Tens of Millions of Errors Per Hour: Investigation Reveals the 'Accuracy Illusion' of Google AI Search

marsbitPublished on 2026-04-10Last updated on 2026-04-10

Abstract

A New York Times investigation, in collaboration with AI startup Oumi, reveals significant accuracy and reliability issues with Google's AI Overviews search feature. Testing over 4,300 queries showed the accuracy rate improved from 85% (Gemini 2) to 91% (Gemini 3). However, given Google's scale of ~5 trillion annual searches, this 9% error rate translates to over 57 million incorrect answers generated hourly. A more critical issue is the prevalence of unsubstantiated citations. For correct answers, the rate of "unfounded citations"—where provided source links do not support the AI's claims—worsened, rising from 37% with Gemini 2 to 56% with Gemini 3. This makes it difficult for users to verify the information. The AI also heavily relies on low-quality sources, with Facebook and Reddit being its second and fourth most cited domains. Furthermore, the system is highly susceptible to manipulation. A BBC journalist successfully "poisoned" it by publishing a fake article; Google's AI began presenting the false information as fact within 24 hours. Google disputed the study's methodology, criticizing the use of the SimpleQA benchmark and an AI model (Oumi's HallOumi) to evaluate its own AI. The company maintains that its internal safeguards and ranking systems improve accuracy beyond the base model's performance.

Author: Claude, Deep Tide TechFlow

Deep Tide Introduction: The latest test by The New York Times in collaboration with AI startup Oumi shows that the accuracy rate of Google Search's AI Overviews feature is about 91%. However, given Google's scale of processing 5 trillion searches annually, this translates to tens of millions of incorrect answers generated every hour. More troublingly, even when the answers are correct, over half of the cited links fail to support their conclusions.

Google is delivering misinformation to users on an unprecedented scale, and most people are completely unaware.

According to The New York Times, AI startup Oumi, commissioned by the publication, used the industry-standard test SimpleQA developed by OpenAI to evaluate the accuracy of Google's AI Overviews feature. The test covered 4,326 search queries, conducting one round in October last year (powered by Gemini 2) and another in February this year (upgraded to Gemini 3). The results showed that Gemini 2's accuracy was about 85%, which improved to 91% with Gemini 3.

91% sounds good, but it's a different story when considering Google's scale. Google processes approximately 5 trillion search queries annually. Calculating with a 9% error rate, AI Overviews generates over 57 million inaccurate answers per hour, nearly 1 million per minute.

Correct Answers, Wrong Sources

More alarming than the accuracy rate is the issue of "unanchored" citation sources.

Oumi's data shows that in the Gemini 2 era, 37% of correct answers had "unsupported citations," meaning the links attached to the AI summaries did not support the information provided. After upgrading to Gemini 3, this proportion increased instead of decreasing, jumping to 56%. In other words, while the model gives correct answers, it's increasingly failing to "show its work."

Oumi CEO Manos Koukoumidis pointedly questioned: "Even if the answer is correct, how do you know it's correct? How do you verify it?"

The problem is exacerbated by AI Overviews' heavy reliance on low-quality sources. Oumi found that Facebook and Reddit are the second and fourth most cited sources for AI Overviews, respectively. In inaccurate answers, Facebook was cited 7% of the time, higher than the 5% in accurate answers.

BBC Journalist's Fake Article "Poisoned" Results Within 24 Hours

Another serious flaw of AI Overviews is its susceptibility to manipulation.

A BBC journalist tested the system with a deliberately fabricated false article. In less than 24 hours, Google's AI Overview presented the false information from the article as fact to users.

This means anyone who understands how the system works could potentially "poison" AI search results by publishing false content and boosting its traffic. Google spokesperson Ned Adriance responded by saying the search AI feature is built on the same ranking and security mechanisms that block spam, and claimed that "most examples in the test are unrealistic queries that people wouldn't actually search for."

Google's Rebuttal: The Test Itself Is Flawed

Google raised several objections to Oumi's research. A Google spokesperson called the study "seriously flawed," citing reasons including: the SimpleQA benchmark itself contains inaccurate information; Oumi used its own AI model HallOumi to judge another AI's performance, potentially introducing additional errors; and the test content doesn't reflect real user search behavior.

Google's internal tests also showed that when Gemini 3 operates independently outside the Google Search framework, it produces false outputs at a rate as high as 28%. But Google emphasized that AI Overviews leverages the search ranking system to improve accuracy, performing better than the model itself.

However, as PCMag's commentary pointed out the logical paradox: If your defense is that "the report pointing out our AI's inaccuracies itself uses potentially inaccurate AI," this probably doesn't enhance users' confidence in your product's accuracy.

Related Questions

QWhat is the accuracy rate of Google's AI Overviews feature according to the Oumi study?

AThe accuracy rate of Google's AI Overviews was found to be approximately 91% when powered by Gemini 3, an improvement from about 85% with Gemini 2.

QHow many inaccurate answers does the article estimate Google's AI Overviews produces per hour?

ABased on Google's annual volume of 5 trillion searches and a 9% error rate, the AI Overviews feature is estimated to produce over 57 million inaccurate answers per hour.

QWhat is the 'unsubstantiated citation' problem identified in the report?

AThe 'unsubstantiated citation' problem refers to instances where the AI Overviews provides a correct answer, but the attached source links do not actually support the information given. This issue increased from 37% with Gemini 2 to 56% with Gemini 3.

QWhich low-quality websites are frequently used as sources by AI Overviews, according to the Oumi data?

AAccording to Oumi's data, Facebook and Reddit are the second and fourth most cited sources by AI Overviews, with Facebook being cited more frequently in inaccurate answers.

QHow did Google respond to the findings of the Oumi study?

AGoogle criticized the study, calling it 'seriously flawed.' Their spokesperson argued that the SimpleQA benchmark itself contains inaccuracies, that using an AI (HallOumi) to judge another AI introduces errors, and that the test queries do not reflect real user search behavior.

Related Reads

Within Strategy's Framework, STRC's Dividend Yield Remains at 12% as Share Price Stays Below Par Value

Michael Saylor, Executive Chairman of Strategy (MSTR), confirmed that the dividend rate for its STRC perpetual preferred shares will remain at 12.00% through August 2026. The rate has increased from 9% at its July 2025 launch to the current high via a "ratchet" mechanism, which permanently raises the rate by 0.5% whenever the share price falls below $95. This mechanism is intended to push the price back toward its $100 par value and support Strategy's "at-the-market" (ATM) program for issuing new shares to fund Bitcoin purchases. However, the mechanism has not worked as intended. STRC shares closed at $89.46 on July 31, remaining about 10-11% below par value despite the record-high dividend. Competition from rival Strive's higher-yielding SATA securities has pressured demand. The persistent discount has forced Strategy to suspend new STRC issuances via its ATM program, limiting this funding channel for Bitcoin acquisitions. STRC's struggles reflect Bitcoin's own volatility, as the preferred shares historically move in tandem. Analysts have warned the ratchet structure carries long-term, one-way risk. A law firm is investigating Strategy's ability to maintain dividend payments if Bitcoin's price stays low. Retail investors own roughly 83% of outstanding STRC shares, a group seen as prone to panic selling during downturns. In response, Strategy has established financial reserves, including a liquidity cushion covering about 26 months of dividend/interest obligations, and a $2 billion share buyback program alongside a Bitcoin monetization framework, though the company emphasized it is not obligated to sell any Bitcoin.

cryptonews.ru5m ago

Within Strategy's Framework, STRC's Dividend Yield Remains at 12% as Share Price Stays Below Par Value

cryptonews.ru5m ago

Analyst: Bitcoin's Price Will Drop to $60k in August, Then Rebound to $70k

Financial analyst Andrey Poroshin has provided a new forecast for Bitcoin's price dynamics in August. Poroshin, an analyst at the Bitbanker exchange, expects the cryptocurrency market to experience a downturn this month, with prices retesting the $60,000 level due to a lack of supportive macroeconomic catalysts. He noted that the recent US Federal Reserve decision to hold interest rates did not significantly impact the market, while inflation remains above the 2% target. Poroshin stated that Bitcoin is ending July under pressure from moderate volatility and a lack of new macroeconomic stimuli, leading to continued market caution. According to his base scenario, Bitcoin will drop to a range of $60,000 to $62,000 before recovering to $70,000. He pointed out that even $70,000 remains below the cost of mining in the US, which has prompted some miners to shift towards AI data center operations. Poroshin cited the winding down of BitMEX's operations as a potential catalyst for a price rebound, suggesting the exit of weaker players often coincides with market reversals and reduced short-term selling pressure. He believes Bitcoin is currently less susceptible to geopolitical shocks, such as the Iran-US conflict, and does not expect significant market changes in August related to the pending CLARITY Act. Looking ahead, Poroshin forecasts that September will bring more active price fluctuations driven by potential Fed rate decisions and possible discussions or approval of the CLARITY Act.

cryptonews.ru5m ago

Analyst: Bitcoin's Price Will Drop to $60k in August, Then Rebound to $70k

cryptonews.ru5m ago

In Jinjiang, Fujian, a Storage Super Unicorn Lies Quiet

In Fujian's Jinjiang, a city known for sportswear, lies a quiet semiconductor giant: Fujian Jinhua Integrated Circuit Co. (JHICC). Once a promising domestic DRAM manufacturer alongside Yangtze Memory and ChangXin Memory Technologies (CXMT), its journey was derailed in 2018 when the U.S. placed it on an Entity List and filed criminal charges for alleged trade secret theft. This halted production for years. A turning point came in February 2024 when a U.S. federal court found JHICC not guilty. However, it had lost crucial time. While CXMT soared to become a top-valued A-share company in 2024, JHICC, with an estimated valuation of 80 billion RMB, was just restarting. Its current output is primarily customized DDR4 chips, not the advanced DDR5/HBM demanded for AI, but it still benefits from the broader memory chip upcycle. JHICC's story is tied to Chen Zhengkun, a veteran engineer who left Micron to lead the venture. Founded in 2016 with state-backed funding, JHICC partnered with Taiwan's UMC to develop DRAM technology. Rapid progress was cut short by the U.S. actions, which Micron initiated, partly due to its heavy reliance on the Chinese market. Post-sanctions, Chen's team worked to rebuild the production line with reduced reliance on U.S. technology. According to its records, JHICC achieved small-scale production and revenue growth under immense pressure. It now focuses on the stable "niche" DRAM market (e.g., TVs, routers) with a monthly capacity of ~40,000 wafers, aiming for 60,000 by 2026. It holds over 1,000 patents but remains on the Entity List. For Jinjiang, investing in JHICC was a bold industrial leap. The local government provided unwavering financial and logistical support during the crisis, helping the company survive. JHICC has become the anchor for a growing local semiconductor cluster. Though its scale lags behind domestic peers, JHICC's persistence symbolizes a hard-won foothold in a global market long dominated by Samsung, SK Hynix, and Micron. Having missed one boom, it seeks a place in the new AI-driven memory supercycle.

marsbit2h ago

In Jinjiang, Fujian, a Storage Super Unicorn Lies Quiet

marsbit2h ago

Trading

Spot
活动图片