Mountain View, California – The Head of Google Search, Liz Reid, has acknowledged that the company’s search engine has been delivering some unusual, inaccurate, or unhelpful AI-generated results to users in the United States. These responses, known as AI Overviews, have sparked concern among users due to their questionable accuracy and relevance.
Reid recently addressed the issue in a blog post, where she explained that Google has implemented new safeguards to improve the quality of its AI-generated responses. She defended the company against fake claims circulating online, emphasizing that some of the more absurd AI Overview answers, such as advising people to leave dogs in cars, are not genuine.
One example cited by Reid involved a query about eating rocks, which led to a satirical response linked to a specific website. While the response may seem bizarre, Reid clarified that it was generated based on the content available on the site. She also highlighted another instance where the AI suggested using glue to stick cheese to pizza, derived from information found on a forum. Despite the potential pitfalls of relying on user-generated content, Reid assured that Google is taking steps to address such issues.
Following widespread criticism of its AI technology, Google has reassessed its approach to generating responses. By analyzing user interactions and feedback, the company has identified areas where improvements are needed. Changes include enhancing the AI’s ability to detect humor and satire, as well as limiting the inclusion of user-generated content in AI-generated responses.
Moreover, Google has taken a proactive stance by introducing restrictions on certain health topics and queries where AI responses proved ineffective. These measures aim to prevent the dissemination of misleading or harmful information through the AI Overview feature. While the initial rollout may have encountered challenges, Google remains committed to refining its AI technologies to provide more accurate and relevant search results.