1/ Stopped saying "write in my voice."

Based on what data? AI knows nothing about how you write.

That's like telling a new barber "make it look like me, but better." You know how that ends.

2/ Stopped editing AI output. Started rewriting it.

Editing = fixing AI's draft.

Rewriting = AI gave you scaffolding, you built the house.
Sounds subtle. Results aren't.

3/ Stopped using AI for first drafts.

Now I give it my ugly rough draft and ask:

"Where is this weak?"
"What am I avoiding saying?"
"Where would a skeptic stop reading?"

Better editor than writer.

4/ Stopped accepting the first output.

First pass = statistical average of everything AI has ever read. Competent. Forgettable. Wallpaper.

Third or fourth pass after specific corrections is where it starts sounding human.

5/ Stopped blaming AI for sounding generic.

Vague input = vague output. "Be conversational" means nothing.

"Short punchy sentences, then one longer thought, parenthetical asides for personality" means something.

6/ Stopped treating every piece the same way.

Some need heavy scaffolding. Some just need a sparring partner for one stubborn paragraph.

Matching collaboration level to the task saves hours.

7/ Stopped hiding that I use AI.

Not because transparency is trendy. Because the HOW is the interesting part.

Anyone can type a prompt. The system around it determines whether you get slop or something worth reading.

8/ The difference between AI delegation and AI collaboration:

Karaoke follows the track.
Jazz responds to what's happening in the room.

9/ 100 articles in. Still calibrating.

But the output is unrecognizable from where I started. And every piece still has my fingerprints on it.

That's the whole point. https://mastodon.social/@nickquick/116261543065524009