There are many of examples on social media of Google's new AI Overview product stating strange things, including advising people to put glue on their pizza or to eat pebbles. Users are reporting that a large number of memes vanish from social media soon after they are uploaded, which is because to Google's disorganized rollout, which forces them to manually disable AI Overviews for particular searches as they are posted.




This is strange because, according to CEO Sundar Pichai, Google has been testing AI Overviews for a year. The feature was first made available in beta as the Search Generative Experience in May 2023. During that time, the business has processed over a billion inquiries. However, Pichai has also stated that, "driven by hardware, engineering, and technical breakthroughs," Google has reduced the cost of providing AI answers by 80% during that same period. It seems possible that early optimization occurred before the technology was ready.





Google maintains that the majority of the "high quality information" that users receive from its AI Overview tool. Where appropriate under our content policies, Farnsworth also confirmed that the company is taking swift action" to delete AI Overviews on specific inquiries. "We are using these examples to develop broader improvements to our systems, some of which have already started to roll out," he added.


An AI founder who preferred to remain anonymous said, "A company known for being at the cutting edge and shipping high-quality stuff is now known for low-quality output that's getting meme'd,"




It's true that Google is in a difficult position. With Satya Nadella's now-famous "we made them dance" comment, Bing pioneered AI search before Google, while OpenAI is purportedly developing its own search engine, a new AI search firm is valued at $1 billion, and younger people are apparently migrating to TikTok in quest of the greatest experience. it to compete is obviously there in the organization, and it is what leads to clumsy AI launches. Marcus draws attention to the fact that Meta released Galactica, an AI system in 2022, but it had to be taken down soon after it was introduced since, among other things, it advised people to consume glass. Sounds recognizable.




Google has grand plans for AI Overviews — the feature as it exists today is just a tiny slice of what the company announced last week. Multistep reasoning for complex queries, the ability to generate an AI-organized results page, video search in Google Lens — there’s a lot of ambition here. But right now, the company’s reputation hinges on just getting the basics right, and it’s not looking great.


“[These models] are constitutionally incapable of doing sanity checking on their own work, and that’s what’s come to bite this industry in the behind,”

 Marcus said.

Google Races to Correct Strange AI Responses in Search Results

 



There are many of examples on social media of Google's new AI Overview product stating strange things, including advising people to put glue on their pizza or to eat pebbles. Users are reporting that a large number of memes vanish from social media soon after they are uploaded, which is because to Google's disorganized rollout, which forces them to manually disable AI Overviews for particular searches as they are posted.




This is strange because, according to CEO Sundar Pichai, Google has been testing AI Overviews for a year. The feature was first made available in beta as the Search Generative Experience in May 2023. During that time, the business has processed over a billion inquiries. However, Pichai has also stated that, "driven by hardware, engineering, and technical breakthroughs," Google has reduced the cost of providing AI answers by 80% during that same period. It seems possible that early optimization occurred before the technology was ready.





Google maintains that the majority of the "high quality information" that users receive from its AI Overview tool. Where appropriate under our content policies, Farnsworth also confirmed that the company is taking swift action" to delete AI Overviews on specific inquiries. "We are using these examples to develop broader improvements to our systems, some of which have already started to roll out," he added.


An AI founder who preferred to remain anonymous said, "A company known for being at the cutting edge and shipping high-quality stuff is now known for low-quality output that's getting meme'd,"




It's true that Google is in a difficult position. With Satya Nadella's now-famous "we made them dance" comment, Bing pioneered AI search before Google, while OpenAI is purportedly developing its own search engine, a new AI search firm is valued at $1 billion, and younger people are apparently migrating to TikTok in quest of the greatest experience. it to compete is obviously there in the organization, and it is what leads to clumsy AI launches. Marcus draws attention to the fact that Meta released Galactica, an AI system in 2022, but it had to be taken down soon after it was introduced since, among other things, it advised people to consume glass. Sounds recognizable.




Google has grand plans for AI Overviews — the feature as it exists today is just a tiny slice of what the company announced last week. Multistep reasoning for complex queries, the ability to generate an AI-organized results page, video search in Google Lens — there’s a lot of ambition here. But right now, the company’s reputation hinges on just getting the basics right, and it’s not looking great.


“[These models] are constitutionally incapable of doing sanity checking on their own work, and that’s what’s come to bite this industry in the behind,”

 Marcus said.

No comments