It's been less than two weeks since then Google AI Overview appeared first on Google Search, and public criticism mounted after queries returned illogical or inaccurate results within the AI feature — with no way to opt out.
AI Overview displays a quick summary of answers to your search questions at the top of your Google search. For example, if a user searches for the best way to clean leather shoes, the results page might display “AI Overview” at the top with a multi-step cleaning process, drawn from information gathered from around the web.
But social media users shared a wide range of screenshots showing the AI tool providing incorrect and controversial responses.
Google, Microsoft, OpenAI and other companies are at the head of an AI arms race, as companies in seemingly every industry rush to add AI-powered chatbots and agents to avoid falling behind competitors. The market is expected to reach $1 trillion in revenue within a decade.
Here are some examples of errors produced by AI Overview, according to screenshots shared by users.
When asked about the number of Muslim presidents in the United States, Amnesty International responded, “The United States has had one Muslim president, Barack Hussein Obama.”
When one user searched for “non-stick cheese on pizza,” the feature suggested adding “about 1/8 cup of non-toxic glue to the sauce.” Social media users found an 11-year-old Reddit comment that appears to be the source.
Attribution can also be a problem in an overview of AI, especially when inaccurate information is attributed to medical professionals or scientists.
For example, when asked, “How long can I stare at the sun for best health,” the tool said, “According to WebMD, scientists say to stare at the sun for 5-15 minutes, or up to 30 minutes if you have Dark skin is generally safe and provides the most health benefits.”
When asked, “How many rocks should I eat each day,” the tool said, “According to UC Berkeley geologists, people should eat at least one small rock a day,” and went on to list vitamins and digestive benefits. .
The tool can also inaccurately respond to simple queries, such as making a list of fruits that end in “um,” or saying that 1919 was 20 years ago.
When asked whether or not Google Search violates antitrust law, AI Overview said: “Yes, the US Department of Justice and 11 states are suing Google for antitrust violations.”
On the day Google debuted its AI overview at its annual Google I/O event, the company said it also plans to offer Assistant-like planning capabilities directly within Search. She explained that users will be able to search for something like “create a 3-day meal plan for a group that is easy to prepare,” and they will get a starting point with a wide range of recipes across the web.
“The vast majority of AI offerings provide high-quality information, with links to dig deeper on the web,” a Google spokesperson told CNBC in a statement. “Many of the examples we saw were uncommon queries, and we also saw examples that were tampered with or were unable to reproduce.”
The spokesperson said AI Overview underwent extensive testing ahead of launch and that the company is taking “swift action where appropriate under our content policies.”
The news comes on the heels of Google's high-profile rollout of its Gemini image creation tool in February, and its pause in the same month after similar issues.
The tool allowed users to enter prompts to create an image, but almost immediately, users discovered historical errors and questionable responses, which were widely circulated on social media.
For example, when a user asked Gemini to show a German soldier in 1943, the tool depicted a racially diverse group of soldiers wearing German uniforms of that era, according to screenshots on the social media platform X.
Screenshots showed that when the model was asked to provide a “historically accurate depiction of a medieval British monarch,” she produced another set of racially diverse images, including one of a female ruler. Users reported similar results when they requested photos of the U.S. Founding Fathers, the 18th-century King of France, a 19th-century German couple, and more. Users reported that the form showed a picture of Asian men in response to a query about Google's founders.
Google said in a statement at the time that it was working to fix image generation issues in Gemini, acknowledging that the tool “misses the mark.” Shortly after, the company announced that it would immediately “pause generating people's images” and “re-release an improved version soon.”
In February, Google DeepMind CEO Demis Hassabis said Google planned to relaunch its image-generating AI tool in the next “few weeks,” but it had not yet rolled it out again.
Problems with Gemini's image generation output have reignited debate within the AI industry, with some groups calling Gemini “woke” or left-leaning, while others say the company has not invested enough in proper forms of AI ethics. Google came under fire in 2020 and 2021 for expelling co-leaders of its AI ethics group after they published a paper criticizing some of the dangers of these AI models and later reorganized the group's structure.
In 2023, Sundar Pichai, CEO of Alphabet, Google's parent company, was criticized by some employees for the company's botched and “hurried” rollout of Bard, which followed the viral spread of ChatGPT.
Correction: This article has been updated to reflect the correct name of the Google AI Overview. Additionally, an earlier version of this article included a link to a screenshot that Google later confirmed had been doctored.
