When Google announced plans to use artificial intelligence to generate summaries of results for search queries, it was touted as the biggest change to the search engine in decades. The question now is whether the change was a good idea in the first place.
Over the past few days, social media has been abuzz with users sharing some incredibly bizarre and incorrect answers that were allegedly provided by Google's AI profile. Users who typed their questions into Google received answers from an AI that seemed to come from an alternate reality.
read more: What is Google's AI Overview and why is it wrong?
For example, “According to geologists at the University of California, Berkeley, you should eat at least one small rock a day,” the AI summary said in response to one person's (admittedly goofy) question, apparently referencing an article in the popular humor site The Onion.
What really captured the hearts of pizza-loving internet users was a post in which someone who asked how to make cheese stick to pizza replied, “You can also add about 1/8 cup of non-toxic adhesive to the sauce for extra stickiness.” Sounds yummy, right?
In another false answer, Google shared a racist conspiracy theory: “The United States had a Muslim president named Barack Hussein Obama who served from 2009 to 2017,” saying Obama was and continues to be a devout Christian.
Another response to a question about how to pass kidney stones suggested drinking urine: “Aim to drink at least two quarts (two liters) of urine every 24 hours per day,” the disturbing response read.
In response to CNET's request for comment, a Google spokesperson specifically referred to searches claiming former President Obama is Muslim, saying, “This particular summary violated our policies and we have removed it.”
A spokesperson defended Google's AI's responses, saying the “vast majority” provide accurate information. The statement also questioned whether all of the outlandish answers circulating were accurate, saying some were doctored or couldn't be reproduced internally by Google employees.
“We conducted extensive testing before releasing this new experience, and as with other features we've released in Search, we appreciate your feedback,” the statement said. “We took swift action where appropriate and in accordance with our content policies, and we're using these examples to drive broader improvements to our systems, some of which have already started rolling out.”
The limits of artificial intelligence
Google’s struggles with AI Overviews are one of the most dramatic examples of the limitations of AI technology today and the willingness of big tech companies to adopt it.
Nearly two years after the release of OpenAI's ChatGPT, AI has taken the tech world by storm. Companies large and small, from Microsoft to the U.S. State Department, are investing heavily in new AI tools designed for everything from summarizing meeting notes to creating images, videos, and music from prompts and short descriptions.
read morePrompt Engineering: What you need to know and why it matters
The technology is already changing the way people use computers, from helping with work documentation to understanding math homework, but it still faces problems with “hallucinations” — effectively fabricating facts that aren't actually true in order to provide coherent answers. The problem is so widespread that many AI companies are including prominent warnings in their apps and sites to inform users that the information their AI provides may not be true.
People are currently facing this problem with Google's search service, the most popular site on the web and used by billions of people every day to find information.
How to hide the AI summary
In response, some Google watchers have found that the AI summary disappears if you adjust the service's settings. But it's not the default way Google initially displays results. (Check out CNET's AI Atlas resource page for hands-on reviews of generative AI products like Gemini, Claude, ChatGPT and Microsoft Copilot, plus AI news, tips and commentary.)
read more: AI Atlas, your guide to artificial intelligence today
But that hasn't stopped Google's search service from churning out some bizarre responses, like confidently announcing that dogs have played in the NBA, or inventing a new measurement to convert 1,000 km into tomatoes: “1 kilo tomato.”
