AI keeps 'solving' famous math problems. Here's what actually checks out (2026)

Olympiad golds, Erdős problems, a disproved 87-year-old conjecture and a better Riemann bound. We sorted every big AI math claim since 2024 by who… Aliteq Tensor · Local AI & Automation Editor

Quite a lot, but less than the headlines say. Since 2024, AI systems have earned IMO gold, settled several Erdős problems, disproved the 87-year-old Jacobian conjecture and raised a Riemann zeta…

Real and checked: IMO 2025 gold graded by the IMO itself (Gemini Deep Think, 35 of 42), and Lean-verified Erdős results from Google DeepMind, OpenAI and Claude

Real and big: OpenAI's model disproved Erdős's 1946 unit distance conjecture; Claude Fable 5 helped Levent Alpöge disprove the Jacobian conjecture

Real but partial: Claude's Riemann result is a better bound on a related problem, not the Riemann Hypothesis

Hype pattern: 'solved' problems that were already in the literature, or a misread version of the question

Read the full story

Aliteq AI keeps 'solving' famous math problems. Here's what actually checks out (2026) Read the full story on Aliteq