Terence Tao predicts an explosion of experimental mathematics as AI maps math difficulty
haider1 · x · 2026-11-04
Mathematician Terence Tao, in his UCLA "Math 2.0" talk, predicts an imminent "explosion of experimental mathematics": math has been almost purely theoretical because proving on paper was easier than collecting millions of data points, and soon there will be a map of which math fields are hard or easy.
- Tao stresses more proofs don't mean more understanding; better questions, deeper insight, and open infrastructure are needed.
- The quoted post also showcases Open Math Model, a platform built on open frontier models (Kimi K3, DeepSeek, GLM, etc.) with parallel agent exploration, Lean formal proofs, LaTeX collaboration, and 100+ preinstalled scientific computing tools, with data sovereignty held by the math community.
- Demo: an agent verified that the sum of the first n odd numbers is n² via Python checks, a Lean induction proof, and an auto-written proposition.
More from coding & agent
- Ubuntu 26.04's Rust sudo isn't compatible with standard sudo — lemire · 2026-11-04
- Secretless private vector search for Foundry Agents via MCP and Cosmos DB — adnan_hashmi · 2026-11-04
- Monomem: open-source MCP memory server with provenance stamps and injection linter — Mother-Side6477 · 2026-11-04
- Two OpenAI employees built a Codex-talking device in under a week with GPT-6 Astra — OpenAI · 2026-11-04
- OpenAI works with Chip Ganassi Racing and Mazda on AI-driven automotive design — OpenAI · 2026-11-04
- OpenAI brings ChatGPT and Codex to hardware engineering, partners with Chip Ganassi Racing and Mazda — OpenAI · 2026-11-04