Archive — US — techcrunch
AI models are starting to crack high-level math problems
AI models have demonstrated significant advancements in solving high-level math problems, particularly through OpenAI's GPT 5.2, which has successfully tackled open conjectures by mathematicians like Erdős. The use of AI in formalizing proofs is evolving, with notable contributions from mathematicians and increasing reliance on AI tools for advancing mathematical knowledge.
Key points
- Neel Somani tested the math skills of OpenAI’s new model and found it could solve complex problems in 15 minutes.
- ChatGPT provided a complete solution to a problem by Paul Erdős, differing from previous solutions by offering a more complete perspective.
- Since the release of GPT 5.2, there has been an increase in the number of problems solved, 11 of which were credited to AI on the Erdős website.
- Mathematician Terence Tao noted AI's meaningful contributions to solving Erdős problems, particularly easier ones.
- Formalization of mathematical proofs is becoming easier with tools like Lean and Harmonic’s Aristotle, enhancing AI's role in mathematics.
Why it matters
- The rise of AI in mathematics reflects broader trends in AI applications across various fields, such as automation and data analysis.
- AI's ability to tackle complex puzzles mirrors current advancements in AI technologies, aligning with 2025 trends emphasizing AI-assisted research and automation.
- Historical reliance on computational methods in academia has paved the way for current AI applications, promoting a shift towards more collaborative human-AI problem-solving.
AI's capability to solve high-level math problems marks a pivotal moment in mathematics, suggesting a future where AI plays a significant role in research, pushing the boundaries of human understanding and formalizing complex proofs.