A mentor once told me “you have my trust, but break it and you will need to prove yourself six times over to win it back”.
Its a good guide to live by, at least at work. Life is difficult and not very much fun if you don’t trust anyone.
Most often in a work context I would have to apply this rule when people let you down. Ask for help on something time critical and they don’t deliver.
These days the rule is seeing new light in the fight against AI slop.
I hate editing AI slop. It usually goes like this:
Someone sends you a draft of a manuscript, meeting notes or other internal doc to edit. You start editing it. Partway through you realise your editing a lost cause, vapid, AI written sentences that look polished by say nothing.
I start deleting whole chunks of text. Then I give up. I’ve just wasted 30 minutes or more of your precious time fixing someone else’s slop.
Every time that happens I feel like a little piece of my soul dies.
If you don’t think its possible to separate AI from human writing, check out this recent pre-print. This analysis of over 60,000 human and AI written stories shows quantitatively that AI stories inhabit a unique set of narrative and style choices. They show you can pick AI written fiction with about 90% accuracy.
I’m not against AI augmented writing. But there’s a fine line between having AI improve your writing versus letting it do too much and deviating into slop.
I see too many cases of people being too rushed, or busy or just lazy and overusing AI to do their hard work. Then passing that off to collaborators to fix.
So all new collaborators, colleagues and students get my trust. I’ll edit their work. If the trust is broken though, I’ll need to read six new revisions before I’ll engage with editing again.