Google will fix the artificial intelligence results it disgraced. The company states that many of the resulting screenshots are fake.
Google wants to bring generative artificial intelligence directly to search results and has been doing this for a long time.Search Generative Experience (SGE)He conducts tests under the name “. Behind these tests, instead of the site giving some results to the search terms as it does now, Generating effective answers through productive artificial intelligence. This useful infrastructure produces detailed answers to people about the topics they are looking for and provides information here. In addition to what is published on the websites, help is received from specially trained large language models.. The system adds sources to the information it receives from websites and even generates possible follow-up questions for people. So how exactly does this system perform? The answers produced by the system are not always good, and sometimes they are really “incredibly bad” results may occur. For example, the system was updated last week. “Cheese does not stick to pizza” in response to his search, which he took from a ridiculous Reddit post that is 11 years old, “Use chemical glue for that.” your suggestion had given.
YOU MAY BE INTERESTED IN
This situation, which exploded when it fell on X, was not the only incorrect response of the system. It was said that it would not be a problem for him to eat one stone a day.. At the same time Here The system, which is making a splash with the incredibly bad/wrong answers you can see, is currently manually remove these bad contents one by one being cleaned. Google, which follows the posts on social media and immediately eliminates the nonsense answers produced, accepts its mistakes within the scope of its statement and promises to make the system better. Google stated that many of the screenshots that emerged were fake.says it has already put in place various measures to improve results.
Among these “nonsense” include better detection mechanisms for queries, limiting the inclusion of humorous content in results, and restricting the use of user-generated content in misleading answers. Company bringing advanced protection systems for critical issues such as healthcare to ensure higher accuracy and reliabilitywill not shut down the system despite all the negative feedback and will continue to develop it. will.