Google's AI Overviews gives users dangerous answers to queries

According to the AI, a dog played in NHL

Reading time icon 2 min. read

Readers help support Windows Report. We may get a commission if you buy through our links. Tooltip Icon

Read our disclosure page to find out how can you help Windows Report sustain the editorial team Read more

An AI generated image of a dog playing NHL based on AI Overviews

Google didn’t learn from its past mistakes. The AI Overviews feature added to the search engine became dangerous from the start. If you didn’t know, the tool generated alarming solutions to simple questions. For example, one user typed Cheese not sticking on pizza, and the answer to that was to add non-toxic glue.

When someone asked how many rocks they should eat, the AI Overview told them to eat at least one small rock daily. There are more unhinged answers, but they are even more dangerous.

Why is Google’s AI Overviews feature not working?

Google’s AI Overviews feature became dangerous because it does what it was programmed to do. It provides answers based on the information gathered from various sources such as Reddit, Quora, and satirical sites. Unfortunately, nobody noticed the problem during its preview phase.

In addition, Google should know best what lingers on the internet and prohibit its AI from using any source. After all, Gemini experienced bugs not too long ago, and the company removed its image generation feature and limited the others.

Besides being dangerous, users dislike the AI Overviews feature. So, plenty of videos show ways to turn it off, and they take just a few minutes.

A Google spokesperson told Business Insider that these results are rare, and most answers use high-quality information. Yet, it is difficult to believe that. Users shared various dangerous solutions provided by the AI Overviews feature. For example, according to the AI, a dog played in the NHL, and 1919 was 20 years ago.

Ultimately, the AI Overviews feature from Google is somewhat dangerous. After all, nobody wants to stumble across misinformation while searching for meaningful answers. In addition, some business owners complain that the new feature could make users interact with articles less.

Do you consider the AI Overviews feature dangerous? Let us know in the comments.

More about the topics: Gemini, Google