- Google introduced AI Overviews in 2024 to provide AI-generated search answers.
- AI Overviews show 90% accuracy but still yield millions of incorrect answers daily.
- Google processes nearly five trillion queries yearly, risking 57 million errors per hour.
In 2024, Google introduced a new feature on its search page called AI Overviews, which provides AI-generated answers to any search query put in by the user. The new product essentially transformed Google from a curator of information into a publisher, which meant the onus of being correct and truthful lay with the search giant. Now, a new analysis by The New York Times has found that although the answers in AI Overviews are accurate 90 per cent of the time, the enormous volume of queries Google processes each day likely still results in millions of incorrect responses.
The report highlighted that approximately one in 10 Google AI search overviews contains false information. Since the search engine processes almost five trillion queries per year, users could be exposed to more than 57 million inaccurate answers each hour or nearly one million per minute.
The analysis was conducted with the help of a startup called Oumi, which probed AI Overviews with the SimpleQA evaluation, a common test to rank the factuality of generative models like Gemini. The tests started last year when Gemini 2.5 powered the AI Overviews, with the benchmark showing an 85 per cent accuracy rate. When the test was rerun this year following the Gemini 3 update, the AI Overviews showed an accuracy rate of 91 per cent.
Quizzed about the results, Google said Oumi's analysis was flawed as it relied on a benchmark test built by OpenAI that itself contained incorrect information.
“This study has serious holes,” Ned Adriance, a Google spokesperson, was quoted as saying by the outlet.
Google AI Overviews Criticism
Last year, in the wake of the tragic Air India crash in Ahmedabad, social media users criticised Google after its AI Overview feature claimed that the incident involved an Airbus plane when it was actually a Boeing 787-8 Dreamliner. Social media users questioned Google's use of AI, stating the tech giant should not push its overview results at the top, knowing how much AI models hallucinate.
As the controversy snowballed, Google applied a quick fix, stating it had manually removed the response from AI overviews.
"As with all Search features, we rigorously make improvements and use examples like this to update our systems," the tech company
"This response is no longer showing. We maintain a high quality bar with all Search features, and the accuracy rate for AI Overviews is on par with other features like Featured Snippets."
