Google is making some changes to its AI Overview after its AI-driven search feature returned what the company called “weird and inaccurate” responses to people’s online searches.
AI Overviews were introduced last month at Google’s annual I/O developer conference. Now, when people use Google Search to find information about certain topics, an AI-generated text box appears at the top of the search results with descriptions containing links to external websites. Traditional search results appear below AI Overviews, marking a major shift in how Google presents information.
AI Overview results are produced using Gemini, the company’s large language model (LLM), and are designed for situations where someone “wants to get both a quick overview of a topic and links to learn from,” according to a blog post from Google VP Liz Reid. More.”
Google tech expert Alex Joseph told ABC Audio that AI Overview can answer more complex questions than traditional Google Search.
“What happens with the Artificial Intelligence Overview? [Google] What we can really do is synthesize a lot of information and get to the answer you’re looking for very quickly,” Joseph said.
Instead of presenting users with pages of links to scan, Joseph said AI Overviews streamline the process by summarizing information and providing users with a concise answer.
“They’ll have less friction, they won’t have to click through a bunch of different websites, which can often be a bit of a pain in the ass if you just want to get some information very quickly,” notes Chris Stokel-Walker of technology. journalist and author of “How AI Ate the World: A Brief History of Artificial Intelligence – and Its Long Future.”
MORE: Video — Google announces product updates at developers conference
However, Stokel-Walker said the new feature makes it difficult for people using Google Search to verify the accuracy of the information they read.
“With the dominance of Google Search over the last two decades, we have become accustomed to jumping to conclusions that a search term is largely correct,” he told ABC Audio. “All of a sudden, if you get away with that, as Google suggests, and actually just send an answer to the search results page generated through generative AI, you lose any real way to identify that information and analyze it in some way, whether it’s accurate or not.”
There are other concerns about the new feature. First, generative AI technology has faced criticism both at Google and elsewhere for “hallucinating,” that is, producing unreliable and inaccurate information.
For example, in the few weeks since the AI Overview was released to the public, people using Google Search have been advised to eat at least one small rock a day, which one user was told was a good way to get the cheese to stick to the pizza. Mixing glue into tomato sauce, both of which are, of course, very bad ideas. It was also stated that the seventh US president, Andrew Jackson, who died in 1845, graduated from the university in 2005.
Any benefits from the AI Overview ultimately depend on a balance between convenience and cost, Stokel-Walker said. “You no longer have to click through five or six different pages and perhaps several pages of search results to find the right answer, but that also means the answer may be wrong or not actually the answer you want to get,” he said.
“We’ve always been very clear about the limitations of Master’s studies, there will be occasional hallucinations,” Google’s Alex Joseph said, adding that AI Overview also cited the websites it used to generate its answers.
“That’s one of the reasons why we present all the information for you holistically,” Joseph said. “These are quick shortcuts to help get some information to you quickly, but they’re tracked with areas where you can go, double-check, verify.”
Joseph also said that not all queries are best served by AI Overview: “We only show them on queries where we have high confidence that they will be useful and really improve the experience.”
MORE: Video Google announces product updates at developers conference
Following unusual responses reported by some social media users, Google announced it has made “more than a dozen technical improvements” to its AI Overview. According to Liz Reid’s blog post, they include limiting the inclusion of user-generated content as well as satirical or humorous web pages in the data used to create AI Overviews. Reid said Google has also “initiated additional trigger improvements to enhance our quality protections” around health content, which is “targeted.”[s] not showing AI Overviews for tough news topics where freshness and authenticity matter.”
The blog post also notes that “AI Overviews generally do not ‘hallucinate’ or make things up like other LLM products do” and that incorrect answers are the result of “misinterpreting the questions, misinterpreting a nuance in the language.” “There’s not a lot of great information available online or on the internet.”
Aside from accuracy concerns, Stokel-Walker said Google’s prioritization of AI Overviews over traditional search results could impact revenue and reshape the way business is done on the web.
“Websites produce content; they try to make it attractive to Google. Google will show them in the search results. As a result, people click on the websites, then they see the ads, and the publisher makes money,” says Stokel-Walker, “which allows them to put new content on their websites.” said.
But by replacing the top of the Google Search results page with AI-generated content, websites could see fewer visitors and therefore generate less ad revenue, Stokel-Walker said.
This is ironic, according to Stokel-Walker. That’s because Gemini, the LLM Google uses to create AI Overviews, relies on the websites it currently appears in front of.
“These websites still need to exist, and they need to have a way to generate revenue, because otherwise AI-generated search results have nothing to rely on,” Stokel-Walker said.
MORE: Google joins efforts to increase online transparency with new digital standard
Google told ABC News that its tests showed that the opposite actually happened: that links included in AI Overviews received more clicks than if the page appeared as it usually did in search results. Google also said it “will continue to focus” on sending valuable traffic to publishers and creators.
Regardless of how concerns about AI Overview ultimately play out, it’s just one of the many features the company has planned for its line of tech products.
“I think it’s alarming to do something like this as quickly as Google does,” said technology reporter C. Scott Brown of the Android Authority website.
Google has announced plans for additional features similar to AI Overviews aimed at answering questions about specific web pages or YouTube videos. Brown says these features will come to market in an increasingly competitive environment.
“The reason he’s doing this is because he feels like he has to. He needs to keep up with companies, especially OpenAI, that are creating prolific AI technologies that threaten Google’s core business of distributing information to people. “Google is allowing them to make billions of dollars through search,” Brown said. “We offer advertisements that recognize people,” he said.
“Google can’t rest easy because it sees these as a threat. It can’t figure out how to do this cleanly and correctly,” Brown added. “He just needs to do it.”
Google is making adjustments to its AI Overview after a rocky rollout that originally appeared on abcnews.go.com