Google’s AI Overviews feature is delivering BIZARRE answers faster than ever.
Imagine this: you’ve set aside an evening to unwind and decide to make a homemade pizza. You assemble your pie, throw it in the oven, and are excited to start eating. But as you prepare to take a bite, you encounter a problem — the cheese falls right off. Frustrated, you turn to Google for a solution.
A Bizarre Suggestion from Google’s AI
“Add some glue,” Google answers. “Mix about 1/8 cup of Elmer’s glue in with the sauce. Non-toxic glue will work.”
As ridiculous as it sounds, that’s what Google’s new AI Overviews feature currently suggests. This feature, designed to provide AI-generated responses, mistakenly advises using glue to solve the cheese problem. The AI-generated advice seems to be based on a joke from a decade-old Reddit thread by a user named “fucksmith.”
This is just one of many mistakes cropping up in the new AI feature that Google rolled out broadly this month. For instance, it also claims that former US President James Madison graduated from the University of Wisconsin 21 times, that a dog has played in the NBA, NFL, and NHL, and that Batman is a cop.
Google spokesperson Meghann Farnsworth stated that these mistakes arose from “generally very uncommon queries and aren’t representative of most people’s experiences.” She mentioned that the company has taken action against policy violations and is using these “isolated examples” to refine the product further.
AI’s Current Limitations
Despite Google’s cautionary “Generative AI is experimental” label at the bottom of AI answers, it’s evident that these tools aren’t ready to provide accurate information at scale. The controlled demo during Google I/O’s big launch of this feature also delivered questionable advice on fixing a jammed film camera, suggesting users “open the back door and gently remove the film,” which would ruin the photos.
It’s not just Google facing these issues; companies like OpenAI, Meta, and Perplexity have also grappled with AI hallucinations and mistakes. However, Google is the first to deploy this technology on such a large scale, and the errors keep surfacing.
Accountability in AI Development
Companies developing artificial intelligence often sidestep accountability for their systems, much like a parent with an unruly child. These companies claim they can’t predict what the AI will generate, so it’s out of their control. For users, though, this unpredictability is problematic. Last year, Google declared that AI was the future of search, but what’s the point if the search results seem less reliable than before?
AI optimists argue that we should embrace the hype because of the rapid progress made so far, trusting that it will continue to improve. While it’s true that AI technology will likely get better, focusing solely on an idealized future ignores the significant issues these technologies currently face. This approach allows companies to continue delivering subpar products.
For now, our search experiences are marred by decade-old Reddit posts in the pursuit of incorporating AI into everything. Many idealists believe we are on the brink of something great and that these issues are simply growing pains of a nascent technology. I sure hope they’re right. But one thing is certain: we’ll likely witness someone putting glue on their pizza soon because that’s the nature of the internet.
The incident underscores the need for better oversight and refinement in AI-generated content. While AI has immense potential, its current limitations can lead to bizarre and even dangerous advice. As companies like Google continue to develop and deploy AI at scale, ensuring the accuracy and reliability of AI-generated information is crucial. Until then, users must remain cautious and critical of the advice provided by AI systems.
More News: Technology News
