Google says AI update will improve quality of search results in 'snippets'
If you've ever Googled something only to come across a little info box highlighting the best answer, you've come across one of Google's "snippets". Featured snippets are the small Google results that the search engine aggregates and displays at the top of the page for many searches.
The problem with featured snippets is that, from the user's perspective, these results appear to be extremely trustworthy: they appear at the top of the results page, after all. Since Google first introduced them years ago, they have become more prevalent over time, but just like the rest of Google search results, snippets are algorithmically populated and not programmed by curators. humans.
Google says it's rolling out an under-the-hood change that should improve the answers people see in those information boxes at the top of many search results pages. According to Google, a new AI model called the "Unified Multitasking Model" allows its search ranking system to check its own work, in a way. The AI model achieves this by cross-referencing the top bold text of a search snippet result with established high-quality search results to see if they say the same thing, even if they do. with different wording.
Image credits: Google
"We found that this consensus-based technique significantly improved the quality and usefulness of featured snippet calls," wrote Pandu Nayak, VP of Google Search, in a blog post.
According to Google, another problem is that sometimes the search engine provides seemingly reasonable answers to a search query that is itself incorrect. Google's latest AI model should also help its results ranking system determine when displaying results in a snippet isn't appropriate because the premise of the question is wrong. The company claims that featured snippets now appear 40% less in these cases.
"This is particularly useful for questions for which there is no answer: for example, a recent search for "When did snoopy murder Abraham Lincoln" yielded a snippet highlighting a date accurate and informative about the Lincoln assassination, but this is clearly not the most useful way to display this result,” Nayak wrote.
Google also announced that it will expand its use of warning messages for searches that do not produce results in which the search engine has "high confidence". The company already uses these content notices for emerging topics that lack established search results, but says it will now deploy them in cases where overall search results don't meet its quality standards.
>
If you've ever Googled something only to come across a little info box highlighting the best answer, you've come across one of Google's "snippets". Featured snippets are the small Google results that the search engine aggregates and displays at the top of the page for many searches.
The problem with featured snippets is that, from the user's perspective, these results appear to be extremely trustworthy: they appear at the top of the results page, after all. Since Google first introduced them years ago, they have become more prevalent over time, but just like the rest of Google search results, snippets are algorithmically populated and not programmed by curators. humans.
Google says it's rolling out an under-the-hood change that should improve the answers people see in those information boxes at the top of many search results pages. According to Google, a new AI model called the "Unified Multitasking Model" allows its search ranking system to check its own work, in a way. The AI model achieves this by cross-referencing the top bold text of a search snippet result with established high-quality search results to see if they say the same thing, even if they do. with different wording.
Image credits: Google
"We found that this consensus-based technique significantly improved the quality and usefulness of featured snippet calls," wrote Pandu Nayak, VP of Google Search, in a blog post.
According to Google, another problem is that sometimes the search engine provides seemingly reasonable answers to a search query that is itself incorrect. Google's latest AI model should also help its results ranking system determine when displaying results in a snippet isn't appropriate because the premise of the question is wrong. The company claims that featured snippets now appear 40% less in these cases.
"This is particularly useful for questions for which there is no answer: for example, a recent search for "When did snoopy murder Abraham Lincoln" yielded a snippet highlighting a date accurate and informative about the Lincoln assassination, but this is clearly not the most useful way to display this result,” Nayak wrote.
Google also announced that it will expand its use of warning messages for searches that do not produce results in which the search engine has "high confidence". The company already uses these content notices for emerging topics that lack established search results, but says it will now deploy them in cases where overall search results don't meet its quality standards.
>What's Your Reaction?
