Comment by nayroclade

So, the AI wrote a NL proof of Navier-Stokes, then incorrectly auto-formalised it to Lean, but still ended up with a verifiable proof of Navier-Stokes? That seems... strange?