Janelle Shane, who in her day job creates computer-controlled holograms for studying the brain, collected genuine messages printed on the heart-shaped sweets to use as training data. She then fed them to a neural network so that it could learn the patterns behind the words. The messages it produced ranged from the endearing “LOVE BUN” to the frankly insulting “MY HAG”.
— Janelle Shane (@JanelleCShane) February 11, 2020 Shane could only find about 360 authentic messages to use as training data, so she decided to perform a follow-up that added her favorite ones produced by the neural network. This increased the total training dataset to almost 500 messages.
Sickly sweet candy hearts
The neural network’s second attempt produced a slightly disturbing fetish for bears, producing messages including “BE MY BEAR”, “TIME BEAR” and “STACK BEAR”. “In fact, I’m seeing worrying signs of a bear-based feedback loop that might lead to 100% bear content after a few more iterations of this,” Shane wrote in a blog post. [Read: Love is a myth, Tinder earned $1.2 billion revenue in 2019] It also produced more complimentary, ambiguous and risky expressions of love — as well as revealing a dirtier side to its tastes. Those X-rated messages have been hidden for the benefit those of a sensitive disposition, but if you’re brave enough to take a peek you can do so by signing up to Shane’s email list. The candy hearts aren’t Shane’s first attempt to mix AI and romance. She previously taught a neural network to invent pickup lines, one of which became the title of her book: You Look Like a Thing and I Love You. You’re here because you want to learn more about artificial intelligence. So do we. So this summer, we’re bringing Neural to TNW Conference 2020, where we will host a vibrant program dedicated exclusively to AI. With keynotes by experts from companies like Spotify, RSA, and Medium, our Neural track will take a deep dive into new innovations, ethical problems, and how AI can transform businesses. Get your early bird ticket and check out the full Neural track.
