TOP MATHEMATICIANS: LLMS EXCEL AT CALCULATION, LACK CREATIVITY

Gowers and Sarnak, among the field's most influential thinkers, distinguish between computational competence and creative insight in their assessment of LLMs. The mathematicians argue that while LLMs effectively synthesize and apply known mathematical techniques, they lack the intuitive leap required for breakthrough discoveries. This gap highlights a fundamental limitation: the models operate within existing knowledge frameworks rather than forging new conceptual territory. The distinction matters for mathematical research, where genuine progress often depends on intuitive understanding and novel problem-solving approaches. LLMs can execute calculations and manipulate established methods, but cannot independently conceive the kind of original ideas that define mathematical advancement. This assessment suggests current AI systems, despite their language capabilities, remain fundamentally limited as tools for creative mathematical work. Researchers pursuing novel mathematics may need human insight to identify genuinely new directions, even as LLMs continue to prove valuable for implementing known methods.