UPDATE: Google AI Overviews are experiencing significant improvements in 2025, but mixed reactions from users reveal a complex relationship with the technology. What started as a notorious recommendation to use glue on pizza has turned into a more reliable tool, showcasing rapid advancements in artificial intelligence.
In a surprising turn of events, Katie Notopoulos, a journalist who went viral for her 2024 experience with Google’s AI, reflects on how the technology has evolved. Once mocked for suggesting that users add glue to pizza, the AI now offers more coherent and useful responses, according to Notopoulos. “Now that it’s the end of 2025, I can see that Google’s AI Overviews, for as much as I goofed on them for being clumsy, are having the last laugh,” she stated.
The dramatic shift in user experience comes after a year of widespread criticism. Back in 2024, the AI Overviews were deemed “abysmal” due to their frequent inaccuracies and absurd recommendations. One such incident saw Notopoulos actually following the AI’s advice to create a pizza with glue, a moment that became an internet sensation.
As of April 2025, users noted ongoing quirks within the AI. Notable was the “You can’t lick a badger twice” phenomenon, where nonsensical phrases were interpreted as legitimate idioms. This issue highlighted the AI’s struggle with context, prompting a response from Google. According to spokesperson Meghann Farnsworth, “When people do nonsensical or ‘false premise’ searches, our systems will try to find the most relevant results based on the limited web content available.”
However, signs of improvement are evident. Recent tests by Notopoulos revealed that when she inputted a made-up idiom like “you can’t tell a yak not to dance,” the AI provided a more accurate explanation, suggesting it was “not a common, established idiom, but rather a playful or poetic expression.” This indicates that Google is actively refining its algorithms to enhance user interaction.
Users now report using Google AI Overviews more frequently instead of traditional search methods, raising concerns about the implications for web traffic to content creators. Notopoulos admitted, “The fact that people are willing to use the AI Overview answer instead of clicking on a link to a human-generated article is not necessarily great.”
As AI technology continues to evolve, the implications for both users and content creators remain significant. The improvements in AI Overviews signal a shift in how information is consumed online, suggesting a growing reliance on AI-driven answers.
Going forward, experts recommend users remain cautious while engaging with AI Overviews. Understanding the limitations and potential inaccuracies is crucial as the technology becomes more integrated into daily search habits.
As Google works to refine its offerings, the ongoing dialogue between developers and users will shape the future of AI in search. The next wave of updates is eagerly anticipated, with many wondering what new features or challenges could emerge.
Stay tuned for more updates on this developing story.