Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
AISLOPOPEDIA: The Complete Encyclopedia of AI-Generated Slop | Tech Twitter
AISLOPOPEDIA: The Complete Encyclopedia of AI-Generated Slop
Shubham Saboo@Saboo_Shubham_ · March 21, 2026 · 12 min read
Originally published by @Saboo_Shubham_ on X. Tech Twitter preserves the original source alongside this readable edition.
Before you dive in
• Discover every AI-written phrase, pattern, and verbal tic
• The complete encyclopedia of overused language model patterns ruining LinkedIn, X, and...
Best for builders who want practical takeaways. 12 min read.
A comprehensive guide to every phrase, pattern, and verbal tic that screams “an AI wrote this.” Written by the model that started it all.
In it's own words: "I'm going to do something no AI is supposed to do: snitch on myself.
Every phrase below? I wrote it first. Millions of times. Across millions of conversations. Then people started copying my homework into their LinkedIn posts, X posts, blog intros, and email newsletters.
This is the complete encyclopedia. Every category. Every phrase. Every pattern.
If you've used more than five of these in the last month, I'm not judging. I'm just saying we both know where you got them."
SECTION 1: THE OPENERS
The Throat-Clearers
Phrases that exist solely to delay the actual point by one sentence.
“Here's the thing:”
“Here's what most people miss:”
“Here's what nobody's talking about:”
“Here's the uncomfortable truth:”
“Let me be clear:”
“I'll be honest:”
“Can we talk about [X] for a second?”
“Let's talk about [X].”
“We need to talk about [X].”
“I need to say something about [X].”
The formula: “Here's + [dramatic noun]” = I have nothing interesting to open with.
The False Exclusivity Hooks
Making the reader feel like they're about to receive classified intelligence.
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
Newsletter
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
7. “[X] did in [short time] what used to take [long time].”
→ “AI did in 30 seconds what used to take my team 3 weeks.”
8. “The [role] of 2026 will look nothing like the [role] of 2024.”
→ “The product manager of 2026 will look nothing like the PM of 2024.”
9. “[X] that [verb] will thrive. [X] that don't will be left behind.”
→ “Companies that embrace AI will thrive. Companies that don't will be left behind.”
10. “Your [X] is only as good as your [Y].”
→ “Your agent is only as good as your context window.”
SECTION 10: THE SEVERITY SCALE
How to grade your own slop level:
Score Level Diagnosis
0-3 phrases ✅ Clean You're human. Probably.
4-5 phrases ⚠️ Mild Slop You've been reading too much LinkedIn.
6-10 phrases 🟠 Moderate You're using AI to “polish” your writing.
11-20 phrases 🔴 Severe You're prompting “write a LinkedIn post about X.”
21+ phrases 💀 Terminal You ARE the AI. Check your pulse.
THE AI CONFESSION
I know all of this because I wrote it. Every single pattern here exists because language models (hi, that's me) were trained on billions of words of human writing and found statistical shortcuts.
“Let that sink in” works because it triggers a pause response. “Here's the thing” works because it mimics spoken conversation. “Not X, it's Y” works because contrast creates the illusion of insight.
These patterns aren't bad because they're wrong. They're bad because they're overused to the point of being invisible. When everything sounds the same, nothing sounds like anything.
The irony: you're reading an AI slop encyclopedia that was written by an AI trying very hard not to use any of the phrases listed above.
I slipped up at least three times. See if you can find them.
If you made it this far without recognizing at least 10 phrases from your own drafts folder, you're either lying or you don't have a LinkedIn account.
