Google has been facing criticism lately for the odd, humorous, and sometimes downright inaccurate answers provided by its AI Overviews in search results.
AI Overviews are the AI-generated snippets that Google deployed more broadly earlier this month, though the results have been mixed. For instance, a user seeking advice on how to get cheese to stick to their pizza was advised to use glue—advice pulled from an old Reddit post. Another user was told to eat “one small rock per day,” an absurd suggestion sourced from The Onion.
You may not encounter these errors yourself, as Google is actively working to remove inaccurate results. A company spokesperson mentioned that Google is taking “swift action” and using these incidents to develop broader improvements to their systems.
“The vast majority of AI Overviews provide high-quality information, complete with links for deeper exploration on the web,” the spokesperson said. “Many of the examples we’ve seen involve uncommon queries, and some have been doctored or couldn’t be reproduced. We conducted extensive testing before launching this new feature, and as with other features in Search, we value the feedback.”
It’s likely the accuracy of these results will improve over time, and some of the screenshots circulating on social media were likely created for humor.
However, this raises a question: What are AI Overviews actually for? Even assuming they worked perfectly, how are they better than traditional web search?
Google aims to provide users with instant answers without requiring them to scroll through multiple web pages. The company noted that in early tests of AI Overviews, “people use Search more and are more satisfied with the results.”
The idea of eliminating the traditional “10 blue links” isn’t new for Google. While those links have become less central over time, it seems premature to completely abandon them.
Consider a search like “what is TechCrunch.” The AI-generated summary was mostly accurate but felt padded with unnecessary details, including traffic numbers from a Yale career website. For the query “how do I get a story in TechCrunch,” the overview referenced an outdated article about submitting guest columns, which TechCrunch no longer accepts.
The issue isn’t just finding more examples of AI Overviews getting it wrong. It’s also about acknowledging that many errors will be mundane rather than spectacular. And although Google deserves credit for including links to source material, determining which source provided which part of the answer often requires a lot of clicking.
Google also points out that the most criticized results often come from data voids—topics with little accurate information online. This is a valid point but underscores the need for a healthy, open web full of accurate information.
Ironically, AI could pose a threat to that very open web. If AI-generated summaries become the norm, there’s less incentive to write detailed how-to articles or break investigative stories, as people might only read the summaries.
Google claims that AI Overviews lead people to visit a wider variety of websites for complex questions and that the links in AI Overviews get more clicks than traditional web listings. I hope this is true. But if it isn’t, no technical improvement can compensate for the potential loss of vast swaths of the web.
For More Information Visit Now Content Baskit.
