Tao-led 25 Fields medalists warn against reducing math to AI benchmarks; daily roundup
APPSO · wechat · 2026-09-12
APPSO's daily roundup is headlined by a joint statement led by Terence Tao and 25 Fields medalists warning that AI labs' use of famous math problems as benchmarks is misaligned with mathematics: the field values understanding and knowledge transfer, and models outpacing human comprehension could disrupt it. Context: OpenAI claimed an internal model solved Navier-Stokes at $M compute costs (1000x prior work), pending verification.
Other items:
- Microsoft plans to grow datacenter capacity from 12GW to 38GW+ by 2032
- Zhipu kicks off $5B refinancing ($2B share placement + $3B convertible bonds)
- Cognition trains SWE-2 on Kimi K3 at a quarter of Astra's cost
- OpenAI to retire GPT-5.3-Codex-Spark; Oracle cloud revenue +62% YoY
- CXMT posts 82% operating margin, 10% global DRAM share
- Enflame debuts on STAR Market, up 188%, market cap over 170B yuan
- Alipay embeds payments into Goertei's robot dog; SenseTime's Xiaohongxiong goes mobile
- Cursor launches Projects with coordinator agents for long-running dev work
- Apple execs explain the iPhone Duo foldable design choices; Tesla Model Y Performance launches at 369,000 yuan
More from coding & agent
- HABIT-Bench: Best agent memory system scores just 33.61% at inferring user habits — SJTU · 2026-10-28
- Regehr's 'Piano Test': would you trust formally verified code with your life? — moyix · 2026-10-28
- Vibe-coded 'AI 浮生记' game satirizes AI startup life with compute black-market trading — vista8 · 2026-10-28
- The legendary fake Go generics: Unicode Canadian Syllabics as angle brackets — rickasaurus · 2026-10-28
- Microsoft Agensh board: a coding agent's "tests passed" is just a message — shoumikgoswami · 2026-10-28
- Ollama and FriendliAI to Host Seoul Meetup on Building Production Agents with Open Models — ollama · 2026-10-28