Google’s artificial intelligence-powered search feature suggested one user to put glue on a pizza and another to jump off a bridge. After the increasing reactions to these false and incorrect answers, a statement came from Google.
Google started offering its artificial intelligence-supported search experience in the USA at the I/O event held last week. “AI OverviewsIn this experience called “, artificial intelligence responded to the searches made by the user and directed them to the relevant links. However, it has not produced good results since its launch.
In a news we shared with you just a few days ago, we saw that the feature gave incorrect answers. Now it has been revealed that this situation continues and shows very serious false results. Following the reactions, a statement came from Google.
Artificial intelligence suggested users “eat stones” and “add glue to pizza”
There are some very strange things indeed among these responses spread on social media. For example, a user asked AI-supported Google Search what he should do to prevent the cheese on the pizza from separating. AI also suggested adding a non-toxic glue. Yes, glue. Of course, such a thing should definitely not be done. According to the information received, the feature gave this answer based on a Reddit comment that was shared as a joke a long time ago.
Other examples are just as interesting. The search provides the user with a quote from a parody news somewhere. to eat stones suggests. One user even asked the artificial intelligence about depression and asked him about it in the USA. Jumping from the Golden Gate Bridge He even says he recommends it.
Google made a statement following the reactions. A company spokesperson said errors are often caused by “from rare queriesHe stated that it was due to Also most people Does not reflect real experience He kind of defended the feature by saying. He added that they will use these samples to improve the feature.
RELATED NEWS
Google Will Start Showing Ads on AI Responses Too
Even though Google says so, it would not be wrong to say that the results are worrying. We know that AI-powered search is still experimental. Therefore, making mistakes can be considered normal. However, it is not acceptable for him to give such completely wrong answers. We will wait and see whether the feature will be improved in the future and whether it will truly provide a better search experience as Google says. Unfortunately, it is not possible to say it is better now.
Source link: https://www.webtekno.com/yapay-zeka-destekli-google-arama-hatali-yanit-h144431.html