When Artificial Intelligence Comes to an End and Breaks the Link: A Case Study from Google, Bing, Meta, and Other AI-Aided Systems
Imagine this: you’ve carved out 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. You run into a problem when you bite into your oily creation because the cheese falls off right away. You have no idea how to get a solution, so you turn to the internet.
Yes, don’t do that. That is what a new feature in the AdWords platform will tell you to do. The feature, while not triggered for every query, scans the web and drums up an AI-generated response. The answer received for the pizza glue query appears to be based on a comment from a user named “fucksmith” in a more than decade-old Reddit thread, and they’re clearly joking.
Google continues to say that its AI Overview product largely outputs “high quality information” to users. An email from an official with the company said many of the examples they have seen have been uncommon queries and that some of them have been doctored or unable to reproduce. When asked about the removal of Artificial Intelligence Overviews on certain queries where appropriate under the company’s content policies, and the use of those examples to develop broader improvements to its systems, Farnsworth said that they had already started to roll out.
Take the launch of this feature by the I/O. The demo was controlled and yet it gave an incorrect answer about how to fix a jammed film camera. Don’t open the back door and remove the film unless you want to ruin your photos.
Look, it’s a tough spot for Google to be in. Bing went big on artificial intelligence before Google did, the startup Openai has become worth $1 billion, and there is a younger generation of users who just want the best experience, all of which are related to the rise of artificial intelligence. The company is feeling the pressure to win and that’s what makes the releases messy. Marcus points out that in 2022, Meta released an AI system called Galactica that had to be taken down shortly after its launch because, among other things, it told people to eat glass. Sounds familiar.
Companies developing artificial intelligence are often quick to avoid accountability for their systems, just like a parent with a rambunctious child. These companies claim that they can not predict the outcome of this machine learning project.
But for users, that’s a problem. Last year, the search engine company said that it was the future of search. What’s the point, though, if the search seems dumber than before?
Gary Marcus, an artificial intelligence expert and an instructor of neural science at New York University, said that a lot of companies are selling technology that can go from 80 percent correct to 100 percent. Achieving the initial 80 percent is relatively straightforward since it involves approximating a large amount of human data, Marcus said, but the final 20 percent is extremely challenging. In fact, Marcus thinks that last 20 percent might be the hardest thing of all.
Marcus said that these models are constitutionally incapable of doing sanity checking on their work.
Google AI Overviews aren’t Memo’d, or Why Google Can’t Feature Memes? Commentary of Marcus, Meta, and Pichai
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. There is a lot of ambition here, with multistep reasoning for complex queries and the ability to generate anai-organized results page. The company’s reputation is dependent on getting the basics right, and it isn’t looking great.
You need to do some reasoning to determine if it’s plausible. Is the source legit? You have to do things like a human fact checker might do, that actually might require artificial general intelligence,” Marcus said. And Marcus and Meta’s AI chief Yann LeCun both agree that the large language models that power current AI systems like Google’s Gemini and OpenAI’s GPT-4 will not be what creates AGI.
“A company once known for being at the cutting edge and shipping high-quality stuff is now known for low-quality output that’s getting meme’d,” one AI founder, who wished to remain anonymous, told The Verge.
It has been claimed by Pichai that the cost of delivering artificial intelligence answers is down 80 percent because of hardware, engineering and technical breakthrough. It appears that the tech might not have been ready for it.
One of the weird things that the new Artificial Intelligence Overview product from Google says is to put glue on pizza. The mess caused by the launch of the new version of the search engine means it has to manually disabling the artificial intelligence overviews for specific searches, which can cause many to disappear soon after being uploaded to social networks.