Google’s new AI system, “AI Overview”, has unleashed a whirlwind of controversy after the emergence of strange and often absurd responses to user queries. On social networks, examples of AI Overview advising actions as crazy as sticking glue on pizza or eating stones have gone viral.
This digital chaos has forced Google to take drastic measures: manually disabling AI Overview responses for specific searches. The speed with which examples of strange responses have spread has generated the feeling that the system has been launched on the market without an exhaustive testing phase, despite the fact that Google assures that the technology has been in development for a year.
The company, known for its obsession with quality and innovation, is now in a difficult situation. Criticism has begun to emerge, with accusations that AI is far from ready to be used in the real world.
“Google, once synonymous with excellence, is now the target of ridicule for the low quality of its AI,” said one AI founder, who preferred not to reveal his identity.
Google, for its part, insists that most AI Overview answers are accurate and useful. The company attributes the strange responses to unusual queries or user manipulations. However, he acknowledged that AI is still developing and that steps are being taken to improve its accuracy.
“We are working to improve the quality and accuracy of AI Overview,” said Google spokesperson Meghann Farnsworth. “We are using the examples of strange responses to identify areas for improvement and are implementing changes to prevent them from occurring in the future.”
Gary Marcus, an AI expert and professor emeritus of neuroscience at New York University, believes that the AI industry is selling false promises. While achieving 80% accuracy in AI is relatively easy thanks to the use of large data sets, reaching 100% is an even greater challenge.
“For AI to really work, you need more than just processing information: you need reasoning and discernment,” Marcus said. “AI needs to be able to discern the plausibility of answers, the quality of sources and apply common sense, something that only artificial general intelligence could achieve.”
Google’s ambitions with AI Overview are big: the company seeks to create a system that can answer complex queries, generate personalized results, and perform visual searches in real time. However, Google’s reputation is now at stake. To regain user trust, the company will need to ensure the accuracy and reliability of its AI, and address the errors that have marred its launch.
“The limitations of AI are obvious,” Marcus concluded. “Current systems lack the capacity for self-assessment and discernment, and this is what is creating problems in the industry.”
