# Navier-Stokes equation

Using OpenAI's September 8 announcement of a finite-time blow-up proof for the forced Navier-Stokes equation as an entry point, Tapio Schneider distinguishes episteme (explanatory understanding) from techne (the craft of prediction), and argues that when predictions must be trusted before they can be empirically verified—as in decadal climate projection or the design of a novel aircraft—trust comes from an auditable causal chain running from assumptions and input data to outcomes, each link of which can be tested individually; AI should therefore be embedded in auditable scaffolds such as physical conservation laws and used to learn closure models that can be checked against high-resolution simulations, observations, or experiments, rather than deployed as end-to-end models.

Using OpenAI's September 8 announcement of a finite-time blow-up proof for the forced Navier-Stokes equation as an entry point, Tapio Schneider distinguishes episteme (explanatory understanding) from techne (the craft of prediction), and argues that when predictions must be trusted before they can be empirically verified—as in decadal climate projection or the design of a novel aircraft—trust comes from an auditable causal chain running from assumptions and input data to outcomes, each link of which can be tested individually; AI should therefore be embedded in auditable scaffolds such as physical conservation laws and used to learn closure models that can be checked against high-resolution simulations, observations, or experiments, rather than deployed as end-to-end models.