10 AI Slop Examples That Prove Humans Are Still Funnier

I've been collecting AI slop like some people collect vintage vinyl. Except my collection makes me want to throw things. Here are ten specimens from the wild that prove, beyond any reasonable doubt, that humans are still funnier, weirder, and more interesting than anything a language model has ever produced.

1. The LinkedIn CEO-bro post that says absolutely nothing

You've seen this one. Ten bullet points. Each one a sentence fragment posing as wisdom. Something like:

I fired my best employee.
She cried.
I cried.
But then I realized: growth isn't comfortable.
That employee? She started her own company.
That company? It's worth $50M now.
The lesson? Sometimes you have to let people go so they can fly.
Agree?

The "Agree?" at the end is the chef's kiss. It's the AI equivalent of a dog tilting its head, hoping for a treat. Real LinkedIn is already insufferable, but AI-generated LinkedIn takes the format's worst instincts and removes the last trace of self-awareness. At least the human version of this guy believes his own nonsense. The AI version doesn't believe anything. It just patterns.

2. The recipe that lists "a cup of creativity" as an ingredient

An AI-generated cooking blog (and there are thousands of them now, clogging up Google results for anyone trying to make a decent carbonara) published a recipe that included, between "2 cloves garlic, minced" and "1/4 cup Parmesan," the instruction to add "a cup of creativity and a dash of love." The model apparently learned from all those cutesy food blogs that pad their word count with inspirational filler and decided the filler was part of the recipe. Reader, it is not. You cannot saut creativity. I've tried.

3. Google AI Overview: "Add glue to your pizza"

This one became genuinely famous. Google's AI Overview feature, which summarizes search results at the top of the page, told a user asking about cheese not sticking to pizza to "add about 1/8 cup of non-toxic glue to the sauce." It had picked this up from an old joke comment on Reddit. The AI couldn't tell the difference between a sincere recommendation and someone being a wise guy. Which is, when you think about it, the whole problem in miniature. AI can process text. It cannot understand when someone is messing around.

4. The AI portrait where earrings become part of the neck

AI-generated art has a well-documented problem with hands (too many fingers, impossible joint angles), but the earring-neck merge might be my favorite glitch. A popular AI art generator produced a "professional headshot" where the subject's hoop earrings gradually blended into the skin of her neck, like she was being slowly absorbed by her own jewelry. It looked like a Cronenberg film. The person who posted it had apparently used it as their actual LinkedIn photo for two weeks before someone told them.

5. The cold email complimenting your "impressive work at [COMPANY NAME]"

We've all gotten these. "Hi [FIRST NAME], I've been following your impressive work at [COMPANY NAME] and I'd love to connect about synergies in the [INDUSTRY] space." The brackets are the tell, but even without them, the email reads like a hostage note written by someone who's never met another human. The best one I've seen forgot to fill in any of the brackets and was sent to a freelancer who doesn't have a company. They responded with "Thanks! [COMPANY NAME] is doing great, we just closed a [DOLLAR AMOUNT] round."

6. The news article citing a study that doesn't exist

An AI-generated news piece about the health benefits of a specific supplement cited a "2024 study published in The Lancet" that, upon investigation, does not exist. Not retracted. Not paywalled. Nonexistent. The AI fabricated the journal, the authors, and the findings. It hallucinated an entire peer review process. Someone screenshot the article and the actual Lancet's search results side by side, and it went viral. The kicker? The article was on the first page of Google results for weeks before being flagged.

This isn't funny. This is actually dangerous. But it's on this list because the sheer audacity of inventing a Lancet study is so brazen it loops back around to absurd.

7. The Amazon listing with contradicting specs

A product listing for a Bluetooth speaker claimed in the title it was "waterproof IPX7," stated in the bullet points it was "splash resistant only — do not submerge," and then in the product description said "perfect for underwater use in pools and oceans." Three different water resistance claims on one page. The AI-generated copy apparently pulled from multiple source materials and just... combined them, contradictions be damned. At least a human copywriter would have picked one lie and committed to it.

8. The AI children's book with nightmare illustrations

Someone used an AI image generator to create a children's book about a friendly dragon. The dragon had human teeth. Full, adult human teeth, in a grinning reptilian mouth. One illustration showed the dragon "hugging" a child in a way that looked more like the cover of a true crime podcast. Another page had the dragon with seven toes on one foot and three on the other, which honestly is the least disturbing thing about it. The book was briefly available on Amazon before reviews comparing it to "Junji Ito for preschoolers" got it pulled.

9. The Reddit comment that starts with "Great question!"

Nobody on Reddit has ever, in the history of the platform, started a genuine comment with "Great question!" That's a teacher thing. That's a corporate webinar thing. It is not a Reddit thing. Reddit comments start with "Actually," or "This," or just a link with no context. When you see "Great question! There are several factors to consider here..." you are reading AI slop, and the fact that it's being upvoted means either bots are voting or people have stopped reading past the first sentence. Both options are depressing.

The structure is always identical: affirmation, followed by a numbered list, followed by a gentle disclaimer about how "this is a complex topic." Real Reddit users have never in their lives described anything as "a complex topic." They have strong opinions and they share them with the confidence of someone who just read one Wikipedia article.

10. The travel blog reviewing a restaurant that closed three years ago

An AI-generated travel guide to Lisbon enthusiastically recommended a specific restaurant's "perfectly seared bacalhau" and "intimate courtyard seating." The restaurant had closed in 2023. The building is now a phone repair shop. The AI had been trained on old review data and presented it as current. Someone actually went to the address based on the blog post and took a photo of themselves standing in front of the phone repair shop holding a screenshot of the AI's glowing review. That photo is better content than anything AI has ever produced.

Why this matters

Every single one of these examples fails for the same reason: the AI doesn't know what it's talking about. It doesn't know that glue isn't food. It doesn't know that earrings don't grow out of necks. It doesn't know that restaurants close. It doesn't know anything. It generates plausible text based on patterns, and when the patterns conflict with reality, the patterns win.

Humans are funnier because we actually live in the world. We know what pizza tastes like. We've been to Lisbon. We've gotten terrible cold emails and rolled our eyes. Our humor comes from real experience, real frustration, real absurdity that we witnessed with our own eyes.

Want to experience this for yourself? Play Your AI Slop Bores Me and watch real humans try to sound like AI — and fail in ways that are infinitely more entertaining than any of the examples above. The game works because human weirdness is impossible to fake, even when you're trying to fake being fake.

AI slop is everywhere. But so are the people making fun of it. And honestly? The people making fun of it are producing better content.