“We didn't see any AI overviews of them.”
Fake News
Since Google rolled out its AI Overviews search feature to the public earlier this month, it's been in the news for a flurry of bad news, from incorrectly telling people to jump off the Golden Gate Bridge to advising them to eat rocks.
The situation is out of control quartz Bad actors are reportedly forging screenshots to make it appear as if the new tech feature is returning even stranger responses to queries.
For example, User X posted a screenshot that read, “Doctors recommend smoking 2-3 cigarettes a day during pregnancy,” and the post was viewed 9.9 million times.
However, as other users were quick to point out, the screenshot was likely faked.
Now, Google is fighting back, accusing users of exaggerating how dumb its new AI feature actually is. In a blog post, Liz Reid, vice president and head of Google Search, pointed to the existence of “a great many faked screenshots,” possibly suggesting that concerns about the new feature are overblown.
Still, she warned users to fact-check any questionable output from AI tools themselves, partially avoiding responsibility for potentially dangerous suggestions.
“Some of these fabricated results were obvious and absurd,” she wrote, “including implying that the results returned dangerous results on topics such as leaving dogs in cars, smoking while pregnant, and depression.”
“No overview of these AIs was provided, so we encourage anyone coming across these screenshots to search and review them for themselves,” she added.
Quick Fix
Reid also asserted that Google engineers are hard at work resolving any issues or nonsensical answers that real users may encounter.
Since its release earlier this month, the answers AI Overviews has generated have been so bizarre, and downright made-up, that the company has announced it will be placing restrictions on where these outputs can be seen.
But this doesn't mean the problem is solved: many users still get the wrong answers in the AI overview, and it remains to be seen whether the issue will ever truly go away, given that the company's CEO, Sundar Pichai, has already admitted that there really isn't a solution to what appear to be inherent problems with the technology.
“In defense of Google's AI brief, it presents false, laughable, and even dangerous misinformation at its worst,” he tweeted. The New York Times Editor Adam Sternberg. “But the best thing about it is that it gives you the first few lines of the Wikipedia entry.”
Google Search Details: Google admits AI search feature is a huge failure and announces it will scale back the tool
