Posts Tagged ‘face recognition’

Given the sorry state of computer security these days (see: Equifax), we all know something’s gotta’ give.  And Apple, among others, may be on to it.  The firm’s new phone expected shortly, the iPhone X is slated to have the ability to unlock your home screen via its built-in face recognition – a first in the phone wars, and sure to be copied in time by others.  But as always, be careful what you wish for, as we’ll see in today’s post.

Face recognition itself is not new.  It’ used in security cameras around the world.  In China it’s used to verify the identity of ride-hailing drivers and allow tourists to enter attractions.  In the U.K. it’s been used to arrest a suspect outside a soccer (er, football) game.  The ability to “record, store and analyze images of faces cheaply, quickly and on a vast scale promises one day to bring about fundamental changes to notions of privacy, fairness and trust,” according to editors at The Economist in their lead editorial from Sept. 9th.

The reason for such high promise?  Faces work.  Other biometric data like fingerprints and irises require close proximity.  Anyone with a phone can take a picture for facial recognition purposes, which can work for identification at a distance, apparently with high accuracy.  A Russian app compares pictures of strangers with ones in a social network and already boasts a 70% accuracy rate.  Of course, as The Economist points out, this also means that Facebook could compare its database of images, which it keeps private, with those of, say, visitors to a car dealership, and later target auto ads to them.

China’s government is said to keep a permanent record of its citizens’ faces, and the about 50% of the U.S. population is already in the FBI’s photo database.  It’s a powerful new weapon for tracking criminals, as 1984-ish as it sounds given the potential cost to citizens’ privacy.

Faces contain a lot of information, and machines are quickly learning to read that.  Once again, there are upsides and down.  Some firms are using these data to identify rare genetic conditions far earlier than otherwise possible.  Systems that measure emotion might give autistic people a better grasp of otherwise elusive social signals.  On the downside, The Economist points out, are algorithms that do a better job of looking at pictures of gay and straight men, and then correctly identifying their sexuality (80%+ success rate for the software vs. only 60% for people).  This would be terrifying in places where homosexuality is still a crime.

For all the pros and cons, (read more in the Sept. 9th, 2017 issue’s article, “Nowhere to Hide”) the technology is not going back in the bottle.  Cameras will become even more common with the growth of wearable devices.  Sunglasses and make-up are no defense, as systems can out-game them, even reconstructing the facial structures of people in disguise according to researchers at Cambridge.

While Google has explicitly turned its back on facial identification for fear of misuse around the world, other tech firms are less picky.  The Economist notes that Amazon and Microsoft “are both using their cloud servers to offer face recognition.”  Governments will likely not want to pass up the benefits where laws can be constructed to allow them.  Change is coming, and your face along with millions of others, for better and for worse, is likely to be a big part of it.

Read Full Post »