A new BuzzFeed quiz is the first in what Director of Product for Quizzes Chris Johanesen said he's hoping will be a series of "stunt-y experiments" that the publisher launches this year.
The quiz, timed for Valentine's Day weekend, promises to "create your perfect boyfriend (or girlfriend) using AI technology." Johanesen said it's designed to "poke fun at the situation we're all in" (quarantine, obviously), as well as the "weird world of online dating."
To take the quiz, you answer a series of multiple-choices questions about what you're looking for in your ideal romantic partner.
The questions will probably feel familiar to anyone who's taken a quiz on BuzzFeed or elsewhere online, but the answer should be a lot more unique: Johanesen noted that in a normal online quiz, there might be "12 or 20 different results that are written, and that’s pretty much it." With this one, "you could retake it dozens of times and never get the same results."
Johanesen explained that the BuzzFeed team generated an enormous variety of different profile images using StyleGAN technology. For the text, BuzzFeed staff contributed personality traits, text messages quotes, hobbies and "weird, dark stuff" that the quiz combines algorithmically.
"I think we're mostly trying to embrace the absurdity of it," he added. (I saw this myself when I tried out a demo earlier this week and was assigned a girlfriend who wanted to show off her "collection of scabs.") "We try to match it a little bit to some of your inputs so that it's not totally random. ... An early version was more realistic, but it wasn't as fun."
Looking ahead, Johanesen said he's hoping to create more quizzes that are "more generative," where a writer might come up with a concept but they don't have to "handwrite every single option." Still, it sounds like this approach requires significant editorial work, which Johanesen doesn't expect to change.
"We could definitely use machine learning models to write a quiz, but it probably wouldn’t be very good," he said. Instead, the team is interested in "that intersection of what technology can do that humans can’t, and what humans can do that technology can't."