Remember when the robots were supposed to take our jobs?
Turns out they're starting with our lunch.
Google's AI Overviews—that cheerful little box of answers at the top of your search results—recently informed the internet that you should add "about ⅛ cup of non-toxic glue" to your pizza sauce to keep the cheese from sliding off.
This was a trillion-dollar company's flagship product confidently reading a decade-old Reddit shitpost as established culinary fact.
For the uninitiated, the glue tip originated from a 2011 comment by a user named fucksmith, who was clearly trolling.
Google's AI scraped it, stripped the context, and served it up like a Paula Deen recipe.
Somewhere, a product manager is having a very bad week, and honestly?
Here's the thing nobody wants to admit: this isn't a bug, it's the entire business model.
It knows what's *statistically likely to appear near other words*.
When you train a model on the whole internet and then ask it questions, you're essentially asking a very well-read toddler to explain quantum physics.
The pizza glue incident is just the appetizer.
Users have since gotten AI to recommend eating rocks for minerals, putting gasoline in your coffee, and—my personal favorite—jumping off a bridge if you're feeling sad.
Google has been quietly scrubbing these results and adding disclaimers, which is the corporate equivalent of putting a Band-Aid on a gunshot wound and calling it a wellness retreat.
But let's zoom out, because the real story isn't that AI is dumb.
We've spent two years breathlessly integrating chatbots into our search engines, our email, our customer service, and our children's homework, all while the technology hallucinates facts like a guy three beers deep at a family reunion.
And when it screws up, we shrug and say "well, it's new." New?
My dude, the Wright brothers' first flight was 12 seconds and nobody asked them to land a 747 the next week.
It's how fast we normalized trusting a system that has no idea what it's saying.
Now we're copy-pasting AI answers into work emails and hoping for the best.
The pizza thing is funny because it's absurd.
It stops being funny when the same tech is summarizing medical studies, legal precedents, or election results.
Suddenly "⅛ cup of glue" becomes "take this dosage" or "this law says you can't vote." Silicon Valley keeps promising that AGI is right around the corner, but right now we can't even get a chatbot to not tell us to eat adhesive.
Maybe the real intelligence was the critical thinking we abandoned along the way. **The Take:** AI isn't ready to be your doctor, your lawyer, or your sous chef—but it's absolutely ready to be your most confident idiot friend.
Final Thoughts
Verify everything, trust nothing, and for the love of God, keep the Elmer's out of the kitchen.