"Make no mistakes" works about as well for humans as it does for AI, and we have decades of lived experiences to draw on for that verdict.
"Make no mistakes" works about as well for humans as it does for AI, and we have decades of lived experiences to draw on for that verdict.
Go look at a nineteenth century mechanical signal box. When you, through inattention or even malevolence, try to release signals so that train B can plough into train A, the lever won't move because it is physically impossible due to mechanical interlocking, you need to move a piece of sold metal through another piece of solid metal and it won't go.
Do not rely on humans to do boring but important things, they are very bad at this.
There is a catch, however. Some languages will not allow you to take some shortcuts you know are safe unless you can prove them safe.
It isn't like C has some "there be dragons" set of features that are hard to use. The source of exploitable memory safety issues is consistently just foundational language features.