Mathematicians working at a London event have used artificial intelligence to accelerate the formal verification of Fermat's Last Theorem, one of mathematics' most famous problems. The effort represents a striking convergence between classical mathematics and modern machine learning.
Fermat's Last Theorem states that no three positive integers x, y, and z can satisfy the equation xn + yn = zn for any integer value of n greater than 2. Pierre de Fermat proposed this in 1637, but a proof eluded mathematicians for 358 years until Andrew Wiles finally proved it in 1995 using advanced algebraic geometry and modular forms. Wiles' proof spans over 100 pages and relies on deep results from multiple mathematical domains.
Formalizing a proof means converting it into a version that a computer can verify step-by-step, checking every logical inference against strict formal rules. This process eliminates ambiguity and guarantees correctness but demands enormous effort. The Wiles proof's complexity made full formalization seem computationally prohibitive.
The London team leveraged AI systems trained on mathematical knowledge to help bridge gaps in formal reasoning and suggest proof strategies. This AI assistance reduced the time and human labor required to translate Wiles' conceptual arguments into machine-verifiable code. The researchers achieved progress faster than previously anticipated, demonstrating that AI can handle intricate mathematical reasoning tasks.
The work builds on earlier efforts to formalize major theorems. In 2021, researchers formalized the prime number theorem using the Lean proof assistant. The success with Fermat's Last Theorem suggests AI could accelerate formalization of other landmark mathematical results.
Limitations remain. The AI required human mathematicians to guide its reasoning and validate its suggestions. The formal proof still represents a substantial translation project from Wiles' original work. Additionally, formalizing mathematics doesn