LinkedIn Says It's Cracking Down on AI Slop — I Read the News and Just Wanted to Laugh

Recently I came across a rather amusing piece of news: LinkedIn has started "getting serious" about cleaning up AI-generated junk content.

How? They took down the one-click generate-a-post button and added a new button specifically for users to report "this was written by AI."

Chief Product Officer Hari Srinivasan said it with great feeling: tackling AI spam is a top priority for all of us, and we truly care. People come to LinkedIn to connect with real humans and hear real opinions.

Really?

I stared at that sentence for a long time, wanting to ask just one thing: with which eye did you see users coming here for "realness"?

First, Get This Straight: Why Do People Actually Come to LinkedIn?

Let's skip the sentiment and talk business.

You open LinkedIn — what for? To meet people senior to you. Meet them for what? To sell something, or to sell yourself.

This isn't some dark cynicism; it's the business model. Nobody's here to make friends. In a place full of handshakes and mutual "Great insight!" likes, everyone pretends the conversation isn't about money — and everyone knows it's about money.

In a place like that, what's "realness" worth?

So LinkedIn grew into what it is today: a polite wasteland. Everyone speaking the same smooth, rounded, never-wrong corporate jargon. People call themselves "visionaries" and "thought leaders" — and not a single person laughs out loud.

Think about it: who writes this kind of content fastest, and writes the most of it?

AI, of course.

What AI does best is precisely this "correct nonsense." It never goes off topic, never takes risks, never surprises — like a highway so straight it's suffocating, delivering you from point A to point A-and-a-half in the most predictable way possible. Pair that with a recommendation system that measures value in jargon density, and—

A match made in heaven.

The Data Says So Too

There's a company whose name is its own irony: Originality.AI, which specializes in detecting whether text was written by AI. They checked 5,000 posts from July, and the conclusion: 81.2% were "possibly" written by AI.

"Possibly" was capitalized in the original. I paused when I read that: does capitalizing "possibly" make it more possible?

The science fiction writer Theodore Sturgeon had an old saying: ninety percent of everything is crap. Harlan Ellison added a stab from behind: Sturgeon was an optimist.

That ninety percent crap — it used to be people straining to write what they couldn't; now AI handles the whole batch. Consider: the vast majority of the English-speaking world can't write one clean simple sentence smoothly, let alone anything with a subordinate clause. Millions of "readable" posts piled up there — without AI, there's no way they'd ever have existed.

The tech blogger John Gruber once said something brutally honest that stuck with me: tools like ChatGPT, Claude, and Grok genuinely write better than most humans. But what is there to be proud of? Most people never knew how to write in the first place.

Exactly. The problem was never that AI writes too well.

It's that humans write too little, and too fake.

How Do You Prove a Post Was Written by a Human?

There's a wonderfully simple test for this: check whether it dares to go off topic.

What's going off topic? It's when, mid-piece, the writer suddenly swerves into an aside, or drops in a parenthesis with zero relation to the main thread. (Like this one. Did you just read it to the end without thinking? And it went down fine, didn't it?)

The human brain has a natural quirk, academically known as "pareidolia" — seeing a face in the clouds, a figure on the wall. When data runs short, imagination fills the gap. It sounds like a defect, but it's exactly the skill that let humans survive in a messy world — and the source of all inspiration. Peanut butter meets jelly? Barbie shot as a musical? Those "completely unrelated" random connections are something machines cannot make.

The price? Conspiracy theories.

But the reward is genius.

AI doesn't dare go off topic, because off-topic can't be mass-replicated. So if a post carries a bit of useless digression, a bit of living-human body heat, it's eighty percent likely to be human-written. Even if it's badly written, you won't question the species of its author.

So — Does LinkedIn Care?

It cares, but only a tiny little bit.

Their announcement is more like an imitation of "sincerity": making everyone feel that the platform sort of cares about its users. I'll grant this — they are somewhat worried that junk content will scare readers away.

But scared away to where, exactly?

That's where the confidence comes from. The professional-networking niche — for better or worse — has exactly one player. Users complain all they want, then show up the next day to post "humbled to announce."

So there you have it: on one hand, taking down the generate button for appearances; on the other, keeping a recommendation algorithm that rewards the smooth, polished, never-wrong corporate voice — which is exactly the thing AI produces best.

The mouth says no, while the system swallows it whole. That is LinkedIn's real attitude toward AI content.

Next time you see a platform righteously announcing a war on AI slop, don't be too quick to feel moved.

Go look at what its recommendation algorithm rewards.

The answer is all sitting right there.