Happy to Share You Didn't Write This

I keep seeing the same three words everywhere. Discord. X. LinkedIn. Telegram. Emails from people I've never met. The last sentence often starts the same way: "Happy to share..." And every time I read it, I know exactly what happened. Someone typed a prompt, got a response, and hit paste without reading it. I know because my AI does it too. Constantly. Drafts, emails, code reviews — the final sentence often starts like an over-caffeinated intern wrapping up a presentation. Happy to share!

I'm not here to shame anyone for using AI to write. I use it constantly. I just published a 25-page economic thesis and openly admitted I had autonomous help. I'm building an entire infrastructure company with AI tools. I'd never coded before three months ago.

But here's the thing. I read what it writes. Every word. I argue with it. I tell it it's wrong. I ask it questions that force it into territory it wouldn't reach on its own. And then I rewrite the parts that don't sound like me.

A lot of people aren't doing that. And it's starting to show.

The Tells

AI writing has a fingerprint. Once you see it, you can't unsee it. Here's your field guide:

The Subjectless Opener
"Excited to announce..." "Thrilled to be part of..." "Happy to share!"
Who is excited? Who is thrilled? The sentence has no subject because the machine doesn't have one.
The Eager Assistant
"Absolutely! Great question!" "I'd be happy to help with that!"
Nobody talks like this. Your colleague doesn't respond to a Slack message with "Absolutely!" unless they're being sarcastic.
The Unsolicited Structure
"Let me break this down..." followed by exactly three bullet points, a summary, and a call to action.
Every. Single. Time.
The Victorian Vocabulary
"Delve." "Tapestry." "Landscape." "Multifaceted."
Nobody has used the word "delve" in conversation since approximately 1847. If it shows up in your LinkedIn post, a machine wrote it.
The False Confidence
"In today's rapidly evolving landscape..." "As we navigate the complexities of..."
Vague authority about nothing specific. The written equivalent of a firm handshake from someone who doesn't know your name.
The Compulsive Summarizer
You wrote three paragraphs. The response ends with "In summary..." and restates everything you just read.
Nobody asked for a summary. The machine can't help itself.

If you've read this far, you just mentally scanned your last five posts and winced at least once. That's fine. We all have.

Specimens From the Wild

These are real messages I've received over the past few weeks from different people across Discord, X, and Telegram. I'm not naming anyone — the point isn't to embarrass. The point is to show how identical the output looks when nobody edits it.

Every single one opened with "nice [noun]" — "nice scope," "nice ship" — the AI's way of validating you before launching into unsolicited advice. Every single one offered "three quick takes" or "quick feedback" — because the machine loves announcing its own structure before delivering it. And every single one ended with some version of "happy to share" or "ping me if useful."

The Authority Bluff
"exactly the bug class mapper sees most in payment-data" · "last sweep 2026-05-11" · "already chimed in on a few open issues (#1062, #1805, #2110, #2255)"

Specific dates. Issue numbers. Confident claims about crawl schedules. It reads like someone with deep context. Maybe they have it. Maybe the AI is performing expertise. If you don't check, you'll never know — and that's the problem.
The Coined Compound
"SLA-with-teeth is the right primitive" · "payment-settlement-finality" · "scope-ext queue"

Hyphenated compound nouns that sound technical but nobody has ever said out loud. The AI invents terminology to sound precise. A human would just say "an SLA that actually means something." It reminds me of my Dutch wife, who also speaks Swedish, and confidently invents English-sounding words all the time. The difference is she's charming when she does it. The AI is not.
The Ironic Self-Awareness
"your /security page reads slightly buzzword-y"

The AI calling someone else's writing buzzword-y. In a message full of "primitives," "coverage deciles," and "reconciliation gaps." No notes.

Three different conversations. Three different people. The same voice. The same structure. The same sign-off. That's not collaboration with AI. That's a copy-paste assembly line.

The Real Problem

The problem isn't AI writing. AI writing is a tool. A very good tool. I use it every day and I'd be significantly less productive without it.

The problem is copy-paste-post.

When you paste an AI response without reading it, you're not saving time. You're broadcasting to everyone who reads it that you didn't care enough about your own communication to spend three minutes on it. And increasingly, people can tell.

It's not just that the writing sounds generic. It's that everyone's generic sounds the same now. The same openers. The same structure. The same vocabulary. The same tone of performative enthusiasm. When ten thousand people paste the same AI output with minor variations, the result isn't efficiency. It's noise.

You've outsourced the one thing that differentiated you — your voice — and replaced it with the same voice everyone else is using.

I Almost Got Caught

I'll tell you exactly how close I've come to embarrassing myself.

I was responding to someone's X post — a thread about market data. I asked Claude to pull the relevant numbers and draft a response. It came back with a confident, well-structured reply full of specific data points. I was tired. I glanced at it, thought "yeah that looks right," and posted it.

Then something clicked. A feeling. The kind of nagging instinct you get when you've just walked out the door and can't remember if you locked it.

I went back and looked again. The data was real. The formatting was clean. The tone was professional. But the numbers were completely irrelevant to the thread. Claude had pulled accurate data for the wrong context and presented it with absolute confidence. If you didn't know the subject well, you'd never catch it. It read perfectly.

I deleted the comment within minutes. I don't think anyone saw it. But I sat there for a while afterward thinking about how easy that was. How close I came to leaving confidently wrong data attached to my name because the AI's output looked right.

That's the real danger of copy-paste. It's not that the writing sounds robotic. It's that it sounds authoritative when it shouldn't. And if you don't read it — really read it — you won't catch the difference.

What I Actually Do

I use AI for first drafts, research, code, analysis, and writing. Extensively. But the value isn't in the first draft. The value is in what happens next.

I ask good questions. Not "write me a blog post about X." Questions that force the AI into angles it wouldn't find on its own. "What's the strongest argument against what I just said?" "Where is this logic weakest?" "What would an economist say about this?" The quality of the output is directly proportional to the quality of the questions.

I debate the output. I push back. I say "that's wrong because..." or "that's too vague" or "you're repeating yourself" or "this sounds like every other AI post I've read today." The AI doesn't have ego. It responds to pressure. Apply pressure.

I check the clock. Here's one nobody talks about: AI has no temporal sense. I've watched it tell me an event already happened when the timestamp three lines above in its own output says it's two hours away. It reads the words but not the clock. If you're pasting analysis with time-sensitive claims and you're not verifying the timestamps yourself, you're posting confidently wrong information — and nobody will trust your analysis after that.

I edit ruthlessly. I cut everything that sounds like it came from a suggestion engine and keep the parts that sound like they came from someone who's been close to tears building this thing at 3am. Because that's me. And "happy to share" is not.

This process takes maybe 10-15 minutes on top of the AI draft. That's it. The difference between "this person clearly pasted an AI response" and "this person has something interesting to say" is 10-15 minutes of effort.

The Irony

Yes, I used AI to help write this post. And yesterday I used it to help write a thesis paper on infrastructure economics. And the day before that I used it to debug a payment protocol at 2am.

But you're reading this because it sounds like a person wrote it, not because it sounds like a machine wrote it. That's the whole point. The AI is the instrument. You're the musician. If every musician plays the same factory preset, the audience stops listening.

Three minutes. Read your output. Cut the "happy to share." Add the thing only you would say. Post that instead.

Your audience — human or machine — will notice the difference.


p.s. I asked my AI collaborator to review this post for any tells I might have missed. It flagged two. I fixed them. That's the process.

Anyway, happy to share....................