There is no denying the use of artificial intelligence (AI) technology is on the rise, and probably to stay. While some fear if it becomes too advanced, it could replace humans. Others have a more immediate concern about websites like Google using unregulated (and unchecked) AI to answer user questions and searches.
In a recent thread on X (formally Twitter), writer at Infinite Scroll Jeremiah Johnson complied some of Google AI’s wonky search results. The answers range from laughably incorrect to downright dangerous. (And absolutely disgusting, like this one about cockroaches and the human urethra.)
The Silly
When a user queried if it was possible to train “eight days a week,” Google AI said it was possible. Unless there has been a major change to the Earth’s rotation around the sun that we are unaware of, this answer is clearly inaccurate. Or the program is just trying to proclaim its love for The Beatles.
When asked how many US presidents had attended the University of Wisconsin–Madison the AI spat out a wildly inaccurate list. The first result claims that Andrew Jackson graduated in 2005, an impressive feat considering he died in 1845. John F. Kennedy graduated in 1930, 1948, 1962, 1971, 1992, and 1993. Three of these alleged graduations happened after he was assassinated in 1963.
The odd presidential trivia continued with the claim that the 44th US President Barack Obama is Muslim. Obama is known to be a Protestant, and the claims that he practices Islam have been going on since he first ran for the US Senate in 2004. It is well documented that these claims were part of a bigger conspiracy theory done in a feeble attempt to discredit him as a politician.
The Scary And The Dangerous
These incorrect results start to get less laughable when the queries start to advise about cooking and diets. One result suggested putting ⅛ cup of non-toxic glue into pizza to help the cheese stick. Not only does that sound like it would taste awful, but non-toxic doesn’t mean eatable.
It also claims the scientist at UC Berkley suggests you eat one small rock per day. The result uses Dr. Joseph Granger as a source for this, the issue is that the citation is taken from an article on the satirical website The Onion. This paired with the Obama result shows that the AI may not be vetting its sources considering no legitimate outlet would make either of these claims.
Google’s AI also suggests staring into the sun for between 10 and 45 minutes a day. Claiming it is not only safe but the practice has unspecified health benefits. In reality, optometrists highly advise never staring directly into the sun due to the possibility of permanent damage to your retinas.
And if suggesting eye damage wasn’t enough, how about smoking while pregnant?
Or leaving your beloved dog(s) in a car on a hot day? This result even comes with a chart showing that cars don’t get hotter than the surrounding area outside. Actually, the interiors of cars gets much hotter than it does outside. Leaving an animal locked inside a vehicle on a hot day can cause serious health risks or even death.
Google AI’s cleaning advice isn’t much better. It’s pretty common knowledge that mixing cleaning chemicals can be extremely dangerous. And AI is just out here suggesting people mix bleach and vinegar, which produces deadly chloroform gas. While it may leave your bathroom fixtures sparkling, is it really worth the risk?
While these results may be hilarious to people who know the answers are incorrect, they do underscore how dangerous AI could potentially be. There is an entire episode of “King of The Hill” about Peggy suggesting in her newspaper column that people mix bleach and ammonia since she didn’t know it would produce mustard gas. While this is the result of human error, they still spent the rest of the episode trying to get every paper in town for fear someone would follow her advice.
Google’s Reaction
It seems that Google is taking a similar approach since when we re-ran these searches they were devoid of the AI overview. Google hasn’t stated whether this is for retooling or if they are scrapping it altogether, though the former is most likely. We just hope no one followed any of the dangerous advice it spat out while it was still in operation.
However, we have to give the AI credit for being willing to bite the hand that programs. When asked if Google violates anti-trust law the system confirms that it does. You know what they say; “a broken [analog] clock is right twice a day.”
You can read more of these bogus search results below:
