Google Blames 'Data Voids' for Inaccurate Results on AI Overviews

Google is blaming so-called "data voids" for the inaccurate, and even sometimes dangerous, answers on its search engine's AI Overviews.

Google Search VP Liz Reid claimed that the "odd and erroneous overviews" shared online are due to lapses in the data available the AI can use as a source.

According to Reid, the lack of content to answer "nonsensical queries" and inaccurate searches cause the AI to misinterpret the question or outsource the response to satirical content.

The statement came after several news outlets and people on social media shared images of the AI feature generating incorrect responses to users like telling them to eat rocks.

One user even noted that the AI told them that "jumping off the Golden Gate Bridge" is a useful strategy for dealing with depression.

Google has since added query restrictions to prevent similar searches where "AI Overviews were not proving to be as helpful."

Beyond the erroneous answers shared online, Google claimed that the AI-powered tool received "higher satisfaction" from user feedback as people ask for "longer, more complex questions that they know Google can now help with."

Google Gemini's AI Still Under Fire

With the renewed attention on Google Search's AI Overviews, more people are taking notice of the lapses over Google's flagship AI product, Gemini.

Before its supposed upgrade during the I/O conference, Gemini AI was criticized for generating racially inaccurate images of historical figures.

The AI was also noted earlier for generating game statistics for the Super Bowl LVIII hours even before the game began.

Google has responded to the criticisms by pausing the feature and promising to fix the issue in later updates.

Both issues in Google Gemini have not yet been directly addressed to the public since the initial response.

Related Article: Google Blocks Certain Search Results for AI Overviews Over Incorrect Suggestions

Google Rolls Out Gemini AI Across its Products, Apps

Despite safety concerns about the technology, the Gemini AI will continue to spread across Google's products and apps.

Several AI-powered features on Gmail, Google Meet, and other workplace apps have since been integrated with the latest Gemini model.

Even other browsers like Opera have started adding Gemini AI to power their chatbots and other customization features.