I imagine in that line of work, SW must consist of like 99% safeguards and checks of various kinds. Like rather double-check the thing the compiler already proved at runtime than discovering a rare compiler bug in 10 years during a complex oribital descent maneuver…

Have to reboot every 136 years, literally unusable.

Yay, more bad programmers, exactly what we need! /s

I dont remember the numbers but iirc it was covered by one of the validation datasets and GPT 4 did quite well on it