Google recently expanded its AI Overviews feature in the US, providing AI-generated summaries for various queries. However, the responses generated were deemed inaccurate, strange, or even dangerous by users. In response to the criticism, Google has stated that they are taking quick action on offensive queries.
According to a statement given to Android Authority, Google acknowledged that the majority of AI overviews offer high-quality information with links for further exploration. They noted that some of the examples seen were unusual queries, while others were manipulated or unable to be reproduced. Google mentioned extensive testing before the feature launch and welcomed feedback to make improvements. They assured that they are addressing inappropriate content based on their policies and using these instances to enhance their systems.
Several AI-generated summaries went viral due to their bizarre and harmful nature. Some examples include suggestions to eat a small stone daily, use non-toxic glue to make cheese stick to pizza, drink two liters of urine daily to pass a kidney stone, and jump off a bridge if feeling depressed. These responses seem to be sourced from general AI descriptions that may reference satirical articles or forums like Reddit.
This latest incident adds to a series of AI errors observed with Google’s AI initiatives. The company has faced criticism for rushing to implement AI in various services. Notably, Gemini’s imaging feature was condemned for generating images of racially diverse World War II-era Nazis instead of historically accurate representations. Previous versions of the Bard, now known as Gemini, have also garnered attention for producing hallucinations and inaccurate answers.
In conclusion, Google’s AI Overviews feature has raised concerns due to the inaccuracies and questionable responses it has provided. As the company strives to improve its systems and address problematic content, users are advised to approach AI-generated information with caution and verify critical details independently. The incident serves as a reminder of the challenges and responsibilities associated with deploying AI technologies in everyday services.
Article Source
https://www.androidauthority.com/google-remove-ai-overviews-3446363/
