Machine learning's most terrifying product yet: An archive of Mark Zuckerberg haircuts
Humanity may not be at a War Games-level crisis when it comes to artificial intelligence and machine learning, but with great computer processing power comes great responsibility—and at least one program’s power is being used in questionable ways.
Using an “open source hair detector” available on GitHub, artist Sam Lavigne created an archive of 387 machine-learning generated images of Mark Zuckerberg’s hair. “Mark Zuckerberg is one of the richest people in the world,” Lavigne told Mashable on Wednesday. “He has nearly limitless haircut options — haircuts we can’t even begin to imagine. Yet he has chosen that haircut. Why?”
Only God and the man behind Facebook—sorry, “F A C E B O O K”—know the answer to that question, but the result of the research is an oddly hypnotic video that has been viewed more than 38,000 times since it was posted on Sunday.
His work also produced what is sure to become someone’s iPhone case pattern.
But Lavigne didn’t stop there. He also used the technology to transform people into Zuckerberg’s coif as a “cautionary tale.”
Fascinated with automation and surveillance, Lavine has used his Twitter feed to share a treasure trove of horrifying/intriguing futures machine learning could provide.
“With all these infamous white collar criminals back on the loose, now is a better time than ever for communities to invest in white collar crime predictive policing technology. Stay safe!” he wrote on Turesday, the day President Trump pardoned or commuted the sentences of a handful of high-profile prisoners.
And, as if schoolteachers don’t have it hard enough with the principal stopping by for classroom assessments, Lavigne recently highlighted a new program that could monitor “the live classroom and provides metrics for Class Attentivity, Class Participation, Happiness Index and also captures automatic attendance.”
We’re just one more step closer to living in the world of The Jetsons.
Send Great Job, Internet tips to [email protected]