Terrible advice forces Google to rein in AI Overviews
Google has intervened after its new AI Overviews feature began giving dangerous advice. This feature, designed to instantly summarise search results, was found to contain some serious errors, which now requires Google to correct manually.
Last week, Google introduced AI Overviews in the United States, a new search feature that uses artificial intelligence to provide short, direct answers to users’ questions. The feature had been tested for months and seemed ready for use, but recent incidents have shown that the system is not yet completely reliable.
Some users shared on social media the dangerous and bizarre advice they received from AI Overviews. Examples include the advice to use ‘non-toxic’ glue to fasten cheese on pizza, and the claim that geologists recommend eating a rock daily.
These answers appeared to be based on information taken from satirical websites such as The Onion and comments on platforms such as Reddit. Google acknowledged these problems, stressing in a response to BBC that these were “isolated examples” and that most of AI Overviews’ answers were of high quality and well-supported with links to reliable sources.
To prevent further damage, Google acted quickly by adjusting the erroneous and dangerous answers. Ironically, this is done manually by employees, demonstrating that the AI cannot yet operate fully independently without human intervention.
Google continues to stress that most AI Overviews answers are reliable, but the company remains alert to potential errors and continues to improve the technology to ensure safety and reliability.
News Wires
Subscribers 0
Fans 0
Followers 0
Followers