Google’s artificial intelligence feature, "AI Overview," has been producing dangerously inaccurate responses to user searches, and the company has yet to find a reliable solution.
In light of reports about the AI system generating incorrect and harmful responses, Google has temporarily disabled certain queries for the "AI Overview" feature.
One widely circulated report involved the AI responding to a query about keeping cheese on pizza by suggesting the use of glue. In another instance, the system cited geologists to recommend eating "at least one small rock per day."
Did you know?
Subscribe - We publish new crypto explainer videos every week!
Paper Hands vs Diamond Hands: Crypto Slang Explained (ANIMATED)
Despite some humorous mistakes, the main issue lies in the AI system's tendency to produce both accurate and inaccurate results with equal confidence, which poses significant risks.
According to Google representative Meghann Farnsworth, the company has been responding to the issue by removing problematic queries as they arise.
Google has shifted some of the blame to users. Farnsworth noted that many of the problematic examples involved "uncommon queries" or were manipulated.
Currently, it remains uncertain how users can avoid making "uncommon queries." Furthermore, Google's AI system, like many large language models, tends to produce varying answers to identical questions when asked repeatedly.
Overall, Google's AI Overview feature has sparked concern due to its unreliable outputs and the company's struggle to address these issues effectively.
In other AI-related news, Robin Li, co-founder and CEO of Baidu, suggests that although many worry about AI's fast progress, artificial general intelligence (AGI) is still more than a decade away.
