The Case Against Facial Recognition
From Wiki: a facial recognition system is a technology capable of identifying or verifying a person from a digital image or a video frame from a video source. There are multiple methods in which facial recognition systems work, but in general, they work by comparing selected facial features from given image with faces within a database.
Thus in order for facial recognition to work, the user must already have a reasonable picture of you, and must be looking for you in order to trigger a recognition. Of course, with society so recklessly putting their images on every social media site they can find, without regards to their personal privacy, finding images of almost every citizen in the world is not unsurmountable. Although we are not at the level of “Person of Interest” where everyone is instantly recognized and tracked wherever they go, a false misconception of the truth of facial recognition, we are getting closer to this reality, but for now, facial recognition is complex at best, and flat out incorrect at worst.
The argument against facial recognition technology is more than just a distinction between communal safety and personal privacy. It is also an argument against the technology’s early adoption before it is ready. Just last week, a second false arrest was attributed to faulty facial recognition. If anyone has set their phone or laptop to use facial recognition, you can attest that even within this microcosm of an environment, it never works 100% of the time. Just Google “facial recognition doesn’t work”, and you will get over 38 million results. Putting such unreliable technology into environments where hundreds of faces are processed simultaneously, and the false positivity rates expand exponentially. As someone who has implemented complex video surveillance technology, the false positives with just motion sensing are astounding, and facial recognition is far more complex than just simple motion detection.
Now add in racial differences. It is a known truth that facial recognition algorithms have a harder time with darker skin tones than lighter tones. Experiments show that persons of darker skin are misidentified five times more than persons of white skin. Five times! We already have social injustice in our world without facial recognition. Are we ready to extend it further with technology that is still in its infancy? Many cities and companies are stopping the use of this technology, particularly around policing, where stakes are higher, due to these disparities an inequalities facial recognition enhances.
There may be many reasons these disparities exist between races when concerning facial recognition. Artificial intelligence systems rely on machine-learning algorithms that are ‘trained’ on sample datasets. If people of darker skin tones are underrepresented in benchmark datasets, then the facial recognition system will be less successful in identifying persons. Or it may be, distinguishing features are just too difficult to decipher with today’s video resolution, or even today’s compute restraints. Higher resolutions and super computer potential may change all of this, but for now, the technology is just not ready to put it in an environment that does not treat everyone equal. It might one day, and when it does there could be some very positive uses for the technology, without removing the arguments for privacy. Come back Thursday as I explore advantages facial recognition can bring to our society – when its ready.