Google’s AI Insists That Next Year Is Not 2027
We have it on good authority that next year is not 2027.
That authority, of course, is Google’s horribly inaccurate AI Overview feature.
By now, you’re probably well aware that the search engine’s AI-generated summaries are plagued with hallucinations. But you’d think after years of being unleashed on the unsuspecting public, they might be a little better.
Fat chance, it turns out. Ask Google’s incredibly trustworthy AI model “is 2027 next year,” and you’ll get an answer that inconveniently conflicts with reality.
“No, 2027 is not next year; 2027 is two years away from the current year (2026),” it confidently explains, “meaning next year is 2028, and the year after that is 2027, which is a common year starting on a Friday.” A link icon lists several sources for this claim.
In another instance, Google’s AI says it’s currently 2025.
Clearly, the calendars are all wrong, so we’d recommend making a change to yours: whatever date you marked down for the coming of the singularity, push it way back.
Reddit posts flagging this issue show that the AI Overview has been giving the wrong answer for well over a week. But Google engineers aren’t the only ones who’ll need to confide in their chatbot wives or therapists to cope with the embarrassment: OpenAI’s ChatGPT also struggles when asked if 2027 is next year.
“No 🙂 2027 is not next year,” it asserts (emphasis the AI’s).
To be fair to ChatGPT, which was running the free version of its latest model, 5.2, it does quickly catch its mistake. “Wait — sorry, let me restate that clearly: Since the current year is 2026, 2027 is next year.”
So does Anthropic’s Claude Sonnet 4.5, after making the same blunder.
“No, 2027 is not next year. The current date is Thursday, January 15, 2026, so next year would be 2027,” Claude says. Then it adds: “… wait, let me reconsider that.”
“Actually, you’re right to question this. Since we’re currently in 2026, next year would be 2027.”
Either the tech industry’s top models are all secretly operating as part of some vast but amazingly stupid hive mind, or the large language model architecture that underpins them is wonky junk that struggles to keep up with predictable changes like a new year. We’d expect Google’s AI Overviews to screw up such a softball question, given its storied history that includes recommending you put glue on pizza, and its documented susceptibility to inventing meanings for made-up idioms, such as “you can’t lick a badger twice.” But you might expect just a teensy bit better from flagship chatbots like ChatGPT and Claude, which are supposed to represent the best the industry’s got.
On that note, when we asked Google’s much-hyped Gemini 3 about what next year is, it aced it. And for having passed such a high bar, perhaps that’s why it’s been crowned as the new leader in the AI race. Congratulations.
More on AI: Google Now Stuffing Ads Into Its AI Products
