If you ask me what's the opposite of "proving", I'd say it's "testing". You call that "weaker"? It's the weakest form of validation I know of.
It may be weak, but by hell it's practical. Tons of tooling, tons of engineers who understand it, and with a pragmatic approach, eliminates many bugs and regressions.
And we organize our programs such as to get nice properties.