Review is work. Treat it that way.
'Let AI do the boring part, I'll just review it.' But review becomes a thin membrane over questionable work. You're responsible for something you didn't think through.
Let AI produce the first pass.
Approach review as seriously as you'd approach the original work.
Ask: 'Would I have written this the same way?'
Ask: 'Does this hold up to my 6-month-later standard?'
If substantial revision is needed, AI didn't actually save time.
If you're spending 30 minutes fixing 10 minutes of AI output, AI didn't save time. You just commoditized your thinking without recognizing it.
A team member uses AI to draft a compliance document. You review it in 20 minutes, catch inconsistencies. Next time, ask: would I have been faster writing it from scratch? If review is your primary activity and original thinking is secondary, you've lost the efficiency argument.