My chemical engineering curriculum at the University of Minnesota included a course on fluid mechanics. All I remember about this tool was that it involved very intense math, including a prominent role for imaginary numbers—stemming from taking the square root of negative 1. I thought that was very cool. The class discussed the equations of Navier-Stokes, but they featured too many partial derivatives for my taste. However, I learned enough to know their value to understanding fluid dynamics. Thus, I can appreciate the achievement by OpenAI to solve the Navier-Stokes problem of the equations breaking down. If you haven’t been tracking this news, I recommend you read this Science briefing on OpenAI’s breakthrough, why it’s important, and how it ignited a controversy .

But there’s one thing about OpenAI’s solution to the Navier-Stokes problem that greatly worries me.  It allows for the laws of physics breaking down under certain extreme conditions, enabling water to spontaneously explode.

“There seems to be nothing in those equations that prevents a wayward eddy from suddenly turning in on itself, tightening into an angry gyre, until the density of the energy at its core becomes infinite: a catastrophic ‘‘singularity.’’ Someone tossing a penny into the fountain by the faculty center or skipping a stone at the Santa Monica beach could apparently set off a chain reaction that would take out Southern California.”

– University of California mathematics professor Terence Tao quoted in an article about him by New York Times Magazine, 7/24/25

Oh, oh!