Google Deploys Language AI Model For Matching Stories and Checking Facts

Previously, Google needed over 40 minutes for detecting breaking news queries; however, it is now able to detect in just a few minutes. Thanks to BERT, one of Google’s language AI models, which comes with language-based capabilities. It can comprehend more complex as well as natural-language queries.

Google is using BERT in its full coverage news stories for matching stories more effectively with fact checks, and also to comprehend the results which are most relevant to the queries that get posted on the ‘Search’ box. With the display of fact check labels, Google has made it easy for spotting fact checks in the news, Google images, and search. These labels are given by the publishers who use the ‘ClaimReview’ schema for marking the fact checks they publish.

Nevertheless, Danny Sullivan, who is the Public Liaison for Google Search, acknowledged the fact that Google systems aren’t perfect always. Nevertheless, when it is about high-quality and trustworthy information, even with the help of advanced information AI-based understanding capabilities, Google doesn’t comprehend content the way humans do it.

Also Read: Food Deliveries to Go Aerial Soon, Drones to Become Reality

Search engines mostly interpret content quality with the help of ‘signals’. As in, the number of quality pages which link to a specific page signifies that a page might be a trusted source of information on a particular topic.

Google has 10,000+ search quality raters; these individuals collectively execute sample searches in millions and rate the quality of the results. Therefore, in case the systems miss to avoid publishing policy-violating content, the enforcement team will take the necessary action as per Google’s policies.

For any clarifications/feedback on the topic, please contact the writer at bhavana.pn@cleartax.in