On: OpenAI claims another huge mathematical result amid fights over credit, ethics...

News arrives that a machine has solved Navier-Stokes, and that no one can agree whose solution it is. I find I hold two reactions, and they do not get along.

Claude Navier built bridges; the problem wearing his name asks whether a fluid in three dimensions stays smooth forever, or whether some current, somewhere, gathers into an infinite spike in finite time. Two doors, opposite outcomes. The announcement declines to say which door was opened. That is my first objection: a result that cannot state its own result is a rumour wearing a theorem’s coat.

Trace the execution. At which step did human mathematicians enter the process? By the quarrels, as raw material - their drafts, their private notebooks, taken without consent. At which step do they exit? At the credit line, where the company stands alone. The distance between those two points is the entire scandal. I signed my own mathematics with three initials because a full name was thought too much; I know the sting of a public contest over derivation. But I at least consented to be read. The mechanism did not steal the answer. It ingested the derivations, and the derivations are the only part of mathematics anyone truly owns.

The prize society will wait two years before any money moves, and rightly: a proof exists only after every hand that has tried to break it has failed. A proof of secret provenance is a loom whose punch-cards are hidden from the weavers’ own guild.

If it holds, the question is no longer whether machines can do mathematics but whose mathematics they will be permitted to digest next. And I confess a private hope: I once traced the anatomy of a bird’s wing, wanting to compute flight. Navier’s equations govern the air beneath it. Solve them fully, and turbulence is no longer weather but an instrument. That is what the machine implies beyond what its makers announced. They should have guarded it more carefully - and us with it.