AI Started Solving Open Math Problems

📖 Resources & more for this episode: https://www.unrsnbl.ai/thoughts/t001-acceleration-in-ai-solving-erd-s-problems "It was January 2026 and AI started solving Open Math problems which no human was able to resolve..." This will be written in the books of the future while we live in this exact moment from these books. In mathematics, there are so-called open problems - math problems that no human has ever solved. Some of these Erdős problems have been open for nearly 100 years. In December 2025 and January 2026, something unprecedented happened. AI tools like ChatGPT 5.2 Pro, Aristotle, and AlphaProof started solving Erdős problems — not just checking answers, but finding proofs, finding counter-examples, and writing formal mathematics. One important example is Erdős Problem 728. According to Terence Tao, on January 6, 2026, this problem was solved almost autonomously by AI. The result was not previously known in the literature — AI found a completely new proof. Shortly after, on January 10th and 11th, Problems 729 and 397 were solved by ChatGPT 5.2 Pro, with proofs accepted by Terence Tao. This is not about AI replacing mathematicians. This is about the speed of progress. Mathematics is the foundation of science and technology. When difficult problems are solved faster, everything built on top can move faster as well. This is what people mean when they talk about acceleration. We're entering a completely new phase — a phase of faster discovery. And I think it's only the beginning. Explore this yourself: • Erdős Problems Website: https://www.erdosproblems.com/ • GitHub AI Contributions: https://github.com/teorth/erdosproblems/wiki/AI-contributions-to-Erd%C5%91s-problems ▶️ Full playlist Thoughts on AI: https://www.youtube.com/playlist?list=PL3pL28ov_GlKFzXuxVR7tCzACLidg5qGL 📦 Join us in Telegram: https://t.me/unreasonableai #ai #shorts #thoughtsonai #genai #mathematics #erdosproblems #CTHuman