9/9
The last paragraph is you not trusting your own work.

Cut it.

The full piece (including why the AI training loop that causes this is kind of fascinating):

https://articles.cowritewithai.com/p/done-doesnt-know-its-done https://mastodon.social/@nickquick/116772363979752410

8/9
It almost always does.

Because the real ending was already there.

Second-to-last paragraph. Every time.

The one where the argument actually lands. The one where the reader already went: "Oh. Yeah. Got it."

You just didn't trust that moment enough to stop.

7/9
The fix takes 30 seconds.

Delete your last paragraph.

Read the piece again.

Ask: does this end stronger now?

6/9
Which means if you co-write with AI and don't catch this, you're not just over-explaining.

You're leaving the single most recognizable AI fingerprint in the piece right where everyone looks last.

You handed them a gift. Then taped the receipt to the front.

5/9
Now every AI output ends with some version of:

"Ultimately, X matters because Y."
"The key takeaway here is..."
"By applying these strategies, you can..."

The phrasing changes. The compulsion never does.

4/9
Here's where it gets worse.

AI does this too. Pathologically.

It was trained to. Human raters in early AI development marked responses "complete" when they ended with a tidy summary.

So every model learned: wrap-up equals finished. Get the treat.

3/9
That paragraph is not for the reader.

It's for the writer.

It's the exhale. The bow on top. The need to feel done.

The reader was already done. You just didn't know it.

2/9
Every piece has a moment where the argument lands.

The reader goes: oh. Yeah. Got it.

Most writers blow past it.

They add one more paragraph to make sure.