Hm. The question half-sounds as if they're asking "do you check AI output over and above your usual review, unit testing, CI, whatever?

If people interpreted it that way, 48% is not at all bad.

We live in a "good enough" era after all.