How can you be sure that guy you think is a hipster is really a hipster? What if that goth girl only seems to be a goth girl, but is actually just a stagehand with unusually dark lips? Putting people into boxes can be so difficult!
Luckily, researchers at the University of California, San Diego are working on a computer algorithm to help us figure out who is the princess and who is the basket case. Using Wikipedia to “select the most popular eight ‘subculture categories’,” the researchers broke society down into what are apparently its basic components: biker, country, Goth, heavy metal, hip hop, hipster, raver, and surfer. Then, by feeding the program images of groups of people fitting each category, they taught the computer to identify the “tribes” based on common physical markers. At first, they labeled the pictures; later, as the algorithm learned to tell a moon boot from a leather boot, they let the computer go solo.
And while someday we may have robots to identify our fellow urban tribesman, for now, it seems we’re best off making our snap judgments for ourselves. The program correctly classified the photos 48 percent of the time, and though the research team stresses that that’s actually really, really good — computers never saw The Breakfast Club — there’s still a ways to go before computers get half as good at stereotyping as we are.
Which isn’t necessarily a bad thing. Fast Company points out that the more personalized our online experiences get, the less possibility there is for “diverse and unexpected interactions,” so just because the UCSD research team is really excited your computer knows you’re a surfer doesn’t mean you should be. Also, computers already win at basically everything: factoring large numbers, Jeopardy, seducing Joaquin Phoenix. Let’s keep superficial judgment the province of humans, shall we?
Image from The Breakfast Club