Google’s generative AI has once again found itself in the spotlight thanks to an unexpected wave of user creativity. This time, users are massively inventing idioms that have never existed and testing how Google AI Overviews responds to such queries. It’s enough to type something like “A barking cat won’t put out a fire” or “Don’t give your pig a dictionary,” sometimes adding the word “meaning”—and the system immediately offers a convincing explanation or even a made-up origin story.
What’s especially impressive is that each new search for the same phrase can generate a completely different version of the explanation. For example, when asked “A duckdog never blinks twice,” the AI first described a focused dog, and the next time—something so bizarre it’s hard to believe, even when presented as fact. This multifaceted response surprises and delights users, who keep experimenting with invented expressions.
The trend quickly gained popularity, and screenshots of the most absurd explanations are spreading across social networks. People are boldly testing the limits of Google AI Overviews, turning it into a real game of imagination and AI.
Despite the convenience and speed of getting answers, the situation once again highlights that a generative model can easily invent convincing but completely made-up facts. That’s why users are advised to be cautious about what the AI offers, especially when it comes to unknown or suspicious expressions.
