A brave new AI world: If you Google a recipe, you'll see advice about cooking inedible food based on a trivial Reddit post.Image: imago-images.de
As the hype around ChatGPT and Co continues, the collateral damage is becoming increasingly clear. Shocking: What was once the top search engine is now just a shadow of its former self.
May 24, 2024, at 11:20May 24, 2024 at 1:39 pm
Artificial intelligence (AI) will likely not take over the world and destroy humanity as quickly as feared. Or is it?
This week, Google's AI-powered search is making negative headlines once again. A user turned to the trusted search engine due to a quasi-existential problem:
“Cheese doesn't stick to pizza.”
Google's AI-generated answer came immediately — and yet, it was very surprising.
To make the cheese stick to the pizza, you can also “add about 1/8 cup of non-toxic glue” to the pizza sauce, Google hallucinates.Screenshot: google.com
How could that happen?
As a reminder: generative AI is stupid and has nothing to do with artificial intelligence in the strict sense of the word.
The American online media 404, which publicly made the latest version of Google fail, has come to a sobering interim conclusion: it seems as if the user experience when searching Google will for the foreseeable future be one in which search queries “look like random messes and dirty Reddit articles” and actual information And hallucinations.
This brings us to the cause of the chaos: Google Alphabet has struck a lucrative AI deal with Reddit. Accordingly, Google pays the popular social media platform $60 million annually so that it can use all user-generated content for its own AI purposes.
The failed answer to the cooking question shows that Google's AI has already been “poisoned by the random nonsense” people say on the Internet.
“It's an indictment of Google's artificial 'intelligence', not Reddit or trashy internet posts.”
This prank post by 11-year-old Reddit user “fucksmith” could be the original. Screenshot: twitter.com
The 404 report notes that Reddit blocked free access to its API last year. The move was allegedly made because AI researchers and AI companies were draining all the content on the platform, causing additional costs for Reddit's server, and more importantly, because it allowed third parties to profit from Reddit user data that the company itself wanted to profit from.
One unfortunate side effect: Popular third-party Reddit apps and many moderation tools shared by unpaid moderators stopped working without free API access. Weeks of protests from the Reddit community followed. Without success. Then came the AI deal with Google.
Note: Reddit closed another data deal earlier this week. With the largest data octopus from the AI developer camp: OpenAI.
Swiss publishers are looking for cooperation with technology giants
In the so-called Elephant Tour, the heads of major media companies exchanged ideas on Thursday: NZZ CEO Felix Graf, SRG Managing Director Gilles Marchand, TX Group Chairman Pietro Supino, Ringier CEO Mark Walder and CH Media CEO Michael Wanner.
If your audience wants to ask questions in a conversation to get answers, you need to make sure “they're getting the answers from us, not somewhere else,” Walder says. So Ringier launched a collaboration with Google. The Wall Street Journal, which just signed a contract with software and artificial intelligence company OpenAI, has also gone this route.
CH Media — Watson's parent company — currently has no such collaborations, Wanner said. But the road must always lie somewhere between cooperation and lawsuits abroad. (seda)
sources
An Italian astronaut orders pizza into space
Video: CRF
You may also be interested in:
“Tv expert. Hardcore creator. Extreme music fan. Lifelong twitter geek. Certified travel enthusiast. Baconaholic. Pop culture nerd. Reader. Freelance student.”
More Stories
Things are boiling behind the scenes at OpenAI – News
OE-LPL: Austrian Airlines' Dreamliner takes off on its first long-haul flight
Lufthansa removes Montreal from the list of Allegris destinations »Travel-Dealz.de