Quantcast
SWEAT POO: These A.I. Candy Hearts Will Endear You to V-Day - February 14, 2018 - SF Weekly
SF Weekly

SWEAT POO: These A.I. Candy Hearts Will Endear You to V-Day

Some of the candy hearts the A.I. made (Eric Pratt)

In Ghostbusters II, when Bill Murray’s character Peter Venkman is reduced to hosting a cable-access show for paranormal cranks, one guest tells him that the end of the world will fall on Feb. 14.

“Valentine’s Day,” he says. “Bummer.”

In reality, the Singularity fell on Feb. 14 because a Colorado researcher named Janelle Shane created an A.I. program that made its own candy hearts based on 360 original messages. Instead of charming-if-chalky sweet nothings like “BE MINE” or “I LUV U,” we have “STANK LOVE,” “U HACK,” “I HONKER,” and “FANG.”

That last one sounds more like a safeword, honestly. CNet elaborates on how it was done:

Machine learning today works by training a neural network — a set of interconnected elements inspired by human brain nerve cells — with source data like photographs, speech samples, or the text on candy hearts. Without understanding any conventional rules, the neural network can spot the kind of patterns that indicate there’s a cat in a photo.

Neural networks for things like Google speech recognition take gargantuan amounts of computing power. But Shane’s candy-heart experiment was a lot simpler. It took only about 10 minutes to train the neural network on her MacBook Pro.

But they’re not all bad: “YOU ARE BABE” and “MY BEAR” are pretty cute. If you shout many of the others out in the throes of lovemaking, you’re likelier than not to get a knuckle sandwich from your one true love.