Google’s Move Against Misleading Search Results

Google has announced that it will be implementing restrictions on AI Overviews, its artificial intelligence-powered search experience. This decision comes in response to numerous instances where the platform delivered unusual and incorrect results to users. The company made this announcement on May 30, highlighting its commitment to improving the accuracy and reliability of its AI-driven […]

AI Overviews: Google's Move Against Misleading Search Results
by Aparajita Sambhaw - May 31, 2024, 4:39 pm

Google has announced that it will be implementing restrictions on AI Overviews, its artificial intelligence-powered search experience. This decision comes in response to numerous instances where the platform delivered unusual and incorrect results to users.

The company made this announcement on May 30, highlighting its commitment to improving the accuracy and reliability of its AI-driven search capabilities.

According to a CNET report, the AI referenced an 11-year-old Reddit comment as a serious suggestion to keep cheese on pizza by adding an eighth of a cup of safe glue. Additionally, it advised consuming “at least one small rock per day,” a recommendation originally from a satirical article by The Onion in 2021.

This type of AI hallucination refers to instances where a generative AI model produces incorrect or deceptive information, presenting it as true. These hallucinations often stem from inaccurate training data, algorithmic glitches, or misunderstandings of the context.

Liz Reid, the head of Google’s search division, mentioned in a blog post on May 30 that the company has developed systems to detect “illogical queries” and has restricted the incorporation of satirical and humorous content.

Meghann Farnsworth, a spokesperson for Google, mentioned that the errors came from “generally very uncommon queries, and aren’t representative of most people’s experiences.” She noted that the company has implemented measures to address policy breaches and is utilizing these “specific instances” to further enhance the product.

Google’s recent AI launch, Gemini, faced criticism for generating historically inaccurate images, including Black Vikings, racially diverse Nazi soldiers, and a female pope. This scrutiny led to Google issuing an apology and pausing the feature, as reported by Forbes.

These recent developments also coincide with a strategic phase for Google, wherein the tech giant is actively exploring avenues to monetize its AI capabilities. One such initiative involves conducting trials for integrating search and shopping advertisements within AI overviews, marking a significant step towards harnessing the commercial potential of its advanced technologies.