rely heavily on human oversight—making this fully autonomous achievement stand out.

Disclaimer: This article is for informational purposes and highlights advancements in artificial intelligence within the field of mathematical research. While it discusses AI-driven breakthroughs, the content should not be used as a substitute for professional academic or technical guidance in commutative algebra or theorem proving.