Hey Google He confirmed that he is putting his foot on the brake regarding use artificial intelligence Get creative with your research tool after errors start to appear. In recent weeks, users who used AI Overview, a new AI-powered summary tool for searches, received responses saying they should use glue to make the cheese stick better PizzaFor example.
The AI has only been used in the US, but users have noticed a decline in application through the platform. Liz Read, vice president of research at Google, confirmed the change in a post on the company’s blog.
“Over the past week, people on social media have shared some strange experiences (there were also a large number of fake photos). We know that people trust Google to provide accurate information, and they have never hesitated to point out errors when they come across them,” he wrote. CEO. According to her, the pictures that showed answers about smoking during pregnancy, for example, were wrong.
According to Google, social media posts will not be used as a source of AI during searches. Moreover, the search configuration with the generative feature can be deactivated when the search concerns health issues – as well as some topics that provided unsatisfactory or incorrect results. Reed also said dozens of technical modifications are being made to the tool.
Liz explained in the article that the tool works differently from other generative intelligences. “AI Overview works very differently from chatbots and other products people may have experienced. It doesn’t simply generate a response based on training data. there summary “Powered by a custom language model, integrated with our core web classification systems, and designed to perform traditional ‘search’ tasks such as identifying relevant, high-quality results in our index.”
According to the CEO, this nuance in AI explains why he doesn’t have hallucinations. Errors will only occur for two reasons: lack of good information and difficulty interpreting web language.
The company’s blog post argued that the tool has been extensively tested, but nothing compares to millions of users and their meaningless questions. For example, “How many stones should I take per day?” was one of the questions that went viral after the resource was launched.
This isn’t the first time Google has launched an AI product in high-profile ads and then backed away from it. The company has banned users from using its photography tool after conservatives accused it of being “anti-white.”
“Friendly zombie fanatic. Analyst. Coffee buff. Professional music specialist. Communicator.”
