We use Haskell so that 99% of the time, the compiler can prove our program safe. Then, the other 1% of the time, we use our human intuition to decide to "gamble" and disagree with the compiler.

In most languages, 100% of your code is that same gamble.