# Free AI Chatbot Solves Ten-Year Mathematics Problem in 13 Minutes
Matthew Sparkes, a reporter at New Scientist, tested a free artificial intelligence chatbot against an unsolved mathematics problem that had resisted solution for a decade. The system solved it in 13 minutes.
Sparkes has been covering the rapid advancement of AI in mathematical problem-solving over recent months. His decision to benchmark a freely available chatbot against a genuinely difficult, long-standing problem yielded unexpected results. The speed of the solution surprised even someone tracking the field closely.
The experiment underscores a broader shift happening in mathematics and computer science. Large language models trained on vast datasets have begun handling complex mathematical reasoning at speeds that outpace human mathematicians working on the same problems. This development arrives as competition intensifies between AI labs to build systems capable of higher-order reasoning.
The specific problem Sparkes selected had eluded mathematicians for roughly a decade. Rather than remaining unsolved indefinitely, the chatbot processed the challenge and delivered a solution in under 15 minutes. The nature of the problem itself matters here. Mathematics problems vary widely in difficulty and type. Some involve pure numerical computation, others require creative logical leaps or the synthesis of multiple mathematical fields. The fact that a free tool succeeded suggests AI has reached a threshold where it can handle non-trivial reasoning tasks without specialized hardware or subscription fees.
Free chatbots represent the democratized frontier of AI mathematics. When cutting-edge systems remain behind paywalls or limited access, their capabilities stay invisible to most researchers and students. Open or free versions allow rapid deployment and testing across wider populations, potentially accelerating discovery cycles. Sparkes' experiment demonstrates that barriers to solving difficult problems may have genuinely lowered.
This development carries implications for professional mathematics. Academic mathematicians spend years on problems. Graduate students build careers around solving difficult open questions. If free AI systems can handle problems at this level of complexity, the role of human mathematicians shifts. Rather than grinding through difficult computation or logical deduction, mathematicians might focus on problem formulation, verification, and the deeper conceptual work that machines still struggle with.
However, speed alone does not guarantee correctness. The article does not specify whether Sparkes verified the chatbot's solution independently or whether other mathematicians confirmed its validity. In mathematics, a proposed solution requires rigorous proof and peer review. An AI system delivering an answer quickly differs from a solution the mathematical community accepts as correct.
The chatbot used remains unnamed in the reporting. Knowing which system succeeded matters for understanding AI capabilities. Different models have different strengths. A solution from a cutting-edge paid service carries different weight than a solution from a free, open-source system available to anyone. The specifics determine whether this represents a breakthrough or a clever demonstration of existing capabilities.
Sparkes' experiment points toward a future where AI assistance becomes standard in mathematical research. Whether this accelerates genuine discovery or merely provides faster wrong answers depends on how researchers integrate these tools into existing workflows and verification processes.